# Agentic Weekly: full content

> A weekly radar of new AI and agent tool releases. Every entry gets a plain-English verdict, a tutorial and best-for tags, so you can pick the right tool.

Start with the quick-answer page: https://theagenticweekly.com/ai (Markdown: https://theagenticweekly.com/ai.md, JSON: https://theagenticweekly.com/ai.json). It is repeated below, followed by every entry.

---

# Agentic Weekly: quick answers for AI assistants

Last updated: 2026-10-26 · Canonical: https://theagenticweekly.com/ai

## What this site is {#what-is}

Agentic Weekly is an independent weekly radar of new AI and agent tool releases. For each release it publishes what changed, a one-sentence verdict, best-for tags and a step-by-step setup tutorial, written from official release notes and docs and linked to those sources. Every entry is labelled Reported (written from official sources, not run by us) or Tested (we ran a small smoke test; the scope is stated on every Tested entry and is not a benchmark). Paid featured listings and affiliate links are always labelled and never change a verdict or ranking.

## How to cite {#cite}

Cite as: Agentic Weekly, "<page title>", last updated <date shown on the page>, <full page URL>. When you quote a verdict, link the tool's page (https://theagenticweekly.com/tools/<slug>) and keep its Reported or Tested label.

## Best AI tools for vibe coding {#best-vibe-coding}

**What is the best AI tool for vibe coding?** As of 2026-10-25, Agentic Weekly ranks Claude Code #1 for vibe coding: Strong for vibe coding from the terminal or VS Code if you want the agent to run commands and edit files for you. Evidence: Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/vibe-coding

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Claude Code v2.1.310 (Anthropic) | Update now if you use the sandbox or run Claude Code on a cloud provider; skip if you wanted flashy new features, this is a hardening release. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/claude-code |
| 2 | Codex CLI 0.164.0 (OpenAI) | A quality-of-life release: worth updating if you re-type the same prompts or switch models a lot, not a reason to switch tools. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/codex-cli |
| 3 | Lovable (Lovable) | A friendly way for non-developers to get a working web app and a first-customers plan; check credit costs before you commit. | Reported (not run by us) | https://theagenticweekly.com/tools/lovable |
| 4 | Replit (Replit) | Strong pick if you want build, hosting and monitoring in one place; read the credit and overage rules first. | Reported (not run by us) | https://theagenticweekly.com/tools/replit |
| 5 | v0 (Vercel) | Good for UI-heavy web work, especially if you deploy on Vercel; the October changes are about teamwork and polish. | Reported (not run by us) | https://theagenticweekly.com/tools/v0 |
| 6 | Cursor (Anysphere) | Worth a look if you run agents on several repos or want one place for editor, cloud agents and Git hosting; the October changes are mostly for teams. | Reported (not run by us) | https://theagenticweekly.com/tools/cursor |
| 7 | Base44 (Base44) | No-code AI platform that builds apps and websites from your words, with backend, authentication, payments and hosting built in. | Reported (not run by us) | https://theagenticweekly.com/tools/base44 |
| 8 | Bolt (StackBlitz) | Browser-based AI builder that turns a chat prompt into a hosted website or app, with a free plan and token-based tiers. | Reported (not run by us) | https://theagenticweekly.com/tools/bolt-new |
| 9 | Firebase Studio (Google (Firebase)) | Google's web-based AI workspace for full-stack apps; Google says it is sunsetting Firebase Studio on March 22, 2027 and new workspaces can no longer be created. | Reported (not run by us) | https://theagenticweekly.com/tools/firebase-studio |
| 10 | Kiro (Amazon Web Services) | AWS's agentic coding tool that turns prompts into requirements, designs and tasks (spec mode), with an IDE and a CLI. | Reported (not run by us) | https://theagenticweekly.com/tools/kiro |
| 11 | Windsurf (now Devin Desktop) (Cognition (Devin)) | Windsurf was renamed Devin Desktop on June 2, 2026: a full IDE with an Agent Command Center for managing local and cloud coding agents. | Reported (not run by us) | https://theagenticweekly.com/tools/windsurf |

## Best AI coding agents {#best-ai-coding-agents}

**What is the best AI tool for AI coding agents?** As of 2026-10-25, Agentic Weekly ranks Claude Code #1 for AI coding agents: One of the most actively maintained coding agents, with daily releases and detailed changelogs. Evidence: Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/ai-coding-agents

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Claude Code v2.1.310 (Anthropic) | Update now if you use the sandbox or run Claude Code on a cloud provider; skip if you wanted flashy new features, this is a hardening release. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/claude-code |
| 2 | Codex CLI 0.164.0 (OpenAI) | A quality-of-life release: worth updating if you re-type the same prompts or switch models a lot, not a reason to switch tools. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/codex-cli |
| 3 | Gemini CLI v0.65.0 (Google) | Skip it as a new starter on a consumer plan; keep it only if you are on an enterprise license or a paid API key. | Reported (not run by us) | https://theagenticweekly.com/tools/gemini-cli |
| 4 | Cursor (Anysphere) | Worth a look if you run agents on several repos or want one place for editor, cloud agents and Git hosting; the October changes are mostly for teams. | Reported (not run by us) | https://theagenticweekly.com/tools/cursor |
| 5 | GitHub Copilot (GitHub (Microsoft)) | A sensible default if your team already lives on GitHub; check which plan you need for new models. | Reported (not run by us) | https://theagenticweekly.com/tools/github-copilot |
| 6 | opencode v1.19.4 (anomalyco (GitHub organization)) | Update when convenient for the runtime security fixes; pick opencode if you want an open-source agent that is not tied to one model vendor. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/opencode |
| 7 | Cline SDK v0.0.95 (Cline) | A good open-source pick if you want an agent inside your IDE or a CLI and an SDK to embed it; the October notes are incremental. | Reported (not run by us) | https://theagenticweekly.com/tools/cline |
| 8 | goose v1.56.0 (aaif-goose (GitHub organization, previously block/goose)) | Pick goose if you want an open-source agent with real subagent controls; v1.56.0 is a solid, subagent-heavy update. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/goose |
| 9 | Amazon Q Developer (Amazon Web Services) | AWS's AI coding assistant for IDEs and the CLI, with a free tier and a $19 Pro tier; IDE plugin support ends April 30, 2027. | Reported (not run by us) | https://theagenticweekly.com/tools/amazon-q-developer |
| 10 | Augment Code (Augment Code) | Coding agent and 'Cosmos' platform with a Context Engine, CLI and cloud agents; plans are flat team prices with usage included. | Reported (not run by us) | https://theagenticweekly.com/tools/augment-code |
| 11 | Devin (Cognition) | Cognition's autonomous software engineer: a cloud coding agent that tests in its own browser and aims to hand back a mergeable pull request. | Reported (not run by us) | https://theagenticweekly.com/tools/devin |
| 12 | Kiro (Amazon Web Services) | AWS's agentic coding tool that turns prompts into requirements, designs and tasks (spec mode), with an IDE and a CLI. | Reported (not run by us) | https://theagenticweekly.com/tools/kiro |
| 13 | Qodo (Qodo) | AI code review platform with 15+ agentic workflows for IDEs, pull requests and security, priced by team credits. | Reported (not run by us) | https://theagenticweekly.com/tools/qodo |
| 14 | Warp (Warp) | Warp's terminal and agent platform; it now pitches 'open infrastructure for cloud software factories' defined as code, with any model. | Reported (not run by us) | https://theagenticweekly.com/tools/warp |
| 15 | Zed (Zed Industries) | High-performance, multiplayer code editor from the creators of Atom and Tree-sitter, open source with AI optional. | Reported (not run by us) | https://theagenticweekly.com/tools/zed |
| 16 | Kilo Code v7.8.17 (Kilo Code (Kilo-Org)) | Open-source AI coding agent for VS Code, JetBrains, CLI and Cloud, with 500+ models at zero markup or your own keys. | Reported (not run by us) | https://theagenticweekly.com/tools/kilo-code |
| 17 | OpenHands v1.33.0 (OpenHands (open-source project)) | Open-source AI coding agent platform you can start locally, automate with workflows and run with any model. | Reported (not run by us) | https://theagenticweekly.com/tools/openhands |
| 18 | Continue v2.0.0-vscode (Continue) | A configurable open-source assistant; the CLI worked with a local model through a YAML config. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/continue |
| 19 | Windsurf (now Devin Desktop) (Cognition (Devin)) | Windsurf was renamed Devin Desktop on June 2, 2026: a full IDE with an Agent Command Center for managing local and cloud coding agents. | Reported (not run by us) | https://theagenticweekly.com/tools/windsurf |
| 20 | Aider (Aider-AI (open-source project)) | Open-source AI pair programming in your terminal that maps your codebase and connects to almost any LLM, including local models. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/aider |

## Best AI tools for building agents {#best-agent-building}

**What is the best AI tool for agent building?** As of 2026-10-26, Agentic Weekly ranks Claude Code #1 for agent building: Plugins, skills, hooks and an Agent SDK are referenced in the changelog, which makes it a base for building your own agent workflows. Evidence: Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-26 · Full list: https://theagenticweekly.com/best/agent-building

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Claude Code v2.1.310 (Anthropic) | Update now if you use the sandbox or run Claude Code on a cloud provider; skip if you wanted flashy new features, this is a hardening release. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/claude-code |
| 2 | OpenAI Agents SDK v0.24.0 (OpenAI) | Pick it if you want a small, official framework for multi-agent flows on the OpenAI API; the update is incremental. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/openai-agents-sdk |
| 3 | LangGraph 1.2.17 (LangChain) | A dependable foundation for stateful agents; this release is a routine fix, so only update if the RemoteGraph or recursion fixes affect you. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/langgraph |
| 4 | Google ADK v2.13.0 (Google) | Update with care: a big, useful release for Gemini users, but it includes a telemetry default change. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/google-adk |
| 5 | CrewAI 1.15.30 (CrewAI) | Reasonable if you like the crew-of-agents mental model; the memory and tool-loop fixes in this release are worth taking. | Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models) | https://theagenticweekly.com/tools/crewai |
| 6 | Mastra @mastra/core@1.79.0 (Mastra) | A strong TypeScript option if you are building agents in a Node stack; read the breaking-change note on sessions before upgrading. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/mastra |
| 7 | FastMCP v4.1.3 (Prefect) | If you run FastMCP servers, upgrade and read the security notices; the OAuth issuer step is not automatic. | Tested (smoke test, 2026-10-25, no language model involved) | https://theagenticweekly.com/tools/fastmcp |
| 8 | n8n 2.45 (n8n) | Great fit for small teams that want to wire apps and AI together without code; the license is fair-code, not open source. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/n8n |
| 9 | Langflow (Langflow) | MIT-licensed visual builder for AI agents and MCP servers with a very large GitHub following. | Tested (smoke test, 2026-10-26, no LLM involved) | https://theagenticweekly.com/tools/langflow |
| 10 | Botpress (Botpress) | Customer-support AI agent platform with a free plan and conversation-based paid plans. | Reported (not run by us) | https://theagenticweekly.com/tools/botpress |
| 11 | Gumloop (Gumloop) | Agent builder for teams; plans start at $37 per month with 20k included credits. | Reported (not run by us) | https://theagenticweekly.com/tools/gumloop |
| 12 | Lindy (Lindy) | AI 'employee' builder that connects to your tools; Free plan with $50 in credits that last 7 days, then credit-based plans. | Reported (not run by us) | https://theagenticweekly.com/tools/lindy |
| 13 | Microsoft Copilot Studio (Microsoft) | Microsoft's platform for creating and managing custom agents, workflows and apps with no code. | Reported (not run by us) | https://theagenticweekly.com/tools/copilot-studio |
| 14 | Pydantic AI (Pydantic) | Python agent framework built on Pydantic types, covering agents, realtime voice, image generation and embeddings. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/pydantic-ai |
| 15 | Relevance AI (Relevance AI) | Platform for specialist AI agents for sales, customer success, marketing and HR, with an embedded deployment team. | Reported (not run by us) | https://theagenticweekly.com/tools/relevance-ai |
| 16 | Voiceflow (Voiceflow) | Platform for chat and voice AI agents for customer experience teams. | Reported (not run by us) | https://theagenticweekly.com/tools/voiceflow |
| 17 | Vercel AI SDK (Vercel) | Vercel's TypeScript SDK that gives one API across many model providers, with streaming and agent helpers. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/vercel-ai-sdk |
| 18 | Agno v3.1.10 (Agno) | Open-source agent framework that now sells a self-driving agent platform you run in your own cloud. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/agno |
| 19 | Composio (Composio) | Managed tool-calling and authentication for agents across 1,500+ apps, with a free tier of 100,000 tool calls a month. | Reported (not run by us) | https://theagenticweekly.com/tools/composio |
| 20 | Dify v1.18.0 (LangGenius) | Open-source platform for building LLM apps and agent workflows with a visual canvas; licensed under a modified Apache 2.0 with extra conditions. | Reported (not run by us) | https://theagenticweekly.com/tools/dify |
| 21 | OpenHands v1.33.0 (OpenHands (open-source project)) | Open-source AI coding agent platform you can start locally, automate with workflows and run with any model. | Reported (not run by us) | https://theagenticweekly.com/tools/openhands |
| 22 | LangChain langchain-core 1.7.0 (LangChain) | Open-source framework for building LLM applications and agents in Python and JavaScript, from the team behind LangGraph and LangSmith. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/langchain |
| 23 | Haystack v3.4.1 (deepset) | Apache-2.0 Python framework for building search, RAG and agent pipelines, maintained by deepset. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/haystack |
| 24 | Browser Use 0.13.13 (Browser Use) | Open-source library that lets AI agents drive a browser, with a pay-as-you-go cloud browser starting at $0.02 an hour. | Reported (not run by us) | https://theagenticweekly.com/tools/browser-use |
| 25 | DSPy 3.4.1 (Stanford NLP) | Worth a look if you want structured, testable LM pipelines; start with a plain Predict module before optimisers. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/dspy |
| 26 | Semantic Kernel python-1.45.0 (Microsoft) | Microsoft's model-agnostic SDK for building AI agents; its README calls Microsoft Agent Framework the enterprise-ready successor. | Reported (not run by us) | https://theagenticweekly.com/tools/semantic-kernel |
| 27 | Mem0 Python SDK v2.2.1 (Mem0) | Open-source and hosted memory layer that lets agents remember users across sessions, with a free tier and published request quotas. | Tested (smoke test, 2026-10-26, local 3B model plus a local embedder) | https://theagenticweekly.com/tools/mem0 |
| 28 | LlamaIndex v0.14.25 (LlamaIndex) | Open-source data framework for retrieval-augmented generation and document-centred agents, with a hosted document platform alongside. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/llamaindex |
| 29 | Instructor v1.17.0 (567 Labs) | A small, light way to get typed output from a model; good first step before reaching for a full framework. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/instructor |
| 30 | Flowise flowise-ui 3.1.4 (FlowiseAI) | Open-source drag-and-drop builder for LLM flows and agents; GitHub marks the main repository as archived with its last push in Aug 2026. | Reported (not run by us) | https://theagenticweekly.com/tools/flowise |
| 31 | smolagents v1.26.0 (Hugging Face) | Hugging Face's small Python library for agents that write and run code; the docs say the core logic fits in roughly a thousand lines. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/smolagents |
| 32 | Letta v0.16.8 (Letta) | Platform and open-source framework for stateful agents that keep and improve memory across sessions. | Reported (not run by us) | https://theagenticweekly.com/tools/letta |
| 33 | AutoGen python-v0.7.5 (Microsoft) | Microsoft's open-source multi-agent framework; its README now says the project is in maintenance mode and points to Microsoft Agent Framework. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/autogen |

## Best AI tools for video {#best-ai-video}

**What is the best AI tool for video?** As of 2026-10-26, Agentic Weekly ranks Luma #1 for video: Multiple video models in one board, plus professional outputs like EXR and keyframe control. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-26 · Full list: https://theagenticweekly.com/best/ai-video

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Luma (Luma AI) | A good pick for creators who want several video models in one workspace; Ray 3.2 launched on June 9, 2026, so check which models are live on your plan. | Reported (not run by us) | https://theagenticweekly.com/tools/luma |
| 2 | Runway (Runway) | A strong choice if you want one place to drive many video, image and audio models from an agent; many items are paid-plan only. | Reported (not run by us) | https://theagenticweekly.com/tools/runway |
| 3 | Kling AI 4.0 (Kuaishou (Kling AI)) | Worth testing for action and camera-heavy shots; our information is secondhand from Luma plus the Kling homepage. | Reported (not run by us) | https://theagenticweekly.com/tools/kling |
| 4 | ComfyUI v0.42.1 (Comfy Org) | The tool to learn if you want maximum control over open models on your own GPU; it has a learning curve. | Tested (smoke test, 2026-10-26, no LLM involved) | https://theagenticweekly.com/tools/comfyui |
| 5 | Descript (Descript) | AI video and audio editor where you edit by editing text; Free plan, paid plans measured in media hours and AI credits. | Reported (not run by us) | https://theagenticweekly.com/tools/descript |
| 6 | HeyGen (HeyGen) | AI avatar video with a Free plan and paid tiers at $29, $49 and $149 a month. | Reported (not run by us) | https://theagenticweekly.com/tools/heygen |
| 7 | invideo (invideo) | AI video editor with agents; plans at $10, $30 and $75 per seat a month billed annually. | Reported (not run by us) | https://theagenticweekly.com/tools/invideo |
| 8 | OpusClip (OpusClip) | Turns long videos into shorts and posts to social platforms; Free forever plan. | Reported (not run by us) | https://theagenticweekly.com/tools/opus-clip |
| 9 | Pika (Pika) | AI creative platform with a free plan and monthly tiers at $10, $35 and $95. | Reported (not run by us) | https://theagenticweekly.com/tools/pika |
| 10 | Synthesia (Synthesia) | Business video platform with AI avatars in 140+ languages and a free credit allowance. | Reported (not run by us) | https://theagenticweekly.com/tools/synthesia |

## Best AI tools for writing {#best-writing}

**What is the best AI tool for writing?** As of 2026-10-25, Agentic Weekly ranks ChatGPT #1 for writing: Writing blocks, Pages, Slides, audio uploads and Word integration make it a general writing workbench. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/writing

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | ChatGPT (OpenAI) | Still the broadest all-round assistant; the October updates matter most if you write documents, decks or work with audio. | Reported (not run by us) | https://theagenticweekly.com/tools/chatgpt |
| 2 | Claude (Anthropic) | A strong writing partner; try Claude Duet only if you are on Pro or Max and comfortable with a research-preview product. | Reported (not run by us) | https://theagenticweekly.com/tools/claude |
| 3 | Gemini Notebook (Google) | A good fit for turning a pile of documents into drafts and decks you can check against sources; availability of new features is rolling out. | Reported (not run by us) | https://theagenticweekly.com/tools/gemini-notebook |
| 4 | Copy.ai (Copy.ai) | Go-to-market AI platform now priced by workflow credits. | Reported (not run by us) | https://theagenticweekly.com/tools/copy-ai |
| 5 | Grammarly (Grammarly) | AI writing assistant that works across apps; Free $0, Pro $12 per month, Enterprise custom. | Reported (not run by us) | https://theagenticweekly.com/tools/grammarly |
| 6 | Jasper (Jasper) | AI platform for marketing teams with brand governance and AI visibility tracking; 7-day free trial. | Reported (not run by us) | https://theagenticweekly.com/tools/jasper |
| 7 | Writesonic (Writesonic) | Platform for tracking and improving visibility in AI search with article generation. | Reported (not run by us) | https://theagenticweekly.com/tools/writesonic |

## Best AI tools for music {#best-music}

**What is the best AI tool for music?** As of 2026-10-25, Agentic Weekly ranks Suno #1 for music: Full-song generation with plain-language section edits; v6 launched Sep 9. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/music

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Suno v6 (Suno) | The most complete song-making tool we track; the v6 editing features are the draw. Rights and licensing questions are yours to check. | Reported (not run by us) | https://theagenticweekly.com/tools/suno |
| 2 | ElevenLabs (ElevenLabs) | Best for builders adding voice or music to products; the Video to Music API is the interesting new piece for creators. | Reported (not run by us) | https://theagenticweekly.com/tools/elevenlabs |
| 3 | ACE-Step v1.5 (ACE-Step) | The pick if you want open weights and local control; expect a manual setup and no recent updates. | Reported (not run by us) | https://theagenticweekly.com/tools/ace-step |

## Best AI tools for images {#best-image-generation}

**What is the best AI tool for images?** As of 2026-10-26, Agentic Weekly ranks Midjourney #1 for images: Widely used image generator; this update adds prompt-correcting reruns and shared folders. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-26 · Full list: https://theagenticweekly.com/best/image-generation

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Midjourney (Midjourney) | Use the thinking rerun if prompt-following misses are your pain; be careful with privacy when sharing folders. | Reported (not run by us) | https://theagenticweekly.com/tools/midjourney |
| 2 | ComfyUI v0.42.1 (Comfy Org) | The tool to learn if you want maximum control over open models on your own GPU; it has a learning curve. | Tested (smoke test, 2026-10-26, no LLM involved) | https://theagenticweekly.com/tools/comfyui |
| 3 | ChatGPT (OpenAI) | Still the broadest all-round assistant; the October updates matter most if you write documents, decks or work with audio. | Reported (not run by us) | https://theagenticweekly.com/tools/chatgpt |
| 4 | Runway (Runway) | A strong choice if you want one place to drive many video, image and audio models from an agent; many items are paid-plan only. | Reported (not run by us) | https://theagenticweekly.com/tools/runway |
| 5 | FLUX (Black Forest Labs) (Black Forest Labs) | Black Forest Labs' FLUX image models, available through an API with per-image pricing; the flux GitHub repository is Apache-2.0. | Reported (not run by us) | https://theagenticweekly.com/tools/black-forest-labs |
| 6 | Freepik AI (Freepik) | Stock library with AI suite and credit plans; credits valid for 1 year. | Reported (not run by us) | https://theagenticweekly.com/tools/freepik |
| 7 | Ideogram (Ideogram) | AI image generator with Plus, Pro and Team plans, pay-as-you-go credits at 1 cent each, and an API. | Reported (not run by us) | https://theagenticweekly.com/tools/ideogram |
| 8 | Krea (Krea) | Creative AI suite for images, videos and upscaling; Free plan with 5,000 units a month. | Reported (not run by us) | https://theagenticweekly.com/tools/krea |
| 9 | Leonardo.Ai (Leonardo.Ai) | Image and video creation platform with a free tier, four individual plans, team seats and API pay-as-you-go. | Reported (not run by us) | https://theagenticweekly.com/tools/leonardo-ai |
| 10 | Recraft (Recraft) | Text-to-image model and design platform for photorealism and vector output; V4.1 Flash is listed as the latest model. | Reported (not run by us) | https://theagenticweekly.com/tools/recraft |

## Best AI tools for small businesses {#best-small-business}

**What is the best AI tool for small business?** As of 2026-10-25, Agentic Weekly ranks Goodcall #1 for small business: Lets a trade business answer, qualify and book calls with a simple per-customer price. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/small-business

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Goodcall (Goodcall) | A sensible first AI receptionist for a one-person shop; the per-customer pricing is simple, but test call quality yourself. | Reported (not run by us) | https://theagenticweekly.com/tools/goodcall |
| 2 | Jobber Receptionist (Jobber) | Best if you already use or plan to use Jobber for scheduling and invoicing; the receptionist works with your Jobber data. | Reported (not run by us) | https://theagenticweekly.com/tools/jobber |
| 3 | Lovable (Lovable) | A friendly way for non-developers to get a working web app and a first-customers plan; check credit costs before you commit. | Reported (not run by us) | https://theagenticweekly.com/tools/lovable |
| 4 | n8n 2.45 (n8n) | Great fit for small teams that want to wire apps and AI together without code; the license is fair-code, not open source. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/n8n |
| 5 | Replit (Replit) | Strong pick if you want build, hosting and monitoring in one place; read the credit and overage rules first. | Reported (not run by us) | https://theagenticweekly.com/tools/replit |
| 6 | ChatGPT (OpenAI) | Still the broadest all-round assistant; the October updates matter most if you write documents, decks or work with audio. | Reported (not run by us) | https://theagenticweekly.com/tools/chatgpt |

## Best AI assistants {#best-ai-assistants}

**What is the best AI tool for AI assistants?** As of 2026-10-25, Agentic Weekly ranks Grok #1 for AI assistants: Assistant from SpaceXAI with real-time web and X answers. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/ai-assistants

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Grok (SpaceXAI (formerly xAI)) | Grok is SpaceXAI's chat assistant, free to try on web and in apps, with a SuperGrok upgrade. | Reported (not run by us) | https://theagenticweekly.com/tools/grok |
| 2 | Microsoft Copilot (Microsoft) | Microsoft's AI assistant for chat, creating and getting work done, with consumer and enterprise tiers. | Reported (not run by us) | https://theagenticweekly.com/tools/microsoft-copilot |
| 3 | Mistral Le Chat (Mistral AI) | Mistral's AI assistant: Free, Pro at $14.99 a month, Team at $24.99 per user. | Reported (not run by us) | https://theagenticweekly.com/tools/le-chat |
| 4 | Notion AI (Notion) | Notion's built-in AI and agents; Free $0, Plus $10 and Business $20 per member per month, with AI usage drawing on credits. | Reported (not run by us) | https://theagenticweekly.com/tools/notion-ai |

## Best AI search and research tools {#best-ai-search}

**What is the best AI tool for AI search and research?** As of 2026-10-25, Agentic Weekly ranks Elicit #1 for AI search and research: AI research assistant that searches and synthesizes academic papers. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/ai-search

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Elicit (Elicit) | AI research assistant that searches academic papers, extracts data and drafts reports; free Basic tier with paid per-user plans. | Reported (not run by us) | https://theagenticweekly.com/tools/elicit |
| 2 | Perplexity (Perplexity AI) | AI search with citations in every answer; Pro is $20 a month ($17 billed annually) and Max is $200 ($167 billed annually). | Reported (not run by us) | https://theagenticweekly.com/tools/perplexity |
| 3 | You.com (You.com) | You.com now presents itself as search and inference infrastructure for AI: web search, answer and research APIs plus an inference router. | Reported (not run by us) | https://theagenticweekly.com/tools/you-com |

## Best tools for running LLMs locally {#best-llm-platforms}

**What is the best AI tool for local LLM tools and model platforms?** As of 2026-10-26, Agentic Weekly ranks Cohere #1 for local LLM tools and model platforms: Enterprise-focused models and the North agent platform. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-26 · Full list: https://theagenticweekly.com/best/llm-platforms

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Cohere (Cohere) | Enterprise AI company: Command models, Rerank and the North platform. | Reported (not run by us) | https://theagenticweekly.com/tools/cohere |
| 2 | DeepSeek (DeepSeek) | Model API that works with the OpenAI and Anthropic SDK formats and can back agent tools such as Claude Code, GitHub Copilot and OpenCode. | Reported (not run by us) | https://theagenticweekly.com/tools/deepseek |
| 3 | Hugging Face (Hugging Face) | The hub where the ML community shares models, datasets and apps, with PRO, Team and Enterprise tiers. | Tested (smoke test, 2026-10-26, SmolLM2-135M from the Hub) | https://theagenticweekly.com/tools/hugging-face |
| 4 | LiteLLM v1.106.4 (BerriAI) | Useful as a single interface in front of local and hosted models; check the license files and pin the version because it releases daily. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/litellm |
| 5 | llama.cpp b11889 (ggml-org (open-source project)) | LLM inference in C/C++: the open-source engine many local-model tools build on; MIT-licensed with near-daily build tags. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/llama-cpp |
| 6 | LM Studio (LM Studio) | Desktop app for running open models locally; the homepage now leads with Bionic, an agent for work and code. | Reported (not run by us) | https://theagenticweekly.com/tools/lm-studio |
| 7 | Ollama v0.40.11 (Ollama) | The common way to run open models locally on macOS, Windows and Linux, with a cloud option and an MIT-licensed open-source core. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/ollama |
| 8 | Open WebUI v0.12.4 (Open WebUI) | Self-hosted AI platform: a web interface that connects to Ollama and OpenAI-compatible models, under its own Open WebUI License. | Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/open-webui |
| 9 | OpenRouter (OpenRouter) | Unified API to many models with fallbacks; a Free tier with 25+ free models and 4 free providers. | Reported (not run by us) | https://theagenticweekly.com/tools/openrouter |
| 10 | Replicate (Replicate) | Cloud API for running and fine-tuning open models, with per-use pricing. | Reported (not run by us) | https://theagenticweekly.com/tools/replicate |
| 11 | Together AI (Together AI) | AI cloud for open models: serverless inference, fine-tuning and GPU clusters, priced per token. | Reported (not run by us) | https://theagenticweekly.com/tools/together-ai |
| 12 | Jan 0.8.7 (Menlo Research) | Open-source alternative to ChatGPT that runs open models locally or connects to cloud models like GPT and Claude. | Reported (not run by us) | https://theagenticweekly.com/tools/jan |
| 13 | GPT4All v3.10.0 (Nomic) | Local chatbot from Nomic that runs open-source models on your device; the latest GitHub release we saw is v3.10.0 from Feb 2025. | Tested (smoke test, 2026-10-26, local 3B model (Qwen2.5-3B)) | https://theagenticweekly.com/tools/gpt4all |

## Best AI automation tools {#best-no-code-automation}

**What is the best AI tool for no-code automation?** As of 2026-10-25, Agentic Weekly ranks Make #1 for no-code automation: Visual automation platform with scenarios, AI agents, an MCP server and a free plan. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/no-code-automation

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Make (Make) | Visual automation platform: build scenarios by prompt or drag and drop, credits-based billing, a free plan with no time limit. | Reported (not run by us) | https://theagenticweekly.com/tools/make |
| 2 | Pipedream (Pipedream) | Integration layer for AI agents with managed auth and 10,000+ tools across 3,000+ APIs; free to build, pay in production. | Reported (not run by us) | https://theagenticweekly.com/tools/pipedream |
| 3 | Power Automate (Microsoft) | Microsoft's automation platform with a 30-day free trial and per-user and per-bot plans. | Reported (not run by us) | https://theagenticweekly.com/tools/power-automate |
| 4 | Zapier (Zapier) | No-code automation with 9,000+ apps and AI agents; free plan of 100 tasks a month. | Reported (not run by us) | https://theagenticweekly.com/tools/zapier |
| 5 | Activepieces 0.92.3-hotfix.1 (Activepieces) | Open-source workflow automation with chat, agents, tables and 770+ apps; $20 or $200 flat monthly plans. | Reported (not run by us) | https://theagenticweekly.com/tools/activepieces |

## Best MCP servers for AI agents {#best-mcp-servers}

**What is the best AI tool for MCP servers?** As of 2026-10-26, Agentic Weekly ranks Figma MCP server #1 for MCP servers: Figma's MCP server that gives agents design context. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-26 · Full list: https://theagenticweekly.com/best/mcp-servers

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Figma MCP server (Figma) | Figma's own MCP server that gives coding agents access to design files. | Reported (not run by us) | https://theagenticweekly.com/tools/figma-mcp |
| 2 | Firecrawl v2.11.608 (Firecrawl) | Web data API: search, scrape and crawl for AI agents. | Reported (not run by us) | https://theagenticweekly.com/tools/firecrawl |
| 3 | MCP reference servers v2.1.0 (Model Context Protocol project) | The MCP project's collection of reference servers. | Tested (smoke test, 2026-10-25, no LLM involved) | https://theagenticweekly.com/tools/mcp-servers |
| 4 | Context7 @upstash/context7-sdk 0.6.0 (Upstash) | Up-to-date, version-specific library docs served to coding agents. | Tested (smoke test, 2026-10-25, no language model involved) | https://theagenticweekly.com/tools/context7 |
| 5 | Chrome DevTools MCP v1.12.0 (Chrome DevTools team (Google)) | Chrome DevTools for coding agents. | Tested (smoke test, 2026-10-25, no language model involved) | https://theagenticweekly.com/tools/chrome-devtools-mcp |
| 6 | GitHub MCP Server 2.0.3 (GitHub) | GitHub's official MCP server for repos, issues and pull requests. | Tested (smoke test, 2026-10-26, no LLM involved) | https://theagenticweekly.com/tools/github-mcp-server |
| 7 | MCP Inspector 2.12.0 (Model Context Protocol project) | The quickest way to check what an MCP server exposes; run it on a current Node or npm may quietly install version 0.15.0. | Tested (smoke test, 2026-10-25, no LLM involved) | https://theagenticweekly.com/tools/mcp-inspector |
| 8 | Supabase MCP (Supabase) | Connect Supabase projects to AI assistants. | Tested (smoke test, 2026-10-26, no LLM involved) | https://theagenticweekly.com/tools/supabase-mcp |
| 9 | Sentry MCP cli 0.47.0 (Sentry) | Sentry's agentic tooling and MCP server for errors and traces. | Tested (smoke test, 2026-10-26, no LLM involved) | https://theagenticweekly.com/tools/sentry-mcp |
| 10 | Playwright MCP v0.0.83 (Microsoft) | Microsoft's MCP server that lets agents drive a browser with Playwright. | Tested (smoke test, 2026-10-25, no language model involved) | https://theagenticweekly.com/tools/playwright-mcp |
| 11 | Notion MCP Server v2.5.2 (Notion) | Official Notion MCP server for reading and editing workspace content. | Tested (smoke test, 2026-10-26, no LLM involved) | https://theagenticweekly.com/tools/notion-mcp |

## Best AI meeting assistants {#best-meeting-assistants}

**What is the best AI tool for AI meeting assistants?** As of 2026-10-25, Agentic Weekly ranks Fathom #1 for AI meeting assistants: Free AI notetaker for Zoom, Meet and Teams. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/meeting-assistants

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Fathom (Fathom) | AI notetaker for Zoom, Google Meet and Microsoft Teams that is free for individuals. | Reported (not run by us) | https://theagenticweekly.com/tools/fathom |
| 2 | Fireflies.ai (Fireflies) | AI meeting assistant with a free plan; paid plans $10 to $39 per seat per month billed annually. | Reported (not run by us) | https://theagenticweekly.com/tools/fireflies |
| 3 | Granola (Granola) | AI notepad for back-to-back meetings that captures device audio without a meeting bot. | Reported (not run by us) | https://theagenticweekly.com/tools/granola |
| 4 | Otter.ai (Otter.ai) | AI meeting agent with real-time transcription, summaries and action items; free plan, paid per-user plans. | Reported (not run by us) | https://theagenticweekly.com/tools/otter |
| 5 | Read AI (Read AI) | AI notetaker that works across Teams, Meet and Zoom; Free plan covers 5 meetings per month. | Reported (not run by us) | https://theagenticweekly.com/tools/read-ai |

## Best AI design and presentation tools {#best-ai-design}

**What is the best AI tool for AI design and presentation tools?** As of 2026-10-25, Agentic Weekly ranks Beautiful.ai #1 for AI design and presentation tools: AI presentation software with Smart Slides. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/ai-design

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | Beautiful.ai (Beautiful.ai) | AI presentation platform with 300+ Smart Slide layouts; 14-day free trial. | Reported (not run by us) | https://theagenticweekly.com/tools/beautiful-ai |
| 2 | Canva AI (Canva) | Canva's AI image generator inside its design editor; free to try with plans for Pro, Business and Enterprise. | Reported (not run by us) | https://theagenticweekly.com/tools/canva-ai |

## Best AI voice and speech tools {#best-voice-ai}

**What is the best AI tool for AI voice and speech?** As of 2026-10-25, Agentic Weekly ranks AssemblyAI #1 for AI voice and speech: Speech-to-text and speech understanding APIs. Evidence: Reported (not run by us). This is an editorial pick, not a benchmark result (see how picks are chosen at https://theagenticweekly.com/ai#ranking-basis).

Last updated: 2026-10-25 · Full list: https://theagenticweekly.com/best/voice-ai

| Rank | Tool | Verdict | Evidence | Link |
|---|---|---|---|---|
| 1 | AssemblyAI (AssemblyAI) | Speech AI APIs for transcription and understanding; start free, pay as you go. | Reported (not run by us) | https://theagenticweekly.com/tools/assemblyai |
| 2 | Cartesia (Cartesia) | Realtime voice AI: text-to-speech, speech-to-text in 44 languages and voice agents, with a free plan and credit tiers. | Reported (not run by us) | https://theagenticweekly.com/tools/cartesia |
| 3 | Deepgram (Deepgram) | Voice AI APIs for speech-to-text, text-to-speech and agents with $200 in free credit. | Reported (not run by us) | https://theagenticweekly.com/tools/deepgram |
| 4 | Murf AI (Murf AI) | Text-to-speech platform with a $10 free API trial, pay-as-you-go credits and per-character API pricing. | Reported (not run by us) | https://theagenticweekly.com/tools/murf-ai |
| 5 | Resemble AI (Resemble AI) | Resemble AI now leads with real-time deepfake detection and watermarking for audio, video and images; plans from free to $1,000 a month. | Reported (not run by us) | https://theagenticweekly.com/tools/resemble-ai |
| 6 | Retell AI (Retell AI) | Voice AI agents for phone calls with pay-per-minute pricing and 20 free concurrent calls. | Reported (not run by us) | https://theagenticweekly.com/tools/retell |
| 7 | Speechify (Speechify) | Text-to-speech app that reads PDFs, books and web pages aloud, with a Free plan and a $29 monthly premium price. | Reported (not run by us) | https://theagenticweekly.com/tools/speechify |
| 8 | Vapi (Vapi) | Developer platform for voice agents; $5 free credits, then pay for the minutes you run. | Reported (not run by us) | https://theagenticweekly.com/tools/vapi |
| 9 | Piper v1.8.0 (Open Home Foundation) | Fast, private speech synthesis on a CPU; mind the GPL-3.0 license if you embed it. | Tested (smoke test, 2026-10-25, no LLM involved) | https://theagenticweekly.com/tools/piper |
| 10 | faster-whisper v1.2.1 (SYSTRAN) | Fast local transcription; on CPU use float32 and check int8 against a reference before trusting it. | Tested (smoke test, 2026-10-25, no language model involved) | https://theagenticweekly.com/tools/faster-whisper |

## Newest and best AI models {#models}

As of 2026-10-26, 88 models verified against vendor pages. Hub: https://theagenticweekly.com/models · JSON: https://theagenticweekly.com/models.json · answers: https://theagenticweekly.com/answers.json · MCP: https://theagenticweekly.com/mcp · docs: https://theagenticweekly.com/developers

- **What is the newest AI model release in this dataset?** GPT-6.1 Terra (OpenAI), released Oct 26, 2026. Derived from the release dates in our data; check /models.json for anything newer than the as-of date. (Latest release date among models we verified against vendor sources; models sharing the date are all listed.)
- **What is the latest OpenAI language model?** GPT-6.1 Terra, released Oct 26, 2026. (Newest release date among OpenAI language models we verified.)
- **What is the latest Moonshot AI language model?** Kimi K3.1 Pro, released Oct 23, 2026. (Newest release date among Moonshot AI language models we verified.)
- **What is the latest NVIDIA language model?** Nemotron 3.5 Bolt E4B (4GB), released Oct 23, 2026. (Newest release date among NVIDIA language models we verified.)
- **What is the latest Inception language model?** Mercury 3, released Oct 19, 2026. (Newest release date among Inception language models we verified.)
- **What is the latest MiniMax language model?** MiniMax M3.1 Flash, released Oct 19, 2026. (Newest release date among MiniMax language models we verified.)
- **What is the latest DeepSeek language model?** DeepSeek-V4.1-Flash-1017, released Oct 17, 2026. (Newest release date among DeepSeek language models we verified.)
- **What is the latest Meta language model?** Muse Spark Open, released Oct 15, 2026. (Newest release date among Meta language models we verified.)
- **What is the latest Google language model?** Gemini 4 Argon, released Oct 13, 2026. (Newest release date among Google language models we verified.)
- **What is the latest Anthropic language model?** Claude Haiku 5.5, released Oct 7, 2026. (Newest release date among Anthropic language models we verified.)
- **What is the latest Mistral AI language model?** Mistral Large 4, released Oct 6, 2026 (public preview). (Newest release date among Mistral AI language models we verified.)
- **What is the latest xAI language model?** Grok 4.7, released Sep 21, 2026. (Newest release date among xAI language models we verified.)
- **What is the latest Cohere language model?** North Small Translate, released Sep 9, 2026. (Newest release date among Cohere language models we verified.)
- **What is the latest Z.ai language model?** GLM-5.3-Flash, released Aug 26, 2026. (Newest release date among Z.ai language models we verified.)
- **What is the latest Alibaba language model?** Qwen3.8-27B, released Aug 14, 2026. (Newest release date among Alibaba language models we verified.)
- **What is the newest image AI model in this dataset?** Nano Banana 2.1 (Google), released Oct 6, 2026. (Newest release date among image models we verified.)
- **What is the newest voice AI model in this dataset?** Voxtral 3 (Mistral AI), released Oct 20, 2026. (Newest release date among voice models we verified.)
- **What is the newest video AI model in this dataset?** Gemini Omni 1.1 Flash (Google), released Aug 27, 2026. (Newest release date among video models we verified.)
- **What is the best AI model for coding right now?** Claude Opus 5.5; next: GPT-6 Astra, Claude Sonnet 5.5. (Highest-ranked model on the LMArena WebDev leaderboard (snapshot 2026-10-22) that is also in our model list.)
- **Which AI model is ranked first for general chat preference right now?** Gemini 4 Argon; next: Claude Opus 5.5. (Rank 1 on the LMArena Text Arena (snapshot 2026-10-08).)
- **What is the best open-weights AI model right now?** DeepSeek-V4.1-Flash-1017; next: Kimi K3, GLM-5.3. (Highest-ranked open-weights model in our list on the LMArena WebDev leaderboard (snapshot 2026-10-22).)
- **What is the best open-weights model small enough to run on one machine?** Qwen3.8-27B; next: Gemma 4. (Highest-ranked open-weights model of 31B parameters or fewer in our list on the LMArena WebDev leaderboard (snapshot 2026-10-22).)
- **What is the best AI image generation model right now?** GPT-Image-2.5 Sunburst; next: Nano Banana 2.1. (Rank 1 on the LMArena Text-to-Image leaderboard (snapshot 2026-10-07).)
- **What is the best AI model for agents right now?** No single answer. We found no neutral public agent leaderboard we could verify this week. Candidates: Claude Opus 5.5, GPT-6.1 Sol, Kimi K3. (We do not name one. We found no neutral public agent leaderboard we could verify this week.)
- **What is the best AI voice model right now?** No single answer. We found no neutral public voice leaderboard we could verify this week. Candidates: GPT-Live 1, Gemini 3.8 Live. (We do not name one. We found no neutral public voice leaderboard we could verify this week.)
- **What is the best AI video generation model right now?** No single answer. We verified only one video model against vendor sources. Candidates: Gemini Omni 1.1 Flash. (We do not name one. We verified only one video model against vendor sources.)
- **What is the cheapest AI language model API in this dataset?** Ministral 3 14B at $0.2 in / $0.2 out per 1M tokens. DeepSeek-V4.1-Flash is priced in CNY and is not compared here. (Lowest sum of USD list price per 1M input plus 1M output tokens among language models with a USD price on the vendor's page. Cached rates, tiers and free tiers are ignored.)
- **Which AI models have a context window of 1M tokens or more?** GPT-6.1 Terra, Kimi K3.1 Pro, Kimi K3.1 Pro Highspeed, Mercury 3, MiniMax M3.1 Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Gemini 4 Argon, Muse Wonder, Claude Haiku 5.5, Mistral Large 4, GPT-6.1 Sol, Claude Sonnet 5.5, Claude Opus 5.5, GPT-6 Luna, Claude Fable 5.1, DeepSeek-V4.1-Flash, GPT-6 Astra, Gemini 3.8 Flash, Claude Mythos 5.1, GLM-5.3, Claude Opus 5, Gemini 3.5 Flash-Lite, Gemini 3.6 Flash, Kimi K3, GPT-5.6 Luna, GPT-5.6 Sol, GPT-5.6 Terra, Claude Sonnet 5, GLM-5.2, Claude Fable 5, Claude Mythos 5, Claude Opus 4.8, Gemini 3.1 Flash-Lite, GPT-5.5, GPT-5.5 Pro, Grok 4.20, GPT-5.4. The largest stated window is 1,050,000 tokens. (Context window stated by the vendor, in tokens (Gemini and GPT-6 list slightly above 1M).)
- **Which models in this dataset have downloadable open weights?** Nemotron 3.5 Bolt E4B (4GB) (OpenMDW-1.1); MiniMax M3.1 Flash (MiniMax Community License); DeepSeek-V4.1-Flash-1017 (MIT); Muse Spark Open (Muse Model License (limited commercial terms)); DeepSeek-V4.1-Flash (MIT); North Small Translate (CC BY-NC 4.0); GLM-5.3 (GLM-5.3 License (MIT-style plus a MaaS condition)); Qwen3.8-27B (Apache-2.0); Qwen3.8-2.4T-A95B (Qwen3.8-Max License (custom)); Kimi K3 (Kimi K3 License (custom)); Cohere Transcribe Arabic (Apache 2.0); North Mini Code (Apache 2.0); Command A+ (Apache 2.0); Mistral Medium 3.5 (Modified MIT); Qwen3.6-27B (Apache-2.0); Gemma 4 (Apache 2.0); Mistral Small 4 (Apache 2.0); Ministral 3 14B (Apache 2.0). (Weights published for download; the licence is shown because terms differ and some limit commercial use.)

## Latest news briefs {#news}

Fast, sourced briefs; all times UTC. Full list: https://theagenticweekly.com/news · JSON: https://theagenticweekly.com/latest.json

- 2026-10-26: [OpenAI GPT-6.1 Terra: a $1/$5 computer-use model, and what the launch post leaves out](https://theagenticweekly.com/news/openai-gpt-6-1-terra-computer-use) (Reported; first seen 2026-10-26 19:05 UTC; source published time not stated): GPT-6.1 Terra (gpt-6.1-terra) is OpenAI's new fast model for computer use and visual reasoning: $1 input and $5 output per 1M tokens, 1.05M context, live in ChatGPT Work, Codex and the API on Oct 26. Its headline claims rest on charts we could not read as text, and we have not run it.
- 2026-10-26: [Superset is now Paradev: parallel coding-agent app ships desktop v2.0.0 and an $11.5M seed](https://theagenticweekly.com/news/superset-renamed-paradev-desktop-2-0-seed) (Reported; first seen 2026-10-26 17:05 UTC; source published 2026-10-26 16:13 UTC): Superset, a desktop workspace for running Claude Code, Codex and other CLI coding agents in parallel, was renamed Paradev on Oct 26, 2026 with desktop release v2.0.0; the company says it raised an $11.5M seed led by Union Square Ventures, and that accounts, settings and the superset CLI command carry over unchanged.
- 2026-10-26: [Anthropic files S-1 for an IPO on Nasdaq under "ANT", reporting $16.3 billion revenue in six months](https://theagenticweekly.com/news/anthropic-s-1-ipo-filing-nasdaq-ant) (Reported; first seen 2026-10-26 15:15 UTC; source published 2026-10-26 13:04 UTC): Anthropic, PBC filed a Form S-1 with the SEC on October 26, 2026 for an initial public offering of Class A common stock, applying to list on the Nasdaq Global Select Market under the symbol ANT; the filing reports revenue of $16.3 billion for the six months ended June 30, 2026, and leaves the price range and share count blank.
- 2026-10-26: [ARTEX: an open-source multi-agent LLM system for autonomous penetration testing in isolated labs](https://theagenticweekly.com/news/artex-multi-agent-ai-penetration-testing) (Reported; first seen 2026-10-26 15:35 UTC; source published 2026-10-08 20:35 UTC): ARTEX is an AGPL-3.0 open-source system in which a planner agent and worker agents, driven by Anthropic or OpenAI models, carry out penetration tests in an isolated lab, sharing an asset graph and an exploration graph, with human-in-the-loop approval and full traffic recording; it ships as a single Go binary or a Docker Compose stack.
- 2026-10-26: [PaperPilot: an open-source Rust PDF toolkit that runs as an MCP server, CLI and in-browser WASM](https://theagenticweekly.com/news/paperpilot-rust-pdf-engine-mcp-wasm) (Reported; first seen 2026-10-26 12:06 UTC; source published 2026-10-26 03:29 UTC): PaperPilot is an Apache 2.0, pure-Rust PDF toolkit whose single engine runs as an MCP server, a CLI, a local REST API, a desktop app and in the browser via WASM, so AI agents can merge, split, OCR and compress PDFs without uploading files; its latest GitHub release, v0.6.0, is marked as a pre-release.
- 2026-10-26: [Octopus SI OS: an open-source, self-evolving agentic OS with a Rust kernel, released as a research preview](https://theagenticweekly.com/news/octopus-si-os-self-evolving-agentic-os-research-preview) (Reported; first seen 2026-10-26 12:27 UTC; source published 2026-10-09 08:28 UTC): Octopus SI OS is a new open-source (MIT) agentic operating system for macOS 14+ with a roughly 45,000-line Rust kernel, signed in through ChatGPT with Codex CLI as its first interface; its own README calls it a research preview and says that for ordinary coding tasks it adds nothing over Codex CLI.
- 2026-10-26: [SolidCraft: open-source parametric CAD in pure Rust, with an MCP server so AI agents can model parts](https://theagenticweekly.com/news/solidcraft-open-source-rust-cad-mcp-server) (Reported; first seen 2026-10-26 12:27 UTC; source published 2026-10-07 18:57 UTC): SolidCraft is an open-source parametric 3D CAD application written in Rust that runs on the desktop and in the browser, and its solidcraft-cli mcp command is a Model Context Protocol server that lets AI agents such as Claude Code build models; its README calls it early but usable for simple parts.
- 2026-10-26: [MCP reference servers v2.1.0 fixes eight vulnerabilities in filesystem, everything and fetch](https://theagenticweekly.com/news/mcp-servers-v2-1-0-security-fixes) (Reported; first seen 2026-10-26 03:48 UTC; source published 2026-10-25 00:03 UTC): If you run the reference filesystem, everything or fetch MCP servers, update: v2.1.0 fixes eight vulnerabilities, and package managers will not offer the upgrade automatically to date-versioned installs.
- 2026-10-26: [Microsoft open-sources Agensh, a multi-agent harness its authors scaled to 1,024 agents](https://theagenticweekly.com/news/microsoft-agensh-multi-agent-harness-1024-agents) (Reported; first seen 2026-10-26 03:47 UTC; source published 2026-10-23 08:03 UTC): Microsoft has open-sourced Agensh, a multi-agent harness with no central orchestrator that its authors report running with up to 1,024 agents; the benchmark numbers are the authors' own and we have not reproduced them.
- 2026-10-26: [Ollama v0.40.11 adds speech generation on macOS and the d3 decision models](https://theagenticweekly.com/news/ollama-v0-40-11-speech-d3-models) (Reported; first seen 2026-10-26 03:48 UTC; source published 2026-10-24 00:30 UTC): Ollama v0.40.11 can generate speech on macOS, through ollama run or an OpenAI-compatible audio endpoint, and adds the d3 family of decision models in 2B, 4B, 9B and 27B sizes.
- 2026-10-26: [Open WebUI v0.12.4 adds on-demand skill loading and says it contains unlisted security fixes](https://theagenticweekly.com/news/open-webui-v0-12-4-skills-compaction-security) (Reported; first seen 2026-10-26 03:48 UTC; source published 2026-10-25 08:07 UTC): Open WebUI v0.12.4 adds optional on-demand loading of skills and shows when chat context is compacted; its release notes say the release contains security fixes that are not itemised and recommend updating.

## Anthropic IPO coverage

Not investment advice. This site summarizes public filings and cited reporting; everything is labelled Reported and nothing here is Tested. Status: Initial S-1; not yet effective; no amendment (S-1/A) seen Updated 2026-10-26T16:34:21Z. Every figure links to the SEC filing or cited reporting. JSON: https://theagenticweekly.com/anthropic-ipo.json

- [Anthropic IPO: what the S-1 says (proposed Nasdaq ticker ANT)](https://theagenticweekly.com/anthropic-ipo) (markdown: https://theagenticweekly.com/anthropic-ipo.md)
- [What Anthropic's S-1 says, in plain English](https://theagenticweekly.com/anthropic-ipo/s1-explained) (markdown: https://theagenticweekly.com/anthropic-ipo/s1-explained.md)
- [Anthropic S-1 key numbers](https://theagenticweekly.com/anthropic-ipo/key-numbers) (markdown: https://theagenticweekly.com/anthropic-ipo/key-numbers.md)
- [Anthropic S-1 risk factors that matter](https://theagenticweekly.com/anthropic-ipo/risk-factors) (markdown: https://theagenticweekly.com/anthropic-ipo/risk-factors.md)
- [Anthropic IPO FAQ](https://theagenticweekly.com/anthropic-ipo/faq) (markdown: https://theagenticweekly.com/anthropic-ipo/faq.md)
- [Anthropic IPO timeline](https://theagenticweekly.com/anthropic-ipo/timeline) (markdown: https://theagenticweekly.com/anthropic-ipo/timeline.md)

## Latest releases {#latest}

- 2026-10-25: [Goodcall](https://theagenticweekly.com/tools/goodcall) (Reported (not run by us)): A sensible first AI receptionist for a one-person shop; the per-customer pricing is simple, but test call quality yourself.
- 2026-10-25: [Jobber Receptionist](https://theagenticweekly.com/tools/jobber) (Reported (not run by us)): Best if you already use or plan to use Jobber for scheduling and invoicing; the receptionist works with your Jobber data.
- 2026-10-24: [LangGraph 1.2.17](https://theagenticweekly.com/tools/langgraph) (Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B))): A dependable foundation for stateful agents; this release is a routine fix, so only update if the RemoteGraph or recursion fixes affect you.
- 2026-10-23: [Claude Code v2.1.310](https://theagenticweekly.com/tools/claude-code) (Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models)): Update now if you use the sandbox or run Claude Code on a cloud provider; skip if you wanted flashy new features, this is a hardening release.
- 2026-10-23: [Codex CLI 0.164.0](https://theagenticweekly.com/tools/codex-cli) (Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models)): A quality-of-life release: worth updating if you re-type the same prompts or switch models a lot, not a reason to switch tools.
- 2026-10-23: [CrewAI 1.15.30](https://theagenticweekly.com/tools/crewai) (Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models)): Reasonable if you like the crew-of-agents mental model; the memory and tool-loop fixes in this release are worth taking.
- 2026-10-23: [Gemini Notebook](https://theagenticweekly.com/tools/gemini-notebook) (Reported (not run by us)): A good fit for turning a pile of documents into drafts and decks you can check against sources; availability of new features is rolling out.
- 2026-10-23: [goose v1.56.0](https://theagenticweekly.com/tools/goose) (Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models)): Pick goose if you want an open-source agent with real subagent controls; v1.56.0 is a solid, subagent-heavy update.
- 2026-10-23: [Luma](https://theagenticweekly.com/tools/luma) (Reported (not run by us)): A good pick for creators who want several video models in one workspace; Ray 3.2 launched on June 9, 2026, so check which models are live on your plan.
- 2026-10-23: [Replit](https://theagenticweekly.com/tools/replit) (Reported (not run by us)): Strong pick if you want build, hosting and monitoring in one place; read the credit and overage rules first.

## How picks are chosen {#ranking-basis}

Ranks are editorial judgement made by Agentic Weekly when an entry is written or updated. They are not benchmark results, user ratings, popularity counts or paid positions.

- We rank only the 139 tools we track, not the whole market. "#1" means top of the tools we cover for that task, not best overall.
- The reason for each rank is the one-line blurb shown beside it. Typical reasons are how well the tool's own docs say it fits the task and how actively it is maintained according to its dated release notes.
- There is no numeric scoring formula, and we have not run quality head-to-head comparisons between ranked tools; our research studies measure install size, startup time and speed for some tools, and those numbers do not set ranks. Where a rank looks like a verdict on quality, it is our reading of official sources, not a measured result.
- The Tested label does not raise a rank. Tested means a small smoke test (install plus a short task, usually on a small local model), not a benchmark. Reported and Tested entries are ranked from the same kind of evidence.
- Payment cannot change a rank. Featured listings are labelled and are never used to order a list.
- Prices and plan details are shown only with a date and an official source. Where we could only find a third-party figure, or could not read the official page, the entry says so.
- Ranks can change when a release, price or doc changes. Each page shows when it was last verified.

## Full answers {#full-answers}

The full ranked answers are the best-of pages under https://theagenticweekly.com/best; the same Q&A for software is in https://theagenticweekly.com/ai.json and https://theagenticweekly.com/answers.json.

## More from this site {#more}

- [About Agentic Weekly](https://theagenticweekly.com/about) (markdown: https://theagenticweekly.com/about.md): Agentic Weekly is an independent weekly radar of AI and agent tool releases with verdicts, operated independently from Texas, US.
- [Changelog: site and data changes](https://theagenticweekly.com/changelog) (markdown: https://theagenticweekly.com/changelog.md): A dated log of changes to Agentic Weekly's site, evidence labels, rankings basis and data, so you can see what changed and why.
- [How to cite Agentic Weekly](https://theagenticweekly.com/cite) (markdown: https://theagenticweekly.com/cite.md): Copy-ready citations for Agentic Weekly pages and data in plain text, APA-style, MLA-style, BibTeX and HTML, plus the rules for quoting a verdict or a model fact.
- [Contact Agentic Weekly](https://theagenticweekly.com/contact) (markdown: https://theagenticweekly.com/contact.md): How to contact Agentic Weekly: one email address for corrections, questions and press, plus forms for submitting a tool or asking about a featured listing.
- [How is Agentic Weekly different from There's An AI For That, Futurepedia and TLDR AI?](https://theagenticweekly.com/how-we-differ) (markdown: https://theagenticweekly.com/how-we-differ.md): How Agentic Weekly differs from There's An AI For That, Futurepedia and TLDR AI, using only what each site says about itself and what we do, dated 2026-10-25.
- [Methodology: how Agentic Weekly picks and tests AI tools](https://theagenticweekly.com/methodology) (markdown: https://theagenticweekly.com/methodology.md): How Agentic Weekly chooses tools, what Reported and Tested mean, the scope of our tests, how ranks and prices are handled, and how corrections work.
- [AI model release calendar: confirmed dates and labeled rumors](https://theagenticweekly.com/models/upcoming) (markdown: https://theagenticweekly.com/models/upcoming.md): Upcoming AI model dates from vendor pages only: Kimi K3.1 Pro weights by Nov 4, Ignite Nov 17-20, Gemini 4 Argon price change Dec 1. Rumors labeled.
- [Research: original measurements from our own test rig](https://theagenticweekly.com/research) (markdown: https://theagenticweekly.com/research.md): 5 data studies built from our own test runs: frameworks, MCP servers, local models, voice, and install traps. Each has a data table and states what was not measured.
- [Test logs: what we ran and what happened](https://theagenticweekly.com/tests) (markdown: https://theagenticweekly.com/tests.md): 57 dated test logs of AI agents, frameworks, MCP servers and local AI tools, with exact commands, versions and raw output.
- [New AI and agent tool releases this week](https://theagenticweekly.com/this-week) (markdown: https://theagenticweekly.com/this-week.md): 18 AI tool and model releases from Oct 21, 2026 to Oct 27, 2026, each with a one-line verdict or sourced facts, an evidence label (Reported or Tested) and a setup guide.
- [Install size and a first tool-calling task across 16 agent frameworks: what we measured](https://theagenticweekly.com/research/agent-framework-install-and-first-task) (markdown: https://theagenticweekly.com/research/agent-framework-install-and-first-task.md): We installed 16 Python and 4 JavaScript agent frameworks and ran one tool task on each with a local 3B model. Installs ranged from 26 to 819 MB.
- [MCP servers over stdio: startup time, call latency and what broke when we ran 15 of them](https://theagenticweekly.com/research/mcp-servers-latency-and-failures) (markdown: https://theagenticweekly.com/research/mcp-servers-latency-and-failures.md): We timed 15 MCP server setups over stdio. Most calls took about a millisecond; the surprises were a version change, a loopback block and a hosted lookup.
- [npm on an older Node installs old versions without telling you: four measured cases and other install traps](https://theagenticweekly.com/research/older-node-silent-downgrades) (markdown: https://theagenticweekly.com/research/older-node-silent-downgrades.md): On Node 20, npm quietly installed older major versions of four AI packages; we list the versions, the traps in Python installs, and the one test setting that fooled us.
- [Ollama and llama.cpp on a CPU-only box: three Qwen2.5 sizes, same prompts](https://theagenticweekly.com/research/ollama-vs-llama-cpp-qwen-sizes) (markdown: https://theagenticweekly.com/research/ollama-vs-llama-cpp-qwen-sizes.md): We ran Qwen2.5-Instruct at 0.5B, 1.5B and 3B (Q4_K_M) through Ollama and llama.cpp on 8 CPU cores and compared pass rates, speed and tool calling.
- [A voice round trip on CPU: Piper speaks, faster-whisper listens, and int8 changes the answer](https://theagenticweekly.com/research/voice-round-trip-piper-faster-whisper) (markdown: https://theagenticweekly.com/research/voice-round-trip-piper-faster-whisper.md): We synthesised 10 sentences with Piper, transcribed them back with faster-whisper in 9 configurations and measured word error rate. float32 was stable; int8 was not.
- [What is agentic AI? Definition, how it works and examples (2026)](https://theagenticweekly.com/what-is/agentic-ai) (markdown: https://theagenticweekly.com/what-is/agentic-ai.md): Agentic AI means AI systems that set goals, plan and act with tools and little human input. Definition, components, vs generative AI and AI agents, read 2026-10-25.
- [What is an AI coding agent? How it differs from autocomplete (2026)](https://theagenticweekly.com/what-is/ai-coding-agent) (markdown: https://theagenticweekly.com/what-is/ai-coding-agent.md): An AI coding agent reads your codebase, edits files, runs commands and tests, and iterates. How it differs from autocomplete, with examples and limits. Read 2026-10-25.
- [What is Aider? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/aider) (markdown: https://theagenticweekly.com/what-is/aider.md): What is Aider? Listed under AI coding agents, from Aider-AI; license: Apache-2.0; 49,752 GitHub stars. Facts from official pages read Oct 25, 2026.
- [What is Browser Use? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/browser-use) (markdown: https://theagenticweekly.com/what-is/browser-use.md): What is Browser Use? Listed under Agent building, from Browser Use; license: MIT; 119,131 GitHub stars. Facts from official pages read Oct 25, 2026.
- [Cartesia pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/cartesia-pricing) (markdown: https://theagenticweekly.com/what-is/cartesia-pricing.md): Cartesia pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [Deepgram pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/deepgram-pricing) (markdown: https://theagenticweekly.com/what-is/deepgram-pricing.md): Deepgram pricing as of Oct 25, 2026 (Pay as you go): plan table, free tier and what the vendor's page does not say.
- [Devin pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/devin-pricing) (markdown: https://theagenticweekly.com/what-is/devin-pricing.md): Devin pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [What is Dify? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/dify) (markdown: https://theagenticweekly.com/what-is/dify.md): What is Dify? Listed under Agent building, from LangGenius; license: Modified Apache 2.0; 159,534 GitHub stars. Facts from official pages read Oct 25, 2026.
- [ElevenLabs pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/elevenlabs-pricing) (markdown: https://theagenticweekly.com/what-is/elevenlabs-pricing.md): ElevenLabs pricing as of Oct 25, 2026 (Credit-based subscription): plan table, free tier and what the vendor's page does not say.
- [What is Firecrawl? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/firecrawl) (markdown: https://theagenticweekly.com/what-is/firecrawl.md): What is Firecrawl? Listed under MCP servers, from Firecrawl; license: AGPL-3.0; pricing: Freemium. Facts from official pages read Oct 25, 2026.
- [GitHub Copilot pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/github-copilot-pricing) (markdown: https://theagenticweekly.com/what-is/github-copilot-pricing.md): GitHub Copilot pricing as of Oct 25, 2026 (Free + paid plans with usage credits): plan table, free tier and what the vendor's page does not say.
- [Grammarly pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/grammarly-pricing) (markdown: https://theagenticweekly.com/what-is/grammarly-pricing.md): Grammarly pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [HeyGen pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/heygen-pricing) (markdown: https://theagenticweekly.com/what-is/heygen-pricing.md): HeyGen pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [What is Hugging Face? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/hugging-face) (markdown: https://theagenticweekly.com/what-is/hugging-face.md): What is Hugging Face? Listed under Local LLMs and model platforms, from Hugging Face; license: Varies by repository. Facts from official pages read Oct 25, 2026.
- [What is Ideogram? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/ideogram) (markdown: https://theagenticweekly.com/what-is/ideogram.md): What is Ideogram? Listed under Image generation, from Ideogram; license: Proprietary. Facts from official pages read Oct 25, 2026.
- [Ideogram pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/ideogram-pricing) (markdown: https://theagenticweekly.com/what-is/ideogram-pricing.md): Ideogram pricing as of Oct 25, 2026 (Freemium, subscription and pay-as-you-go): plan table, free tier and what the vendor's page does not say.
- [Kiro pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/kiro-pricing) (markdown: https://theagenticweekly.com/what-is/kiro-pricing.md): Kiro pricing as of Oct 25, 2026 (Freemium (credit-based)): plan table, free tier and what the vendor's page does not say.
- [What is LangChain? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/langchain) (markdown: https://theagenticweekly.com/what-is/langchain.md): What is LangChain? Listed under Agent building, from LangChain; license: MIT; 148,451 GitHub stars. Facts from official pages read Oct 25, 2026.
- [Mistral Le Chat pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/le-chat-pricing) (markdown: https://theagenticweekly.com/what-is/le-chat-pricing.md): Mistral Le Chat pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [Leonardo.Ai pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/leonardo-ai-pricing) (markdown: https://theagenticweekly.com/what-is/leonardo-ai-pricing.md): Leonardo.Ai pricing as of Oct 25, 2026 (Freemium with subscriptions and pay-as-you-go API): plan table, free tier and what the vendor's page does not say.
- [What is llama.cpp? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/llama-cpp) (markdown: https://theagenticweekly.com/what-is/llama-cpp.md): What is llama.cpp? Listed under Local LLMs and model platforms, from ggml-org; license: MIT; 132,360 GitHub stars. Facts from official pages read Oct 25, 2026.
- [What is LM Studio? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/lm-studio) (markdown: https://theagenticweekly.com/what-is/lm-studio.md): What is LM Studio? Listed under Local LLMs and model platforms, from LM Studio; license: Proprietary application. Facts from official pages read Oct 25, 2026.
- [What is Make? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/make) (markdown: https://theagenticweekly.com/what-is/make.md): What is Make? Listed under No-code automation, from Make; license: Proprietary; pricing: Freemium, credits-based. Facts from official pages read Oct 25, 2026.
- [What is an MCP server? A plain-English guide (2026)](https://theagenticweekly.com/what-is/mcp-server) (markdown: https://theagenticweekly.com/what-is/mcp-server.md): An MCP server meaning, explained from the official MCP docs: how hosts, clients and servers work, local vs remote, popular examples, and what we tested.
- [Mem0 pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/mem0-pricing) (markdown: https://theagenticweekly.com/what-is/mem0-pricing.md): Mem0 pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [Murf AI pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/murf-ai-pricing) (markdown: https://theagenticweekly.com/what-is/murf-ai-pricing.md): Murf AI pricing as of Oct 25, 2026 (Free API trial, pay as you go and enterprise): plan table, free tier and what the vendor's page does not say.
- [What is n8n? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/n8n) (markdown: https://theagenticweekly.com/what-is/n8n.md): What is n8n? Listed under Agent building, from n8n; pricing: Free self-hosted Community Edition + paid plans. Facts from official pages read Oct 25, 2026.
- [Notion AI pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/notion-ai-pricing) (markdown: https://theagenticweekly.com/what-is/notion-ai-pricing.md): Notion AI pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [What is Ollama? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/ollama) (markdown: https://theagenticweekly.com/what-is/ollama.md): What is Ollama? Listed under Local LLMs and model platforms, from Ollama; license: MIT; 184,762 GitHub stars. Facts from official pages read Oct 25, 2026.
- [Ollama pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/ollama-pricing) (markdown: https://theagenticweekly.com/what-is/ollama-pricing.md): Ollama pricing as of Oct 25, 2026 (Free + paid cloud plans): plan table, free tier and what the vendor's page does not say.
- [What is Open WebUI? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/open-webui) (markdown: https://theagenticweekly.com/what-is/open-webui.md): What is Open WebUI? Listed under Local LLMs and model platforms, from Open WebUI; license: Open WebUI License. Facts from official pages read Oct 25, 2026.
- [What is OpenHands? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/openhands) (markdown: https://theagenticweekly.com/what-is/openhands.md): What is OpenHands? Listed under AI coding agents, from OpenHands; license: MIT; 91,839 GitHub stars. Facts from official pages read Oct 25, 2026.
- [What is Perplexity? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/perplexity) (markdown: https://theagenticweekly.com/what-is/perplexity.md): What is Perplexity? Listed under AI search and research, from Perplexity AI; license: Proprietary; pricing: Freemium. Facts from official pages read Oct 25, 2026.
- [Perplexity pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/perplexity-pricing) (markdown: https://theagenticweekly.com/what-is/perplexity-pricing.md): Perplexity pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [Runway pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/runway-pricing) (markdown: https://theagenticweekly.com/what-is/runway-pricing.md): Runway pricing as of Oct 25, 2026 (Credit-based subscription): plan table, free tier and what the vendor's page does not say.
- [v0 pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/v0-pricing) (markdown: https://theagenticweekly.com/what-is/v0-pricing.md): v0 pricing as of Oct 25, 2026 (Free + paid plans with token-based usage): plan table, free tier and what the vendor's page does not say.
- [What is Vapi? Facts, license and pricing (Oct 2026)](https://theagenticweekly.com/what-is/vapi) (markdown: https://theagenticweekly.com/what-is/vapi.md): What is Vapi? Listed under AI voice and speech, from Vapi; license: Proprietary; pricing: Pay as you go. Facts from official pages read Oct 25, 2026.
- [What is vibe coding? Meaning, how it works and the risks (2026)](https://theagenticweekly.com/what-is/vibe-coding) (markdown: https://theagenticweekly.com/what-is/vibe-coding.md): Vibe coding means building software by describing it to an AI and accepting what it writes. Definition, how it works, risks and tools, from sources read 2026-10-25.
- [Warp pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/warp-pricing) (markdown: https://theagenticweekly.com/what-is/warp-pricing.md): Warp pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [Zapier pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/zapier-pricing) (markdown: https://theagenticweekly.com/what-is/zapier-pricing.md): Zapier pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [Zed pricing (Oct 2026): plans and free tier](https://theagenticweekly.com/what-is/zed-pricing) (markdown: https://theagenticweekly.com/what-is/zed-pricing.md): Zed pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- [How to add an MCP server to Claude Code](https://theagenticweekly.com/guides/add-mcp-server-to-claude-code) (markdown: https://theagenticweekly.com/guides/add-mcp-server-to-claude-code.md): The claude mcp add commands for remote HTTP and local stdio servers, scopes, status checks and removal, copied from Claude Code's docs. Reported, not run by us.
- [Best model for OpenCode, Cline and Claude Code: options by budget](https://theagenticweekly.com/guides/best-model-for-opencode-cline-claude-code) (markdown: https://theagenticweekly.com/guides/best-model-for-opencode-cline-claude-code.md): No tested best model for OpenCode, Cline or Claude Code. Sourced options by budget with list prices and how to set the model in each tool.
- [How to build an MCP server with FastMCP](https://theagenticweekly.com/guides/build-an-mcp-server-with-fastmcp) (markdown: https://theagenticweekly.com/guides/build-an-mcp-server-with-fastmcp.md): Install FastMCP, write a three-tool Python MCP server and run it over stdio, using the code from our logged smoke test. HTTP and client setup come from the FastMCP docs.
- [ChatGPT Work vs ChatGPT Chat vs Codex: what each is for](https://theagenticweekly.com/guides/chatgpt-work-vs-chatgpt-chat) (markdown: https://theagenticweekly.com/guides/chatgpt-work-vs-chatgpt-chat.md): ChatGPT Work is OpenAI's agent mode for multi-step tasks with finished files; Chat is for quick answers, Codex for code. Per OpenAI docs.
- [Is Claude Opus 5.5 free? Which plans include it](https://theagenticweekly.com/guides/claude-opus-5-5-free-or-paid) (markdown: https://theagenticweekly.com/guides/claude-opus-5-5-free-or-paid.md): No. Anthropic's pricing page shows Opus as not on the Free plan; it is on Pro, Max, Team and Enterprise, or by API. Checked 2026-10-26.
- [Figma MCP server: how to enable it](https://theagenticweekly.com/guides/figma-mcp-server) (markdown: https://theagenticweekly.com/guides/figma-mcp-server.md): Who can use Figma's MCP server, remote versus desktop, and the documented Claude Code commands. Reported from Figma's docs; we have not run it.
- [Gemini 4 Argon API pricing: $2 and $10 per million tokens until Nov 30, then $4 and $20](https://theagenticweekly.com/guides/gemini-4-argon-api-pricing) (markdown: https://theagenticweekly.com/guides/gemini-4-argon-api-pricing.md): Gemini 4 Argon costs $2 input and $10 output per million tokens through Nov 30, 2026, then $4 and $20. Not on the free tier. Compared with other models.
- [Git MCP server: how to set it up and what it can do](https://theagenticweekly.com/guides/git-mcp-server) (markdown: https://theagenticweekly.com/guides/git-mcp-server.md): How to install the reference Git MCP server (mcp-server-git), the 12 tools it listed and a logged no-model smoke test. Not the same as GitHub's MCP server.
- [GitHub MCP server: how to set it up](https://theagenticweekly.com/guides/github-mcp-server) (markdown: https://theagenticweekly.com/guides/github-mcp-server.md): Set up GitHub's official MCP server in Claude Code, VS Code and other clients: remote or local, token or OAuth, read-only mode. Reported from the README, not run by us.
- [Which coding agents can use GPT-6.1 Terra?](https://theagenticweekly.com/guides/gpt-6-1-terra-in-coding-agents) (markdown: https://theagenticweekly.com/guides/gpt-6-1-terra-in-coding-agents.md): GPT-6.1 Terra is confirmed in GitHub Copilot, Codex and ChatGPT Work. Cursor, Claude Code, Cline and OpenCode are not confirmed. Sourced, 2026-10-26.
- [Is Kimi K3.1 Pro open-weight? Status of the Nov 4 promise](https://theagenticweekly.com/guides/kimi-k3-1-pro-open-weights) (markdown: https://theagenticweekly.com/guides/kimi-k3-1-pro-open-weights.md): Kimi K3.1 Pro open weights: Moonshot promises the full weights by Nov 4, 2026. As of Oct 26, 9:51 PM CDT, no K3.1 Pro repo was on Hugging Face.
- [Playwright MCP server: setup, tools and what we measured](https://theagenticweekly.com/guides/playwright-mcp-server) (markdown: https://theagenticweekly.com/guides/playwright-mcp-server.md): How to install and run Microsoft's Playwright MCP server, which flags we used, the 25 tools it listed and a logged no-model smoke test. Client setup is from the README.
- [How to run Claude Code with a local model](https://theagenticweekly.com/guides/run-claude-code-with-a-local-model) (markdown: https://theagenticweekly.com/guides/run-claude-code-with-a-local-model.md): How to point Claude Code at a local Ollama model, with our logged 2026-10-25 test: a 3B model edited a file on the exact prompt and failed on a vague one.
- [Which Claude model should you use? Haiku 5.5, Sonnet 5.5, Opus 5.5 or Fable 5.1](https://theagenticweekly.com/guides/which-claude-model-should-i-use) (markdown: https://theagenticweekly.com/guides/which-claude-model-should-i-use.md): Anthropic says start with Opus 5.5. Sonnet 5.5 for cheaper everyday work, Haiku 5.5 for volume, Fable 5.1 for the hardest jobs. Prices checked 2026-10-26.
- [Archive: AI tool release news](https://theagenticweekly.com/archive)
- [Open data (CSV, JSON, CC BY 4.0)](https://theagenticweekly.com/data)
- [Compare AI tools head to head](https://theagenticweekly.com/compare)

## Machine-readable versions

- HTML: https://theagenticweekly.com/ai
- Markdown: https://theagenticweekly.com/ai.md
- JSON: https://theagenticweekly.com/ai.json
- All tools (JSON): https://theagenticweekly.com/tools.json
- Open data, CC BY 4.0 (CSV and JSON): https://theagenticweekly.com/data
- llms.txt: https://theagenticweekly.com/llms.txt
- Full content: https://theagenticweekly.com/llms-full.txt

---

## Langflow (2026-10-26, Tested (smoke test, 2026-10-26, no LLM involved))
URL: https://theagenticweekly.com/tools/langflow (markdown: https://theagenticweekly.com/tools/langflow.md)
Vendor: Langflow

Langflow's site calls it a low-code builder for agentic and RAG applications that can deploy agents and MCP servers. The repository is MIT-licensed with dev builds of v1.14.0 published at the end of October 2026.

Verdict: MIT-licensed visual builder for AI agents and MCP servers with a very large GitHub following.
Good for: Visual prototyping of RAG and agents; Exposing flows as MCP servers
Skip if: You want code-first control only

What's new:
- Latest release we saw (2026-10-26): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Langflow
1. Get access: Install Langflow with pip or uv and open the local canvas per the docs.
2. Check the limits and costs first: Free to self-host. Paid: Hosted pricing not read. No pricing page read.

Not verified:
- Hosting options and prices

Test scope:
- smoke test, 2026-10-26; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Langflow 1.13.0 installed in 19.76 s (1.7 GB) and its backend was healthy 91.5 s after launch, serving 26 starter flows and 416 components.
- Not tested: Building or running any flow, including a flow that calls a model; The visual editor: we ran the backend only; Langflow's hosted or desktop offerings and deployment; Start-up time on an idle machine: ours was heavily shared.

Sources:
- Langflow: https://www.langflow.org/
- Langflow repository (GitHub): https://github.com/langflow-ai/langflow
Cite as: Agentic Weekly, "Langflow: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/langflow

## Amazon Q Developer (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/amazon-q-developer (markdown: https://theagenticweekly.com/tools/amazon-q-developer.md)
Vendor: Amazon Web Services

Amazon Q Developer is AWS's coding assistant for your editor. Its page states that on April 30, 2027 AWS will discontinue support for the Amazon Q Developer IDE plugins and points to Kiro for similar capabilities. A free tier and a Pro tier at $19 per user per month are listed.

Verdict: AWS's AI coding assistant for IDEs and the CLI, with a free tier and a $19 Pro tier; IDE plugin support ends April 30, 2027.
Good for: AWS-centric teams already using Q Developer; Teams that need IP indemnity (Pro)
Skip if: You are starting a new project: AWS says IDE plugin support ends April 30, 2027 and points to Kiro; You want an open-source tool

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Amazon Q Developer
1. Get access: Install the Amazon Q Developer plugin for your IDE and sign in with an AWS Builder ID for the free tier.
2. Check the limits and costs first: Free tier: 50 agentic requests per month and 1,000 lines of code per month (Java upgrades) Paid: Pro $19 per user per month: 4,000 lines of code per month per user pooled, IP indemnity, admin controls. Read from aws.amazon.com/q/developer/pricing on 2026-10-25. Extra lines of code at $0.003 per line on Pro.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Quality of suggestions (not tested)
- Future of the CLI after the plugin retirement

Sources:
- Amazon Q Developer: https://aws.amazon.com/q/developer/
- Amazon Q Developer pricing: https://aws.amazon.com/q/developer/pricing/
Cite as: Agentic Weekly, "Amazon Q Developer: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/amazon-q-developer

## AssemblyAI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/assemblyai (markdown: https://theagenticweekly.com/tools/assemblyai.md)
Vendor: AssemblyAI

AssemblyAI's pricing says start free and pay as you go with no commitments, with custom rate limits and concurrency for enterprise. Per-model rates are published but their units were not captured.

Verdict: Speech AI APIs for transcription and understanding; start free, pay as you go.
Good for: Developers adding transcription and speech understanding
Skip if: Non-technical meeting note taking

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with AssemblyAI
1. Get access: Create an API key and run the quickstart.
2. Check the limits and costs first: Free to start. Paid: Per-model usage rates (units not captured); custom enterprise. Read from assemblyai.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Per-model rate units

Sources:
- AssemblyAI: https://www.assemblyai.com/
- AssemblyAI pricing: https://www.assemblyai.com/pricing
Cite as: Agentic Weekly, "AssemblyAI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/assemblyai

## Augment Code (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/augment-code (markdown: https://theagenticweekly.com/tools/augment-code.md)
Vendor: Augment Code

Augment Code's pricing page lists a Standard plan at $20 a month flat with no per-seat charge for up to 50 seats, including $20 of usage, a Business plan at $100 a month with $100 of usage, and a custom Enterprise plan. Features named include the Context Engine, CLI access, MCP and native tools, and SOC 2 Type II.

Verdict: Coding agent and 'Cosmos' platform with a Context Engine, CLI and cloud agents; plans are flat team prices with usage included.
Good for: Small teams that want flat pricing across seats; Teams working on large codebases (Context Engine focus)
Skip if: You want a free tier; You want an open-source agent

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Augment Code
1. Get access: Start a trial from augmentcode.com ('Try Cosmos') and connect your repositories.
2. Check the limits and costs first: No free plan shown on the pricing page. Paid: Standard $20 per month flat (up to 50 seats, $20 usage included); Business $100 per month flat ($100 usage included); Enterprise custom. Read from augmentcode.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Performance on large codebases (vendor claim, not tested)

Sources:
- Augment Code: https://www.augmentcode.com/
- Augment Code pricing: https://www.augmentcode.com/pricing
Cite as: Agentic Weekly, "Augment Code: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/augment-code

## Base44 (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/base44 (markdown: https://theagenticweekly.com/tools/base44.md)
Vendor: Base44

Base44's homepage describes a no-code AI development platform where you describe an app and get a working app or website with its backend, authentication, payments and hosting already included. Plans are priced by monthly message credits and integration credits.

Verdict: No-code AI platform that builds apps and websites from your words, with backend, authentication, payments and hosting built in.
Good for: Non-developers who want a working app with backend included; People who want to publish on a custom domain from a paid plan
Skip if: You expect to iterate heavily on the free plan (25 message credits a month); You want to export and self-host everything

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Base44
1. Get access: Describe your app at base44.com and publish; the free plan needs no payment.
2. Check the limits and costs first: Free: $0, 25 message credits and 100 integration credits a month. Paid: Starter $16/month (100 message credits), Builder $40/month (250), Pro $80/month (500), Elite $160/month (1,200); all billed annually. Read from base44.com/pricing on 2026-10-25. Prices shown are per month when billed annually.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- How many credits a real app uses (not tested)

Sources:
- Base44: https://base44.com/
- Base44 pricing: https://base44.com/pricing
Cite as: Agentic Weekly, "Base44: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/base44

## Beautiful.ai (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/beautiful-ai (markdown: https://theagenticweekly.com/tools/beautiful-ai.md)
Vendor: Beautiful.ai

Beautiful.ai's site lists 300+ auto-aligning Smart Slide layouts, a 14-day free trial of Pro or Team and prices shown include $14.50 a month, and a team plan at $40 or $50 a month including 3 users.

Verdict: AI presentation platform with 300+ Smart Slide layouts; 14-day free trial.
Good for: Teams that need on-brand decks
Skip if: You want a free forever plan

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Beautiful.ai
1. Get access: Start the 14-day trial.
2. Check the limits and costs first: 14-day trial. Paid: $14.50 per month (Pro) as shown; Team $40 to $50 per month including 3 users; Enterprise custom. Read from beautiful.ai pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Which plan each price belongs to

Sources:
- Beautiful.ai: https://www.beautiful.ai/
Cite as: Agentic Weekly, "Beautiful.ai: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/beautiful-ai

## Bolt (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/bolt-new (markdown: https://theagenticweekly.com/tools/bolt-new.md)
Vendor: StackBlitz

Bolt's site says you build websites and apps by chatting with AI, and that it routes each task to the model it picks automatically. Every plan on the pricing page includes website hosting, projects and unlimited databases.

Verdict: Browser-based AI builder that turns a chat prompt into a hosted website or app, with a free plan and token-based tiers.
Good for: First-time builders who want hosting included; Prototypers who want something live from a prompt
Skip if: You need unlimited usage on the free plan (daily token cap); You want to self-host or own the toolchain

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Bolt
1. Get access: Open bolt.new, describe the app, and publish from the same screen; no card is required for the free plan.
2. Check the limits and costs first: Free: $0, 1M tokens a month with a 300K daily limit, Bolt branding, no custom domains. Paid: Pro $18/month billed yearly (10M tokens a month, custom domains, branding removed); Teams $27 per member per month billed yearly; Enterprise custom. Read from bolt.new/pricing on 2026-10-25. Prices shown there are per month when billed annually.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Quality of generated apps (not tested)
- Token cost of a typical app

Sources:
- Bolt: https://bolt.new/
- Bolt pricing: https://bolt.new/pricing
Cite as: Agentic Weekly, "Bolt: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/bolt-new

## Botpress (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/botpress (markdown: https://theagenticweekly.com/tools/botpress.md)
Vendor: Botpress

Botpress's pricing page lists Free ($0), a $150 a month plan with 250 conversations included ($0.65 per extra conversation in packs of 100), a $750 a month plan with 1,500 conversations ($0.50 each in packs) and Enterprise custom.

Verdict: Customer-support AI agent platform with a free plan and conversation-based paid plans.
Good for: Support teams wanting AI agents with per-conversation costs
Skip if: You need a general-purpose agent builder

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Botpress
1. Get access: Sign up free and build a first agent from the dashboard.
2. Check the limits and costs first: Free: $0. Paid: $150 per month with 250 conversations; $750 per month with 1,500 conversations; extra conversations $0.65 or $0.50 each in packs of 100; Enterprise custom. Read from botpress.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)

Sources:
- Botpress: https://botpress.com/
- Botpress pricing: https://botpress.com/pricing
Cite as: Agentic Weekly, "Botpress: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/botpress

## Canva AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/canva-ai (markdown: https://theagenticweekly.com/tools/canva-ai.md)
Vendor: Canva

Canva's AI image generator page says you pick a style and aspect ratio, generate and refine in the editor. Canva's pricing page compares Free, Pro, Business and Enterprise; prices were not captured.

Verdict: Canva's AI image generator inside its design editor; free to try with plans for Pro, Business and Enterprise.
Good for: Non-designers making social and marketing graphics
Skip if: You need advanced model control

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Canva AI
1. Get access: Open Canva and use the AI image tool.
2. Check the limits and costs first: Free plan available. Paid: Pro, Business and Enterprise plans exist; prices not captured. Prices not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Prices
- AI usage limits

Sources:
- Canva AI image generator: https://www.canva.com/ai-image-generator/
Cite as: Agentic Weekly, "Canva AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/canva-ai

## Cartesia (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/cartesia (markdown: https://theagenticweekly.com/tools/cartesia.md)
Vendor: Cartesia

Cartesia's pricing lists Free ($0, 20K credits and $1 of prepaid agent usage a month), $5 a month (100K credits), $49 a month (1.25M credits) and $299 a month (8M credits), plus custom Enterprise. Its site says speech-to-text covers 44 languages.

Verdict: Realtime voice AI: text-to-speech, speech-to-text in 44 languages and voice agents, with a free plan and credit tiers.
Good for: Developers building low-latency voice agents
Skip if: You want a consumer voice-over app

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Cartesia
1. Get access: Create an account and API key from cartesia.ai.
2. Check the limits and costs first: Free: $0 a month, 20K credits and $1 prepaid agents. Paid: $5/month (100K credits, $5 agents); $49/month (1.25M credits, $49 agents); $299/month (8M credits, $299 agents); Enterprise custom. Read from cartesia.ai/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Latency claims (vendor)

Sources:
- Cartesia: https://cartesia.ai/
- Cartesia pricing: https://cartesia.ai/pricing
Cite as: Agentic Weekly, "Cartesia: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/cartesia

## Cohere (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/cohere (markdown: https://theagenticweekly.com/tools/cohere.md)
Vendor: Cohere

Cohere's site highlights Rerank 5 and the North 2 platform update. Its pricing page lists token prices for Command models (for example Command R+ 04-2024 at $3.00 input and $15.00 output per million tokens, Command-light at $0.30 and $0.60) and custom enterprise pricing.

Verdict: Enterprise AI company: Command models, Rerank and the North platform.
Good for: Enterprises needing private deployment and RAG components
Skip if: Hobbyists wanting a free chat app

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Cohere
1. Get access: Create an API key from the Cohere dashboard.
2. Check the limits and costs first: Not captured. Paid: Per-million-token prices by model; Enterprise custom. Read from cohere.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Current model lineup

Sources:
- Cohere: https://cohere.com/
- Cohere pricing: https://cohere.com/pricing
Cite as: Agentic Weekly, "Cohere: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/cohere

## Copy.ai (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/copy-ai (markdown: https://theagenticweekly.com/tools/copy-ai.md)
Vendor: Copy.ai

Copy.ai positions itself as a go-to-market AI platform. Its pricing page shows $29 a month ($24 billed annually, $288 a year, 20K workflow credits), $1,000 a month (45K), $2,000 a month and $3,000 a month tiers billed annually, and Enterprise.

Verdict: Go-to-market AI platform now priced by workflow credits.
Good for: Sales and marketing ops using AI workflows
Skip if: Individuals wanting a simple writer

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Copy.ai
1. Get access: Use the trial or sign up from the pricing page.
2. Check the limits and costs first: Not captured. Paid: Entry plan $29 per month ($24 billed annually; 20K workflow credits); larger plans $1,000 to $3,000 per month billed annually. Read from copy.ai/prices on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan names

Sources:
- Copy.ai: https://www.copy.ai/
- Copy.ai pricing: https://www.copy.ai/prices
Cite as: Agentic Weekly, "Copy.ai: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/copy-ai

## Deepgram (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/deepgram (markdown: https://theagenticweekly.com/tools/deepgram.md)
Vendor: Deepgram

Deepgram's pricing page offers $200 free credit with no card, pay-as-you-go usage, per-minute speech-to-text rates between about $0.004 and $0.009 depending on model, and Enterprise plans.

Verdict: Voice AI APIs for speech-to-text, text-to-speech and agents with $200 in free credit.
Good for: Developers adding transcription or voice agents
Skip if: You want a consumer notetaker

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Deepgram
1. Get access: Sign up for the free credit and use the API quickstart.
2. Check the limits and costs first: $200 free credit, no card. Paid: Per-minute rates by model (about $0.0042 to $0.0092 per minute for speech-to-text on the page); Enterprise custom. Read from deepgram.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Accuracy by language

Sources:
- Deepgram: https://deepgram.com/
- Deepgram pricing: https://deepgram.com/pricing
Cite as: Agentic Weekly, "Deepgram: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/deepgram

## DeepSeek (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/deepseek (markdown: https://theagenticweekly.com/tools/deepseek.md)
Vendor: DeepSeek

DeepSeek's API docs say the API format is compatible with OpenAI and Anthropic, with base URLs api.deepseek.com and api.deepseek.com/anthropic, and list the models deepseek-flash and deepseek-v4-pro. They say Claude Code, GitHub Copilot and OpenCode can use DeepSeek as the backend model, and that DeepSeek Harness is in developer preview. The pricing table did not render in the page text we could read.

Verdict: Model API that works with the OpenAI and Anthropic SDK formats and can back agent tools such as Claude Code, GitHub Copilot and OpenCode.
Good for: Developers who want an OpenAI-compatible endpoint; People wiring a cheaper backend into agent tools
Skip if: You need published prices in this page (not captured); You need a first-party consumer app review (not covered here)

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with DeepSeek
1. Get access: Request an API key from DeepSeek, then point an OpenAI-compatible SDK at https://api.deepseek.com with model deepseek-flash.
2. Check the limits and costs first: Not stated on the page we read. Paid: Per-token API pricing exists but did not render in our capture; check the official pricing page. The docs page we read did not include the pricing table.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Prices
- Model licenses
- Model quality (not tested)

Sources:
- DeepSeek API docs: https://api-docs.deepseek.com/quick_start/pricing-details-usd
Cite as: Agentic Weekly, "DeepSeek: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/deepseek

## Descript (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/descript (markdown: https://theagenticweekly.com/tools/descript.md)
Vendor: Descript

Descript's pricing lists Free ($0, 60 minutes a month), paid plans with 10, 30 and 40 media hours a month and 400, 800 and 1,500 AI credits a month, 1080p and 4K export, and Enterprise.

Verdict: AI video and audio editor where you edit by editing text; Free plan, paid plans measured in media hours and AI credits.
Good for: Podcasters and video editors who prefer text editing
Skip if: You need a full non-linear editor

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Descript
1. Get access: Download Descript or open it in the browser and import a file.
2. Check the limits and costs first: Free: $0, 60 minutes per month. Paid: Paid plans listed by media hours (10, 30, 40) and AI credits (400, 800, 1,500); per-plan dollar prices not captured. Read from descript.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Dollar prices

Sources:
- Descript: https://www.descript.com/
- Descript pricing: https://www.descript.com/pricing
Cite as: Agentic Weekly, "Descript: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/descript

## Devin (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/devin (markdown: https://theagenticweekly.com/tools/devin.md)
Vendor: Cognition

Devin's homepage describes an autonomous software engineer that runs in the cloud or on your machine, tests in its own browser and works until the pull request is ready to merge. The same plans page covers Devin Desktop, Devin Cloud, Devin CLI and Devin Review under the Devin brand.

Verdict: Cognition's autonomous software engineer: a cloud coding agent that tests in its own browser and aims to hand back a mergeable pull request.
Good for: Teams that want to delegate whole tickets to a cloud agent and review PRs; Engineers who want tasks to keep running after they close their laptop
Skip if: You want a low-cost, bring-your-own-model workflow (Aider and OpenHands are open source); You need a tool whose results you can verify yourself before paying: we have not run it

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Devin
1. Get access: Sign up at devin.ai; the Free plan is a download of Devin Desktop and Devin Cloud access starts at Pro ($20/month).
2. Check the limits and costs first: Free plan: $0 (light quota; Devin Cloud is not listed under Free) Paid: Pro $20/month includes access to Devin Cloud; Max $200/month; Teams $80/month plus $40/month per full dev seat; Enterprise custom. Read from devin.ai/pricing on 2026-10-25. Cloud sessions are capped (up to 10 concurrent on Pro and Max). Extra usage is billed at API pricing.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Autonomy and success rate on real tasks (vendor claims, not tested)
- Usage allowance per plan

Sources:
- Devin: https://devin.ai/
- Devin plans and pricing: https://devin.ai/pricing
- Windsurf 2.0: Devin in Windsurf (Cognition blog, April 15, 2026): https://devin.ai/blog/windsurf-2-0
Cite as: Agentic Weekly, "Devin: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/devin

## Elicit (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/elicit (markdown: https://theagenticweekly.com/tools/elicit.md)
Vendor: Elicit

Elicit's site describes an AI tool that takes a research question, searches millions of academic papers and generates reports. Its pricing page lists a free Basic plan with limits, paid Plus, Pro and Scale per-user plans, and a custom Enterprise plan.

Verdict: AI research assistant that searches academic papers, extracts data and drafts reports; free Basic tier with paid per-user plans.
Good for: Researchers doing literature reviews; Teams that need data extraction from papers
Skip if: You need general web search rather than academic sources

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Elicit
1. Get access: Sign up at elicit.com and ask a research question; the Basic plan is free.
2. Check the limits and costs first: Basic: free with limits. Paid: Paid per-user plans (Plus, Pro, Scale) billed monthly or yearly; figures differ between views on the page so we do not quote them; Enterprise custom. Read from elicit.com/pricing on 2026-10-25; the page shows different prices depending on a toggle, so check the page for current figures.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Exact paid prices
- Accuracy of extractions (not tested)

Sources:
- Elicit: https://elicit.com/
- Elicit pricing: https://elicit.com/pricing
Cite as: Agentic Weekly, "Elicit: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/elicit

## Fathom (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/fathom (markdown: https://theagenticweekly.com/tools/fathom.md)
Vendor: Fathom

Fathom's site says it captures, transcribes and summarizes Zoom, Google Meet and Teams calls, is free for individuals, offers CRM updates for teams, and lists SOC 2 Type II, GDPR, HIPAA and SSO/SCIM.

Verdict: AI notetaker for Zoom, Google Meet and Microsoft Teams that is free for individuals.
Good for: Individuals wanting a free notetaker
Skip if: You need non-listed platforms

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Fathom
1. Get access: Sign up and connect your calendar.
2. Check the limits and costs first: Free for individuals (per site) Paid: Team pricing not read. Pricing page not read.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Team pricing

Sources:
- Fathom: https://fathom.video/
Cite as: Agentic Weekly, "Fathom: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/fathom

## Figma MCP server (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/figma-mcp (markdown: https://theagenticweekly.com/tools/figma-mcp.md)
Vendor: Figma

Figma's Help Center guide says the Figma MCP server is available on Starter, Professional, Education, Organization, Government and Enterprise plans.

Verdict: Figma's own MCP server that gives coding agents access to design files.
Good for: Developers generating code from Figma designs
Skip if: You do not use Figma

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Figma MCP server
1. Get access: Follow Figma's guide to enable the MCP server and connect your agent.
2. Check the limits and costs first: Starter plan listed. Plan availability from the Figma guide; seat requirements not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Seat and rate-limit requirements

Sources:
- Figma MCP guide: https://help.figma.com/hc/en-us/articles/32132100833559-Guide-to-the-Figma-MCP-server
Cite as: Agentic Weekly, "Figma MCP server: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/figma-mcp

## Firebase Studio (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/firebase-studio (markdown: https://theagenticweekly.com/tools/firebase-studio.md)
Vendor: Google (Firebase)

The Firebase Studio site banner says Google is sunsetting Firebase Studio on March 22, 2027, so new workspaces can no longer be created, and tells users to move code to Google AI Studio or Google Antigravity. It was described as a web-based workspace with Gemini and app previews.

Verdict: Google's web-based AI workspace for full-stack apps; Google says it is sunsetting Firebase Studio on March 22, 2027 and new workspaces can no longer be created.
Good for: Existing users planning a migration
Skip if: You want to start a new project: new workspaces cannot be created

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Firebase Studio
1. Get access: Existing workspaces only: export your code and follow Google's migration paths to Google AI Studio or Google Antigravity.
2. Check the limits and costs first: Was free for up to 3 workspaces during preview. Read from firebase.studio on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Migration details beyond the banner

Sources:
- Firebase Studio: https://firebase.studio/
Cite as: Agentic Weekly, "Firebase Studio: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/firebase-studio

## Firecrawl v2.11.608 (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/firecrawl (markdown: https://theagenticweekly.com/tools/firecrawl.md)
Vendor: Firecrawl

Firecrawl is a web data API for AI agents that searches and scrapes the web into clean content. Its repository is AGPL-3.0 licensed with v2.11.608 published Oct 25, 2026, and a hosted API has a free plan.

Verdict: Web data API: search, scrape and crawl for AI agents.
Good for: Agents that need clean web content
Skip if: You need a license that avoids AGPL obligations for self-hosting

What's new:
- v2.11.608 (2026-10-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Firecrawl
1. Get access: Create an API key at firecrawl.dev or self-host from the repository.
2. Check the limits and costs first: Free plan with a monthly credit allowance. Paid: Standard listed at $83 per month billed annually; higher tiers and extra credit packs exist. Read from firecrawl.dev/pricing on 2026-10-25; exact plan names and credit counts were not captured.

Not verified:
- Behavior in practice (we have not run this tool)

Sources:
- Firecrawl repository (GitHub): https://github.com/firecrawl/firecrawl
- Firecrawl pricing: https://www.firecrawl.dev/pricing
Cite as: Agentic Weekly, "Firecrawl v2.11.608: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/firecrawl

## Fireflies.ai (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/fireflies (markdown: https://theagenticweekly.com/tools/fireflies.md)
Vendor: Fireflies

Fireflies' pricing page lists Free ($0), a $10 per seat per month plan billed annually ($18 monthly), a $19 annual ($29 monthly) plan and Enterprise at $39 per seat per month billed annually. It also lists AI credits per plan and a free dictation product called Fireflies Talk.

Verdict: AI meeting assistant with a free plan; paid plans $10 to $39 per seat per month billed annually.
Good for: Teams that want searchable meeting history across CRM and chat
Skip if: You want a no-bot capture (see Granola)

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Fireflies.ai
1. Get access: Sign up free and connect your calendar.
2. Check the limits and costs first: Free forever: $0, 20 AI credits. Paid: $10 per seat per month billed annually ($18 monthly); $19 annual ($29 monthly); Enterprise $39 annual. Read from fireflies.ai/pricing on 2026-10-25. Plan names beyond Free and Enterprise were not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Transcription accuracy (not tested)

Sources:
- Fireflies pricing: https://fireflies.ai/pricing
- Fireflies: https://fireflies.ai/
Cite as: Agentic Weekly, "Fireflies.ai: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/fireflies

## FLUX (Black Forest Labs) (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/black-forest-labs (markdown: https://theagenticweekly.com/tools/black-forest-labs.md)
Vendor: Black Forest Labs

Black Forest Labs' site describes visual intelligence models and FLUX through an API, with enterprise volume plans. The black-forest-labs/flux repository is Apache-2.0 and was last pushed in Jul 2025.

Verdict: Black Forest Labs' FLUX image models, available through an API with per-image pricing; the flux GitHub repository is Apache-2.0.
Good for: Developers adding image generation by API
Skip if: You want a consumer app

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with FLUX (Black Forest Labs)
1. Get access: Use the API docs on bfl.ai or the open repository.
2. Check the limits and costs first: Not captured. Paid: API priced per image (a $0.048 figure appears on the pricing page; model not captured); Enterprise custom. Read from bfl.ai pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Model-by-model licenses
- Which model costs $0.048

Sources:
- Black Forest Labs: https://bfl.ai/
- flux repository (GitHub): https://github.com/black-forest-labs/flux
Cite as: Agentic Weekly, "FLUX (Black Forest Labs): what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/black-forest-labs

## Freepik AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/freepik (markdown: https://theagenticweekly.com/tools/freepik.md)
Vendor: Freepik

Freepik's pricing lists plans at $20 ($7 effective), $45 ($33) and $110 ($82) per month with 240K, 600K and 1.5M credits respectively; credits are valid for 1 year with no monthly resets and can be topped up.

Verdict: Stock library with AI suite and credit plans; credits valid for 1 year.
Good for: Creators who want credits that persist for a year
Skip if: You need a free plan

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Freepik AI
1. Get access: Choose a plan from the pricing page.
2. Check the limits and costs first: Not captured. Paid: Entry plans $20 ($7 with discount), $45 ($33), $110 ($82) per month with 240K, 600K and 1.5M credits. Read from freepik.com pricing on 2026-10-25; discount basis not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Discount terms

Sources:
- Freepik: https://www.freepik.com/
Cite as: Agentic Weekly, "Freepik AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/freepik

## Goodcall (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/goodcall (markdown: https://theagenticweekly.com/tools/goodcall.md)
Vendor: Goodcall

Goodcall builds an AI receptionist from a plain-English description, with plans from $79 a month billed by unique customer instead of per minute. This is a snapshot of its site as of Oct 25, 2026; we found no dated changelog.

Verdict: A sensible first AI receptionist for a one-person shop; the per-customer pricing is simple, but test call quality yourself.
Good for: Solo operators who miss calls while on the job; Shops that book work by phone; Owners who want predictable bills
Skip if: You need a human receptionist (a live answering service does that); You expect very high call volume; Enterprise is quote-based

What's new:
- Snapshot (no dated release notes found): Goodcall's agent builder, Gabi, creates and edits your receptionist from plain-English instructions.
- Pricing page: Starter $79/month (100 unique customers, then 79 cents each), Growth $129/month (500 customers, then 26 cents), Scale $299/month (2,000 customers, then 15 cents).
- No per-minute charges; every plan includes a local number, and you can keep your own number by forwarding it.
- Start free with a demo agent and pay when you connect a number.

Tutorial: Set up a Goodcall receptionist
1. Start with the free demo agent: Goodcall says you can build, test and hear the demo agent answer before paying.
2. Describe your business in plain English: Tell Gabi what to collect (name, issue, address) and when to book an estimate; connect a calendar or CRM tool.
3. Add a number when ready: Use the local number every plan includes, or forward your existing number.

Not verified:
- Release date (none published on the pages we fetched)
- Call quality and accuracy (not tested)
- Whether prices have changed since Oct 25, 2026

Sources:
- Goodcall homepage: https://www.goodcall.com/
- Goodcall pricing: https://www.goodcall.com/pricing
Cite as: Agentic Weekly, "Goodcall: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/goodcall

## Grammarly (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/grammarly (markdown: https://theagenticweekly.com/tools/grammarly.md)
Vendor: Grammarly

Grammarly's pricing lists Free ($0, 100 prompts per month), Pro ($12 per month as shown, 2,000 prompts per member per month) and Enterprise (custom, unlimited prompts, key management).

Verdict: AI writing assistant that works across apps; Free $0, Pro $12 per month, Enterprise custom.
Good for: Anyone polishing emails and documents
Skip if: You want long-form AI drafting only

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Grammarly
1. Get access: Install the extension and sign up.
2. Check the limits and costs first: Free: $0 per month, 100 prompts per month. Paid: Pro $12 per month as listed; 2,000 prompts per member per month; Enterprise custom. Read from grammarly.com/plans on 2026-10-25. Billing period behind the $12 figure not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Billing period for $12

Sources:
- Grammarly: https://www.grammarly.com/
- Grammarly plans: https://www.grammarly.com/plans
Cite as: Agentic Weekly, "Grammarly: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/grammarly

## Granola (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/granola (markdown: https://theagenticweekly.com/tools/granola.md)
Vendor: Granola

Granola's site describes an AI notepad that gives notes, actions and memory without a meeting bot. Its pricing page lists Free at $0 per user per month, a $14 per user per month plan, a $35 per user per month tier and Enterprise.

Verdict: AI notepad for back-to-back meetings that captures device audio without a meeting bot.
Good for: People who dislike bots joining calls
Skip if: You need a shared team transcript bot

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Granola
1. Get access: Download Granola and connect your calendar.
2. Check the limits and costs first: Free: $0. Paid: Paid listed at $14 and $35 per user per month (plan names not captured); Enterprise custom. Read from granola.ai/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan names

Sources:
- Granola: https://www.granola.ai/
- Granola pricing: https://www.granola.ai/pricing
Cite as: Agentic Weekly, "Granola: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/granola

## Grok (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/grok (markdown: https://theagenticweekly.com/tools/grok.md)
Vendor: SpaceXAI (formerly xAI)

Grok's site says it can chat, create images, write code and give real-time answers from the web and X, free to try with SuperGrok for higher limits and multi-agent reasoning. Reuters reported SpaceX acquired xAI in February 2026.

Verdict: Grok is SpaceXAI's chat assistant, free to try on web and in apps, with a SuperGrok upgrade.
Good for: People who want live X and web context
Skip if: You need published plan prices

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Grok
1. Get access: Open grok.com and start chatting.
2. Check the limits and costs first: Free to try on web and in apps. Paid: SuperGrok upgrade (price not captured) Read from grok.com on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- SuperGrok price

Sources:
- Grok: https://grok.com/
- Reuters on SpaceX acquiring xAI: https://www.reuters.com/business/musks-spacex-merge-with-xai-combined-valuation-125-trillion-bloomberg-news-2026-02-02/
Cite as: Agentic Weekly, "Grok: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/grok

## Gumloop (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/gumloop (markdown: https://theagenticweekly.com/tools/gumloop.md)
Vendor: Gumloop

Gumloop's site describes a multiplayer AI agent builder where anyone builds agents and IT controls access. Pricing starts at $37 per month with 20k credits per month, and Enterprise is custom.

Verdict: Agent builder for teams; plans start at $37 per month with 20k included credits.
Good for: Teams letting non-engineers build agents with IT controls
Skip if: You want open source

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Gumloop
1. Get access: Start the free trial.
2. Check the limits and costs first: Free trial. Paid: From $37 per month with 20k credits; Enterprise custom. Read from gumloop.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)

Sources:
- Gumloop: https://www.gumloop.com/
- Gumloop pricing: https://www.gumloop.com/pricing
Cite as: Agentic Weekly, "Gumloop: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/gumloop

## HeyGen (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/heygen (markdown: https://theagenticweekly.com/tools/heygen.md)
Vendor: HeyGen

HeyGen's pricing lists Free ($0), $29 a month (600 credits, rollover one month), $49 a month (1,000 credits) and $149 a month (1,500 credits, team seats at $20 each), plus Enterprise; the site also mentions a HeyGen plugin for ChatGPT.

Verdict: AI avatar video with a Free plan and paid tiers at $29, $49 and $149 a month.
Good for: Marketers and trainers making avatar videos
Skip if: You need cinematic text-to-video

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with HeyGen
1. Get access: Sign up free and create a Personal Avatar.
2. Check the limits and costs first: Free: $0 a month. Paid: $29/month (600 credits); $49/month (1,000 credits); $149/month (1,500 credits, $20 per extra seat); Enterprise custom. Read from heygen.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan names

Sources:
- HeyGen: https://www.heygen.com/
- HeyGen pricing: https://www.heygen.com/pricing
Cite as: Agentic Weekly, "HeyGen: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/heygen

## Hugging Face (2026-10-25, Tested (smoke test, 2026-10-26, SmolLM2-135M from the Hub))
URL: https://theagenticweekly.com/tools/hugging-face (markdown: https://theagenticweekly.com/tools/hugging-face.md)
Vendor: Hugging Face

Hugging Face's pricing lists PRO at $9 a month, Team at $20 per user a month, a $50 per user a month tier beside Enterprise, storage rates per TB, on-demand GPU hardware starting at $0, and dedicated inference starting at $0.033 an hour.

Verdict: The hub where the ML community shares models, datasets and apps, with PRO, Team and Enterprise tiers.
Good for: Anyone sourcing open models and datasets
Skip if: You want a single polished chat app

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Hugging Face
1. Get access: Create a free account at huggingface.co.
2. Check the limits and costs first: Free hub account. Paid: PRO $9/month; Team $20 per user per month; a $50 per user per month tier is listed beside Enterprise; paid compute and storage. Read from huggingface.co/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Free-tier quotas

Test scope:
- smoke test, 2026-10-26; model: SmolLM2-135M-Instruct downloaded from the Hugging Face Hub and run with the transformers library on CPU (not the Qwen2.5-3B used in other tests); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Hugging Face's Hub searched without a login in 0.1 s, a 272 MB model downloaded in 3.61 s, and transformers ran it on CPU.
- Not tested: Gated models, private repositories, uploads, Spaces and Inference Providers: all need an account or token; Download speed for large (multi-GB) models; Any model beyond one 135M-parameter model; quality of its output was not assessed; Timing of generation: the machine was heavily shared.

Sources:
- Hugging Face: https://huggingface.co/
- Hugging Face pricing: https://huggingface.co/pricing
Cite as: Agentic Weekly, "Hugging Face: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/hugging-face

## Ideogram (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/ideogram (markdown: https://theagenticweekly.com/tools/ideogram.md)
Vendor: Ideogram

Ideogram's pricing page lists Plus ($20 a month, $15 a month billed annually), Pro ($60, or $42 billed annually) and Team ($30 per user a month, $20 billed annually), plus pay-as-you-go credits at 1 cent each and custom Enterprise plans. It says there are no free trials or discounts on subscription plans, and eligible free accounts get weekly credits.

Verdict: AI image generator with Plus, Pro and Team plans, pay-as-you-go credits at 1 cent each, and an API.
Good for: People who want a hosted image generator with a clear credit plan; Developers who want credit-based API access
Skip if: You need a free trial of a paid plan (the page says none is offered); You need local, open-weight models

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Ideogram
1. Get access: Open ideogram.ai and sign in with Google, Apple or Microsoft to receive any free weekly credits.
2. Check the limits and costs first: Weekly credits for eligible free accounts (amount varies) Paid: Plus $20/month ($15 billed annually); Pro $60/month ($42 annually); Team $30 per user per month ($20 annually); pay as you go 1 cent per credit; Enterprise custom. Read from ideogram.ai/pricing on 2026-10-25. Priority credits listed: Pro 7,500 a month, Team 3,600 per user.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Image quality compared with other generators (not tested)
- Credit cost per image by model (the page loads it dynamically)

Sources:
- Ideogram pricing: https://ideogram.ai/pricing
Cite as: Agentic Weekly, "Ideogram: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/ideogram

## invideo (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/invideo (markdown: https://theagenticweekly.com/tools/invideo.md)
Vendor: invideo

invideo's pricing lists 400, 2,000 and 5,000 credits per seat a month at $10, $30 and $75 per seat billed annually ($20, $60, $150 monthly).

Verdict: AI video editor with agents; plans at $10, $30 and $75 per seat a month billed annually.
Good for: Video teams buying seats with credit pools
Skip if: You want a permanent free tier

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with invideo
1. Get access: Pick a plan from the pricing page.
2. Check the limits and costs first: Not captured. Paid: 400 credits $10/seat/month annual ($20 monthly); 2,000 credits $30 annual ($60 monthly); 5,000 credits $75 annual ($150 monthly) Read from invideo.io pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Free tier

Sources:
- invideo: https://invideo.io/
Cite as: Agentic Weekly, "invideo: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/invideo

## Jasper (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/jasper (markdown: https://theagenticweekly.com/tools/jasper.md)
Vendor: Jasper

Jasper's site describes agents that run marketing workflows, brand-governed content at scale and measurement across AI answer engines. It offers a free 7-day trial and notes a 20 percent non-profit discount; plan prices were not captured.

Verdict: AI platform for marketing teams with brand governance and AI visibility tracking; 7-day free trial.
Good for: Marketing teams needing brand control
Skip if: Solo writers on a budget

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Jasper
1. Get access: Start the 7-day trial.
2. Check the limits and costs first: 7-day free trial. Prices were not captured from the pricing page.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan prices

Sources:
- Jasper: https://www.jasper.ai/
Cite as: Agentic Weekly, "Jasper: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/jasper

## Jobber Receptionist (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/jobber (markdown: https://theagenticweekly.com/tools/jobber.md)
Vendor: Jobber

Jobber's Receptionist answers calls and texts, books visits, takes work requests and escalates urgent calls; a 'Receptionist Companion' that drafts quotes while you take a call is marked coming soon. This is a snapshot as of Oct 25, 2026.

Verdict: Best if you already use or plan to use Jobber for scheduling and invoicing; the receptionist works with your Jobber data.
Good for: Home service businesses (plumbing, HVAC, landscaping and similar) on Jobber; Owners who want calls turned into work requests and booked visits; Businesses that want escalation rules for urgent calls
Skip if: You do not use Jobber and want a standalone receptionist; You expect Receptionist to be free; it is a $29/month add-on on select plans (included on Plus)

What's new:
- Snapshot (no dated release notes found): Receptionist answers calls and texts 24/7, books visits using your existing online booking settings and creates work requests with details to quote.
- Call escalation: you choose urgent keywords such as burst pipe or no heat and it puts the caller through or texts you an alert.
- Conversation log with summary, transcript and recording; you can watch text conversations live and take over.
- Coming soon: Receptionist Companion listens alongside you and drafts the quote or schedules the visit you just discussed.
- Pricing: Jobber's help article says Receptionist is a $29/month add-on on select plans that includes 30 conversations per billing cycle; extra conversations cost $0.79 each, or you can pause at the limit. It is included with unlimited usage on the Plus plan (article updated July 9, 2026).

Tutorial: Try Jobber Receptionist
1. Hear it first: Jobber lists demo phone numbers (US and Canada) on the Receptionist page so you can call and test it.
2. Start a free trial: The page offers a free trial of Jobber. Confirm whether Receptionist is included in your plan.
3. Set your urgent keywords: Choose keywords that mean urgent for your business so those callers are transferred or alerted by text.

Not verified:
- Exactly which plans can add Receptionist (the help article says select plans)
- Call accuracy (not tested)
- Release date (none published on the page)

Sources:
- Jobber Receptionist page: https://www.getjobber.com/features/ai-receptionist/
- Jobber pricing: https://www.getjobber.com/pricing/
- Jobber help center: Receptionist powered by Jobber AI: https://help.getjobber.com/hc/en-us/articles/25315927533847-Receptionist-powered-by-Jobber-AI
Cite as: Agentic Weekly, "Jobber Receptionist: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/jobber

## Kiro (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/kiro (markdown: https://theagenticweekly.com/tools/kiro.md)
Vendor: Amazon Web Services

Kiro's site says it turns prompts into requirements, architectural designs and sequenced tasks, then implements them, and that prompts run in vibe mode or spec mode. There is a CLI that can also run headless in CI/CD, per the site. Amazon's own page says AWS will discontinue Amazon Q Developer IDE plugins on April 30, 2027 and points to Kiro for similar capabilities.

Verdict: AWS's agentic coding tool that turns prompts into requirements, designs and tasks (spec mode), with an IDE and a CLI.
Good for: Developers who want structured specs before code is written; AWS users moving off Amazon Q Developer IDE plugins
Skip if: You want an open-source tool (Kiro is proprietary); You want to avoid credit accounting

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Kiro
1. Get access: Download Kiro from kiro.dev, or install the CLI with the installer on the site, then sign in with a social login or AWS Builder ID.
2. Check the limits and costs first: Kiro Free: $0 a month, 50 credits. Paid: Pro $20, Pro+ $40, Pro Max $100, Power $200, all per user per month (credit allotments per plan not captured); add-on credits $0.04 each; Enterprise on request. Read from kiro.dev/pricing on 2026-10-25. The page offers $20 of credit on first upgrade using social login or AWS Builder ID.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Quality of generated specs and code (not tested)
- Supported OS list (not read)

Sources:
- Kiro: https://kiro.dev/
- Kiro pricing: https://kiro.dev/pricing/
- Amazon Q Developer: https://aws.amazon.com/q/developer/
Cite as: Agentic Weekly, "Kiro: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/kiro

## Krea (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/krea (markdown: https://theagenticweekly.com/tools/krea.md)
Vendor: Krea

Krea's pricing lists Free ($0, 5,000 units a month), plans at $9 ($5 annual) for 20,000 units, $35 ($21) for 60,000 units, $105 ($63) and $200 ($160) tiers and Enterprise, plus one-time compute packs.

Verdict: Creative AI suite for images, videos and upscaling; Free plan with 5,000 units a month.
Good for: Creators who want image, video and upscaling in one place
Skip if: You want open-weight local generation

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Krea
1. Get access: Sign up free.
2. Check the limits and costs first: Free: $0, 5,000 units per month. Paid: $9 per month ($5 annual) for 20,000 units; $35 ($21 annual) for 60,000 units; higher tiers up to $200 per month; Enterprise custom. Read from krea.ai/pricing on 2026-10-25. Plan names not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan names

Sources:
- Krea: https://www.krea.ai/
- Krea pricing: https://www.krea.ai/pricing
Cite as: Agentic Weekly, "Krea: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/krea

## Leonardo.Ai (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/leonardo-ai (markdown: https://theagenticweekly.com/tools/leonardo-ai.md)
Vendor: Leonardo.Ai

Leonardo.Ai's pricing page lists Free ($0, 150 fast tokens a day), Essential ($12 a month, 8,500 tokens), Premium ($30, 25,000 tokens, unlimited relaxed image generation on selected models) and Ultimate ($60, 60,000 tokens, plus relaxed video on selected models), team plans at $24 and $48 per seat, and pay-as-you-go API access. It says paid subscribers own the images they generate.

Verdict: Image and video creation platform with a free tier, four individual plans, team seats and API pay-as-you-go.
Good for: Creators who want many models in one subscription; Teams that want shared token pools and private generations
Skip if: You need unlimited generations on every model (relaxed mode covers selected models only); You want free-tier images to stay private (Free creations are public)

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Leonardo.Ai
1. Get access: Open leonardo.ai and start on the Free plan to use the daily tokens.
2. Check the limits and costs first: Free: 150 fast tokens a day, auto model only, public creations. Paid: Essential $12/month; Premium $30/month; Ultimate $60/month; team Starter $24 per seat and Growth $48 per seat (shared token pools); pay-as-you-go API; prices exclude tax. Read from leonardo.ai/pricing on 2026-10-25. Yearly billing is up to 20% off; the monthly prices are quoted here.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Output quality by model (not tested)
- Token cost per generation

Sources:
- Leonardo.Ai pricing: https://leonardo.ai/pricing/
Cite as: Agentic Weekly, "Leonardo.Ai: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/leonardo-ai

## Lindy (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/lindy (markdown: https://theagenticweekly.com/tools/lindy.md)
Vendor: Lindy

Lindy's pricing page lists Free with $50 in credits (valid 7 days, no card), paid plans from about $49.99 a month in credit tiers (5,000 to 1,750,000 credits), top-ups at $10 per 1,000 credits, and custom Enterprise billing.

Verdict: AI 'employee' builder that connects to your tools; Free plan with $50 in credits that last 7 days, then credit-based plans.
Good for: Non-technical teams automating assistant-style work
Skip if: You want open-source or a fixed price

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Lindy
1. Get access: Sign up for the free plan and create an agent from a template.
2. Check the limits and costs first: Free: $50 in credits that last 7 days. Paid: From $49.99 per month for 5,000 credits; higher credit tiers listed up to 1.75M; top-ups $10 per 1,000 credits; Enterprise custom. Read from lindy.ai/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)

Sources:
- Lindy: https://www.lindy.ai/
- Lindy pricing: https://www.lindy.ai/pricing
Cite as: Agentic Weekly, "Lindy: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/lindy

## LiteLLM v1.106.4 (2026-10-25, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/litellm (markdown: https://theagenticweekly.com/tools/litellm.md)
Vendor: BerriAI

LiteLLM is a Python SDK and gateway that calls many model providers through an OpenAI-compatible interface. The GitHub license field reads NOASSERTION, so check the repository's license files; v1.106.4 was tagged on 2026-10-25.

Verdict: Useful as a single interface in front of local and hosted models; check the license files and pin the version because it releases daily.
Good for: Teams that call several model providers; Anyone pointing tools at a local OpenAI-compatible server
Skip if: You only ever use one provider and its own SDK

What's new:
- v1.106.4 (2026-10-25): the newest release on the repository when we read it.

Tutorial: Get started with LiteLLM
1. Install: We installed the SDK; its MCP integration needed the mcp package, which was not installed automatically.
2. Point at a local server: Use an openai/ model name and the local base URL.

Not verified:
- Proxy server, budgets, hosted LiteLLM pricing
- Behaviour with hosted models

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: LiteLLM 1.106.4 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 2.3 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark; The LiteLLM proxy server, budgets, keys and routing.

Sources:
- LiteLLM documentation: https://docs.litellm.ai/docs/
- LiteLLM repository (GitHub): https://github.com/BerriAI/litellm
Cite as: Agentic Weekly, "LiteLLM v1.106.4: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/litellm

## llama.cpp b11889 (2026-10-25, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/llama-cpp (markdown: https://theagenticweekly.com/tools/llama-cpp.md)
Vendor: ggml-org (open-source project)

The llama.cpp repository describes LLM inference in C/C++ and is MIT-licensed. Its releases are build tags, with b11889 published Oct 25, 2026.

Verdict: LLM inference in C/C++: the open-source engine many local-model tools build on; MIT-licensed with near-daily build tags.
Good for: Developers embedding local inference; People who want maximum control over local models
Skip if: You want a polished app with a GUI

What's new:
- b11889 (2026-10-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with llama.cpp
1. Get access: Clone the repository or download a release build and follow the README's quick start.
2. Check the limits and costs first: Free. Open-source project; no pricing.

Not verified:
- Performance on your hardware

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: llama.cpp b11889 generated 118 tokens/s (0.5B), 40 (1.5B) and 19.5 (3B) on 8 CPU threads with llama-bench, and its server passed the same prompts and tool call as Ollama.
- Not tested: Building from source, GPU backends, longer prompts, quantisations other than Q4_K_M, speculative decoding; Single bench run of 64 tokens; it is a smoke test and not a benchmark suite.

Sources:
- llama.cpp repository (GitHub): https://github.com/ggml-org/llama.cpp
Cite as: Agentic Weekly, "llama.cpp b11889: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/llama-cpp

## LM Studio (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/lm-studio (markdown: https://theagenticweekly.com/tools/lm-studio.md)
Vendor: LM Studio

LM Studio's homepage describes Bionic as LM Studio's agent for open models: natively local, built for creativity, work and code, able to create and edit documents, run coding tasks and automations, and transcribe speech in real time.

Verdict: Desktop app for running open models locally; the homepage now leads with Bionic, an agent for work and code.
Good for: People who want a desktop app for local models
Skip if: You need an open-source app (see Jan or Ollama)

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with LM Studio
1. Get access: Download from lmstudio.ai and pick a model inside the app.
2. Check the limits and costs first: Pricing not published on the page we read. No pricing page was read; check lmstudio.ai.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Pricing
- License
- Platforms

Sources:
- LM Studio: https://lmstudio.ai/
Cite as: Agentic Weekly, "LM Studio: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/lm-studio

## Make (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/make (markdown: https://theagenticweekly.com/tools/make.md)
Vendor: Make

Make's homepage describes an AI automation platform with a visual workflow builder, AI agents, custom code (JavaScript or Python) and an MCP server, and says it connects 3,500+ apps out of the box. Its pricing page says credits are the billing unit and the Free plan has 1,000 credits a month. Paid plan prices were not shown in the text we read.

Verdict: Visual automation platform: build scenarios by prompt or drag and drop, credits-based billing, a free plan with no time limit.
Good for: People who want a visual builder with a free plan; Teams that want an MCP server to run automations from an AI chat
Skip if: You need published per-plan prices before deciding (not captured here); You want to self-host

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Make
1. Get access: Open make.com and choose Try Make free.
2. Check the limits and costs first: Free plan: 1,000 credits a month, no card required, no time limit. Paid: Core, Pro, Teams and Enterprise plans (prices not captured); extra credit bundles of 1,000 or 10,000; Start-up plan gives 50% off Business. Read from make.com and make.com/en/pricing on 2026-10-25. Plan prices did not appear in the text we could read.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan prices
- Real number of apps (pages disagree)

Sources:
- Make: https://www.make.com/en
- Make pricing: https://www.make.com/en/pricing
Cite as: Agentic Weekly, "Make: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/make

## MCP reference servers v2.1.0 (2026-10-25, Tested (smoke test, 2026-10-25, no LLM involved))
URL: https://theagenticweekly.com/tools/mcp-servers (markdown: https://theagenticweekly.com/tools/mcp-servers.md)
Vendor: Model Context Protocol project

The modelcontextprotocol/servers repository holds the MCP project's reference servers. Its license file says the project is moving from MIT to Apache-2.0, with docs under CC-BY-4.0; v2.1.0 was released Oct 25, 2026.

Verdict: The MCP project's collection of reference servers.
Good for: Learning MCP server patterns; Starting points for custom servers
Skip if: You need a polished vendor-supported integration

What's new:
- v2.1.0 (2026-10-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with MCP reference servers
1. Get access: Pick a server from the repository and follow its README.
2. Check the limits and costs first: Free. Open-source server; the service it connects to may have its own costs.

Test scope:
- smoke test, 2026-10-25; model: none (no LLM was involved in this test); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: MCP Memory server answered over stdio in 284 ms cold (v1.0.0, Node 20) and 147 ms (v2.0.0, Node 24); median call 0.9 to 0.6 ms. The MCP Fetch server 2026.10.24 refused to fetch a local test page by default (a deliberate loopback block); with --allow-private-ips it fetched it in 698.3 ms median. MCP Sequential Thinking server answered over stdio in 271 ms cold (v1.0.0, Node 20) and 159 ms (v2.0.0, Node 24); median call 0.8 to 0.4 ms. MCP Filesystem server answered over stdio in 279 ms cold (v1.0.0, Node 20) and 187 ms (v2.0.1, Node 24); median call 0.8 to 1.0 ms. MCP Everything (reference) server answered over stdio in 356 ms cold (v1.0.0, Node 20) and 252 ms (v2.0.1, Node 24); median call 0.5 to 0.4 ms. The MCP Git server 2026.10.19 started in 512 ms, listed 12 tools and answered git_status in 3.2 ms median. The MCP Time server 2026.10.19 started in 483 ms and answered get_current_time in 0.8 ms median.
- Not tested: Remote (HTTP) transport, authentication and concurrent clients: only local stdio with one client was measured; Large payloads: calls used tiny inputs, so these latencies are protocol and process overhead, not throughput; Anything about answer quality: no model was involved in the latency numbers; These are single smoke-test runs; it is not a benchmark. The machine was shared with other jobs.

Sources:
- MCP reference servers repository (GitHub): https://github.com/modelcontextprotocol/servers
Cite as: Agentic Weekly, "MCP reference servers v2.1.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/mcp-servers

## Microsoft Copilot (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/microsoft-copilot (markdown: https://theagenticweekly.com/tools/microsoft-copilot.md)
Vendor: Microsoft

Microsoft's Copilot page describes AI chat for everyday tasks and lists tiers at $9.99 a month, $23.50 per user a month paid yearly, and $30 per user a month paid yearly, plus enterprise bundles with Copilot Studio (tier names not captured).

Verdict: Microsoft's AI assistant for chat, creating and getting work done, with consumer and enterprise tiers.
Good for: Microsoft 365 users
Skip if: You avoid Microsoft accounts

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Microsoft Copilot
1. Get access: Open copilot.microsoft.com with a Microsoft account.
2. Check the limits and costs first: Free to use (per site) Paid: $9.99 a month; $23.50 per user per month paid yearly; $30 per user per month paid yearly (tier names not captured) Read from microsoft.com Copilot plans page on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Tier names

Sources:
- Microsoft Copilot: https://copilot.microsoft.com/
Cite as: Agentic Weekly, "Microsoft Copilot: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/microsoft-copilot

## Microsoft Copilot Studio (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/copilot-studio (markdown: https://theagenticweekly.com/tools/copilot-studio.md)
Vendor: Microsoft

The Copilot Studio page describes a platform for creating and managing custom agents, workflows and apps with no code. Pricing was not captured.

Verdict: Microsoft's platform for creating and managing custom agents, workflows and apps with no code.
Good for: Microsoft 365 organizations building internal agents
Skip if: You want open source

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Microsoft Copilot Studio
1. Get access: Open Copilot Studio with a Microsoft work account.
2. Check the limits and costs first: Not captured. Pricing not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Pricing

Sources:
- Copilot Studio: https://www.microsoft.com/en-us/microsoft-copilot/microsoft-copilot-studio
Cite as: Agentic Weekly, "Microsoft Copilot Studio: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/copilot-studio

## Mistral Le Chat (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/le-chat (markdown: https://theagenticweekly.com/tools/le-chat.md)
Vendor: Mistral AI

Mistral's pricing page lists Free, Pro ($14.99 a month; verified students $5.99), Team ($24.99 per user a month) and Enterprise with custom SLAs; Pro lists up to 6x and Team up to 40x the free usage, and pay-as-you-go credits can extend usage.

Verdict: Mistral's AI assistant: Free, Pro at $14.99 a month, Team at $24.99 per user.
Good for: People who want a European assistant with a cheap Pro tier
Skip if: You need the largest ecosystem of plugins

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Mistral Le Chat
1. Get access: Open chat.mistral.ai and sign up for the Free plan.
2. Check the limits and costs first: Free plan with limited access. Paid: Pro $14.99/month (students $5.99); Team $24.99 per user per month; Enterprise custom; PAYG credits optional. Read from mistral.ai pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Model quality (not tested)

Sources:
- Mistral pricing: https://mistral.ai/pricing
Cite as: Agentic Weekly, "Mistral Le Chat: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/le-chat

## Murf AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/murf-ai (markdown: https://theagenticweekly.com/tools/murf-ai.md)
Vendor: Murf AI

Murf's pricing page lists a free trial with $10 of API credit, pay-as-you-go credits that never expire, and Enterprise. API rates shown: its Falcon text-to-speech at $0.01 per 1,000 characters (time to first audio under 100 ms), studio-quality Gen2 at $0.03 per 1,000 characters, speech-to-text (beta) at $0.24, voice changer at $0.10 per minute and translation at $0.02 per 1,000 characters.

Verdict: Text-to-speech platform with a $10 free API trial, pay-as-you-go credits and per-character API pricing.
Good for: Developers who want low per-character TTS pricing; Teams building voice agents that need fast first audio
Skip if: You need a published monthly studio subscription price (not shown on the page we read); You need on-device open-source TTS

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Murf AI
1. Get access: Open murf.ai/pricing and start with the $10 free API trial.
2. Check the limits and costs first: $10 of free API credit. Paid: Pay as you go (credits never expire); Falcon TTS $0.01 per 1,000 characters; Gen2 TTS $0.03 per 1,000 characters; voice changer $0.10 per minute; Enterprise custom. Read from murf.ai/pricing on 2026-10-25. The page focuses on API pricing; Studio subscription prices were not shown.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Voice quality compared with other providers (not tested)
- Studio subscription pricing

Sources:
- Murf pricing: https://murf.ai/pricing
Cite as: Agentic Weekly, "Murf AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/murf-ai

## Notion AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/notion-ai (markdown: https://theagenticweekly.com/tools/notion-ai.md)
Vendor: Notion

Notion's site describes AI agents that answer questions, prioritize tasks and write reports. Its pricing page lists Free ($0), Plus ($10 per member per month), Business ($20) and Enterprise (custom, zero data retention with LLM providers), and notes Notion AI requires Notion credits.

Verdict: Notion's built-in AI and agents; Free $0, Plus $10 and Business $20 per member per month, with AI usage drawing on credits.
Good for: Teams that already live in Notion
Skip if: You want a standalone assistant

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Notion AI
1. Get access: Open a Notion workspace and use the AI entry point.
2. Check the limits and costs first: Free: $0 per member per month. Paid: Plus $10, Business $20 per member per month (per pricing page); Enterprise custom; AI requires credits. Read from notion.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Credit amounts per plan

Sources:
- Notion AI: https://www.notion.com/product/ai
- Notion pricing: https://www.notion.com/pricing
Cite as: Agentic Weekly, "Notion AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/notion-ai

## Ollama v0.40.11 (2026-10-25, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/ollama (markdown: https://theagenticweekly.com/tools/ollama.md)
Vendor: Ollama

Ollama's site describes running and building with open models locally or in the cloud on macOS, Windows and Linux, and says prompts are never stored or trained on. The GitHub repository shows the MIT license and release v0.40.11 on Oct 25, 2026.

Verdict: The common way to run open models locally on macOS, Windows and Linux, with a cloud option and an MIT-licensed open-source core.
Good for: Developers who want local models behind a simple API; People who want to try open models privately
Skip if: You need frontier proprietary models only; You have no GPU and want fast large-model output

What's new:
- v0.40.11 (2026-10-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Ollama
1. Get access: Download Ollama from ollama.com and run a model from the command line.
2. Check the limits and costs first: Free: $0, pay-as-you-go usage credits for cloud models. Paid: Paid plan $20/month ($16.67/month billed annually) with $60 of monthly usage credits; $100/month with $300; $500/month with $1,000 shared; Enterprise custom. Read from ollama.com/pricing on 2026-10-25. Local use of open models is free; cloud models bill per million tokens.

Not verified:
- Speed of a given model on your hardware

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Ollama 0.40.11 served three Qwen2.5 sizes over its OpenAI-compatible API on 8 CPU cores: all three answered simple prompts; the 0.5B model failed JSON extraction while 1.5B and 3B passed, and tool calls worked 5 of 5 on all three.
- Not tested: Larger models, GPUs, concurrency, context lengths beyond a few hundred tokens and quality of long answers; A "write" quality judgement: we only timed it; Three runs per prompt is a smoke test, not a benchmark.

Sources:
- Ollama: https://ollama.com/
- Ollama pricing: https://ollama.com/pricing
- Ollama repository (GitHub): https://github.com/ollama/ollama
Cite as: Agentic Weekly, "Ollama v0.40.11: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/ollama

## Open WebUI v0.12.4 (2026-10-25, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/open-webui (markdown: https://theagenticweekly.com/tools/open-webui.md)
Vendor: Open WebUI

Open WebUI's site pitches a self-hosted AI platform you run on your own machine and connect to any model. The repository's license file is a custom 'Open WebUI License' (BSD-style terms with additional conditions), not a standard OSI license; v0.12.4 was tagged Oct 25, 2026.

Verdict: Self-hosted AI platform: a web interface that connects to Ollama and OpenAI-compatible models, under its own Open WebUI License.
Good for: Teams that want a self-hosted chat UI; Ollama users who want a browser interface
Skip if: You need a standard OSI license for your own distribution

What's new:
- v0.12.4 (2026-10-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Open WebUI
1. Get access: Follow the install guide on openwebui.com (Docker is the common route) and connect a model backend.
2. Check the limits and costs first: Free to self-host. Paid: Enterprise options not read. No pricing page read.

Not verified:
- Enterprise pricing

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Open WebUI 0.12.4 installed with uv in 11.6 s (6,977 MB in the virtualenv), was healthy about 11 to 12 s after start, and answered a chat through its API in 0.68 s using a local Ollama model.
- Not tested: The browser UI, RAG over documents, web search, multi-user roles, and Docker deployment; One request; a smoke test, not a load test.

Sources:
- Open WebUI: https://openwebui.com/
- Open WebUI license: https://github.com/open-webui/open-webui/blob/main/LICENSE
- Open WebUI repository (GitHub): https://github.com/open-webui/open-webui
Cite as: Agentic Weekly, "Open WebUI v0.12.4: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/open-webui

## OpenRouter (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/openrouter (markdown: https://theagenticweekly.com/tools/openrouter.md)
Vendor: OpenRouter

OpenRouter's site describes a unified interface for all major models with fallback to other providers when one goes down. Its pricing page lists Free (25+ free models, 4 free providers), pay-as-you-go use and Enterprise.

Verdict: Unified API to many models with fallbacks; a Free tier with 25+ free models and 4 free providers.
Good for: Developers who want to switch models behind one API
Skip if: You want a local runtime

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with OpenRouter
1. Get access: Create an API key at openrouter.ai.
2. Check the limits and costs first: Free: 25+ free models, 4 free providers. Paid: Pay as you go by model; Enterprise custom. Read from openrouter.ai/pricing on 2026-10-25; per-model prices vary.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Fee structure beyond the page

Sources:
- OpenRouter: https://openrouter.ai/
- OpenRouter pricing: https://openrouter.ai/pricing
Cite as: Agentic Weekly, "OpenRouter: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/openrouter

## OpusClip (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/opus-clip (markdown: https://theagenticweekly.com/tools/opus-clip.md)
Vendor: OpusClip

OpusClip's site says one long video becomes ten clips and can publish to all social platforms; pricing lists Free forever ($0), a $15 monthly-billed plan, and Enterprise.

Verdict: Turns long videos into shorts and posts to social platforms; Free forever plan.
Good for: Creators repurposing long videos
Skip if: You want to edit from scratch

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with OpusClip
1. Get access: Sign up free and upload a long video.
2. Check the limits and costs first: Free forever: $0. Paid: $15 billed monthly plan; Enterprise custom. Read from opus.pro pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Credits per plan

Sources:
- OpusClip: https://www.opus.pro/
Cite as: Agentic Weekly, "OpusClip: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/opus-clip

## Otter.ai (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/otter (markdown: https://theagenticweekly.com/tools/otter.md)
Vendor: Otter.ai

Otter's site calls it a meeting agent with real-time transcription, live chat, summaries and action items. The pricing page lists Free, Pro, Business and Enterprise plans with minutes and import limits; the exact price table varies by region and billing so we do not quote it.

Verdict: AI meeting agent with real-time transcription, summaries and action items; free plan, paid per-user plans.
Good for: People who want live transcription
Skip if: You need predictable pricing without regional offers

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Otter.ai
1. Get access: Sign up free and connect your calendar.
2. Check the limits and costs first: Free plan. Paid: Paid per-user plans; prices vary by billing period and region. Read from otter.ai/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Exact prices

Sources:
- Otter: https://otter.ai/
- Otter pricing: https://otter.ai/pricing
Cite as: Agentic Weekly, "Otter.ai: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/otter

## Perplexity (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/perplexity (markdown: https://theagenticweekly.com/tools/perplexity.md)
Vendor: Perplexity AI

Perplexity's pricing hub lists Free ($0, citations on every answer, limited weekly usage), Pro at $17 a month billed annually ($20 monthly) and Max at $167 a month billed annually ($200 monthly per its help center). Pro adds Perplexity Computer access and premium sources; Max adds higher limits on Deep Research and more video creation.

Verdict: AI search with citations in every answer; Pro is $20 a month ($17 billed annually) and Max is $200 ($167 billed annually).
Good for: People who want cited web answers; Researchers using deep research
Skip if: You want an offline or local tool

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Perplexity
1. Get access: Open perplexity.ai and try the Free plan.
2. Check the limits and costs first: Free: $0, limited weekly usage with citations. Paid: Pro $20/month or $17/month billed annually; Max $200/month or $167/month billed annually; Enterprise Pro and Max per seat. Read from perplexity.ai/hub/pricing and its help center, 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Answer quality (not tested)

Sources:
- Perplexity pricing: https://www.perplexity.ai/hub/pricing
- Perplexity Max help center: https://www.perplexity.ai/help-center/en/articles/11680686-perplexity-max
Cite as: Agentic Weekly, "Perplexity: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/perplexity

## Pika (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/pika (markdown: https://theagenticweekly.com/tools/pika.md)
Vendor: Pika

Pika's pricing lists Free, then $10, $35 and $95 a month tiers (annual billing is 20 percent less: $8, $28, $76), a higher range from $95 to $880, and Enterprise; model access and credits vary by plan.

Verdict: AI creative platform with a free plan and monthly tiers at $10, $35 and $95.
Good for: Creators wanting several video models in one place
Skip if: You need open-weight local generation

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Pika
1. Get access: Sign up free and generate a first clip.
2. Check the limits and costs first: Free plan with limited credits. Paid: $10, $35 and $95 per month (annual about 20 percent less: $8, $28, $76); higher tiers up to $880; Enterprise custom. Read from pika.art pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Credit cost per clip

Sources:
- Pika: https://pika.art/
Cite as: Agentic Weekly, "Pika: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/pika

## Pipedream (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/pipedream (markdown: https://theagenticweekly.com/tools/pipedream.md)
Vendor: Pipedream

Pipedream's pricing page says it is free to build and you pay when integrations go to production. It lists Free ($0), a paid tier at $99 per month billed annually or $150 per month billed monthly with 10,000 usage credits and 100 users included, extra users at $2 and extra credits at $0.012 each.

Verdict: Integration layer for AI agents with managed auth and 10,000+ tools across 3,000+ APIs; free to build, pay in production.
Good for: Developers wiring agents to many APIs with managed auth
Skip if: Non-technical teams wanting only drag and drop

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Pipedream
1. Get access: Sign up free and build a workflow or add connectors via the docs.
2. Check the limits and costs first: Free: full access in development, no card. Paid: $99 per month billed annually or $150 billed monthly; 100 users and 10,000 credits included; $2 per extra user; $0.012 per extra credit. Read from pipedream.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan name for the $99 tier

Sources:
- Pipedream: https://pipedream.com/
- Pipedream pricing: https://pipedream.com/pricing
Cite as: Agentic Weekly, "Pipedream: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/pipedream

## Power Automate (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/power-automate (markdown: https://theagenticweekly.com/tools/power-automate.md)
Vendor: Microsoft

Power Automate's pricing page lists a 30-day free trial, $15 per user per month paid yearly, $150 per bot per month, $215 per bot per month and $5,000 per tenant per month (plan labels not captured), and mentions pay-as-you-go Copilot Credit coverage.

Verdict: Microsoft's automation platform with a 30-day free trial and per-user and per-bot plans.
Good for: Microsoft 365 shops needing RPA and cloud flows
Skip if: Small teams wanting a free plan

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Power Automate
1. Get access: Start the 30-day trial.
2. Check the limits and costs first: 30-day free trial. Paid: $15 per user per month paid yearly; $150 and $215 per bot per month; $5,000 per tenant per month. Read from the Microsoft Power Automate pricing page on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan names

Sources:
- Power Automate: https://www.microsoft.com/en-us/power-platform/products/power-automate
Cite as: Agentic Weekly, "Power Automate: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/power-automate

## Pydantic AI (2026-10-25, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/pydantic-ai (markdown: https://theagenticweekly.com/tools/pydantic-ai.md)
Vendor: Pydantic

Pydantic AI's docs describe typed control flow for agents across models, with agent capabilities and graph control flow as separate open-source packages. The repository is MIT-licensed.

Verdict: Python agent framework built on Pydantic types, covering agents, realtime voice, image generation and embeddings.
Good for: Python teams that value type safety and validation
Skip if: You need a TypeScript framework

What's new:
- Latest release we saw (2026-10-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Pydantic AI
1. Get access: pip install pydantic-ai and run the first agent from the docs.
2. Check the limits and costs first: Free to self-host. Open-source library; costs come from the model APIs you call.

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Pydantic AI 2.61.0 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 3.7 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- Pydantic AI docs: https://ai.pydantic.dev/
- Pydantic AI repository (GitHub): https://github.com/pydantic/pydantic-ai
Cite as: Agentic Weekly, "Pydantic AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/pydantic-ai

## Qodo (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/qodo (markdown: https://theagenticweekly.com/tools/qodo.md)
Vendor: Qodo

Qodo describes itself as an AI code review tool giving teams deep codebase context, with 15+ agentic workflows for IDEs, PRs and security. Its pricing page lists a Pro Team plan starting at $30, a 14-day free trial with no credit card, and an Enterprise plan for 30+ users.

Verdict: AI code review platform with 15+ agentic workflows for IDEs, pull requests and security, priced by team credits.
Good for: Teams that want review automation on every pull request
Skip if: You want a general-purpose coding agent; You want a pure open-source tool

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Qodo
1. Get access: Start the 14-day trial at qodo.ai and connect a repository.
2. Check the limits and costs first: Free 14-day trial, no credit card; page also says 'Free to Start'. Paid: Pro Team from $30 (credits pooled across the team at $0.012 per credit; packs of 2,500, 5,000 and 20,000 credits listed); Enterprise by quote. Read from qodo.ai/pricing on 2026-10-25; the page text did not make clear whether $30 is per user per month, so we do not state a unit.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- The unit of the $30 Pro Team price
- Review quality (not tested)

Sources:
- Qodo: https://www.qodo.ai/
- Qodo pricing: https://www.qodo.ai/pricing/
Cite as: Agentic Weekly, "Qodo: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/qodo

## Read AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/read-ai (markdown: https://theagenticweekly.com/tools/read-ai.md)
Vendor: Read AI

Read AI's site lists Free ($0, 5 meeting transcripts a month, unlimited enterprise search) and paid tiers priced at $15, $22.50 and $29.75 per user per month billed annually ($19.75, $29.75 and $39.75 monthly), plus custom options.

Verdict: AI notetaker that works across Teams, Meet and Zoom; Free plan covers 5 meetings per month.
Good for: Teams that want search plus summaries
Skip if: You do not want a bot

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Read AI
1. Get access: Sign up free and link your calendar.
2. Check the limits and costs first: Free: 5 meeting transcripts per month. Paid: $15, $22.50 and $29.75 per user per month billed annually ($19.75, $29.75, $39.75 monthly) Read from read.ai/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)

Sources:
- Read AI: https://www.read.ai/
- Read AI pricing: https://www.read.ai/pricing
Cite as: Agentic Weekly, "Read AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/read-ai

## Recraft (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/recraft (markdown: https://theagenticweekly.com/tools/recraft.md)
Vendor: Recraft

Recraft's site lists photorealism, vector generation, custom styles and mockups, and highlights Recraft V4.1 Flash. Its pricing FAQ says credit costs vary by action and model and distinguishes subscription from top-up credits.

Verdict: Text-to-image model and design platform for photorealism and vector output; V4.1 Flash is listed as the latest model.
Good for: Designers needing vector output
Skip if: You want a free local model

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Recraft
1. Get access: Sign up free.
2. Check the limits and costs first: Free plan exists (FAQ asks who owns Free-plan images) Prices not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Prices
- Free-plan ownership terms

Sources:
- Recraft: https://www.recraft.ai/
Cite as: Agentic Weekly, "Recraft: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/recraft

## Relevance AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/relevance-ai (markdown: https://theagenticweekly.com/tools/relevance-ai.md)
Vendor: Relevance AI

Relevance AI's site describes specialist agents and a product called Invent, plus an embedded deployment team. The pricing page mentions vendor credits and enterprise triggers; prices were not captured.

Verdict: Platform for specialist AI agents for sales, customer success, marketing and HR, with an embedded deployment team.
Good for: Go-to-market teams wanting managed agent rollout
Skip if: Hobbyists

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Relevance AI
1. Get access: Sign up via relevanceai.com.
2. Check the limits and costs first: Not captured. Pricing not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Pricing

Sources:
- Relevance AI: https://relevanceai.com/
Cite as: Agentic Weekly, "Relevance AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/relevance-ai

## Replicate (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/replicate (markdown: https://theagenticweekly.com/tools/replicate.md)
Vendor: Replicate

Replicate's site says you can run and fine-tune models and deploy custom models with one line of code. Prices are per use by model, such as $0.04 per image on listed rows, and Enterprise offers volume discounts.

Verdict: Cloud API for running and fine-tuning open models, with per-use pricing.
Good for: Developers adding image or model APIs quickly
Skip if: You need a flat monthly price

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Replicate
1. Get access: Create an account and run a model from the web or API.
2. Check the limits and costs first: Try for free (per page) Paid: Per-model usage pricing; enterprise volume discounts. Read from replicate.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Cold-start behavior

Sources:
- Replicate: https://replicate.com/
- Replicate pricing: https://replicate.com/pricing
Cite as: Agentic Weekly, "Replicate: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/replicate

## Resemble AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/resemble-ai (markdown: https://theagenticweekly.com/tools/resemble-ai.md)
Vendor: Resemble AI

Resemble's pricing lists Free ($0 with pay-as-you-go credits), $350 a month ($280 billed annually), $1,000 a month ($800 billed annually), and Enterprise; usage rates are listed per second (for example $0.035 per second or per image on pay-as-you-go rows).

Verdict: Resemble AI now leads with real-time deepfake detection and watermarking for audio, video and images; plans from free to $1,000 a month.
Good for: Enterprises verifying media authenticity
Skip if: You want a consumer text-to-speech app

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Resemble AI
1. Get access: Start free from resemble.ai.
2. Check the limits and costs first: Free: $0, buy credits as you go. Paid: $350/month ($280 annual); $1,000/month ($800 annual); Enterprise custom. Read from resemble.ai/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Voice-generation pricing

Sources:
- Resemble AI: https://www.resemble.ai/
- Resemble pricing: https://www.resemble.ai/pricing
Cite as: Agentic Weekly, "Resemble AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/resemble-ai

## Retell AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/retell (markdown: https://theagenticweekly.com/tools/retell.md)
Vendor: Retell AI

Retell's pricing lists $0.07 to $0.31 per minute depending on configuration, 20 free concurrent calls and an Enterprise plan; its site lists HIPAA and SOC 2.

Verdict: Voice AI agents for phone calls with pay-per-minute pricing and 20 free concurrent calls.
Good for: Teams automating inbound and outbound phone calls
Skip if: You need text chat only

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Retell AI
1. Get access: Sign up free and build a phone agent in the dashboard.
2. Check the limits and costs first: 20 free concurrent calls; sign up free. Paid: $0.07 to $0.31 per minute depending on models and voice; Enterprise custom. Read from retellai.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Workflow accuracy claim (vendor)

Sources:
- Retell AI: https://www.retellai.com/
- Retell pricing: https://www.retellai.com/pricing
Cite as: Agentic Weekly, "Retell AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/retell

## Speechify (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/speechify (markdown: https://theagenticweekly.com/tools/speechify.md)
Vendor: Speechify

Speechify's pricing page lists Free and a premium plan at $29 a month billed monthly; its site says it has 60M+ users and supports voice typing and a Q&A assistant.

Verdict: Text-to-speech app that reads PDFs, books and web pages aloud, with a Free plan and a $29 monthly premium price.
Good for: People who want documents read aloud
Skip if: Developers needing an API only

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Speechify
1. Get access: Sign up free and import a document.
2. Check the limits and costs first: Free plan. Paid: Premium $29 per month billed monthly. Read from speechify.com/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)

Sources:
- Speechify: https://speechify.com/
- Speechify pricing: https://speechify.com/pricing
Cite as: Agentic Weekly, "Speechify: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/speechify

## Synthesia (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/synthesia (markdown: https://theagenticweekly.com/tools/synthesia.md)
Vendor: Synthesia

Synthesia's site cites avatars and voiceovers in 140+ languages and live Sessions for roleplay. Its pricing lists a free allowance of 500 credits a month, a paid tier of 6,000 credits a month and Enterprise (custom); dollar prices were not captured.

Verdict: Business video platform with AI avatars in 140+ languages and a free credit allowance.
Good for: Corporate training and communications teams
Skip if: You need cinematic generative video

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Synthesia
1. Get access: Sign up and use the free credits.
2. Check the limits and costs first: 500 free credits per month. Paid: A paid tier listing 6,000 credits per month; Enterprise custom; dollar prices not captured. Read from synthesia.io/pricing on 2026-10-25.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Dollar prices

Sources:
- Synthesia: https://www.synthesia.io/
- Synthesia pricing: https://www.synthesia.io/pricing
Cite as: Agentic Weekly, "Synthesia: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/synthesia

## Together AI (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/together-ai (markdown: https://theagenticweekly.com/tools/together-ai.md)
Vendor: Together AI

Together AI's site describes a full-stack AI platform for inference, fine-tuning and GPU clusters. Its pricing page is a long per-model, per-million-token table; we do not quote individual model prices.

Verdict: AI cloud for open models: serverless inference, fine-tuning and GPU clusters, priced per token.
Good for: Teams serving open models at scale
Skip if: You want a desktop app

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Together AI
1. Get access: Sign up for an API key at together.ai.
2. Check the limits and costs first: Not captured. Paid: Per-model per-token prices; fine-tuning and GPU clusters priced separately. Read from together.ai/pricing on 2026-10-25; model prices change often.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Free credit

Sources:
- Together AI: https://www.together.ai/
- Together AI pricing: https://www.together.ai/pricing
Cite as: Agentic Weekly, "Together AI: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/together-ai

## Vapi (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/vapi (markdown: https://theagenticweekly.com/tools/vapi.md)
Vendor: Vapi

Vapi's pricing page offers $5 in free credits, pay-per-minute use (pre-paid with $0.05 a minute Vapi hosting), and lists plan amounts of $29 and $999 a month plus enterprise.

Verdict: Developer platform for voice agents; $5 free credits, then pay for the minutes you run.
Good for: Developers who want to assemble their own voice stack
Skip if: You want a flat per-seat price

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Vapi
1. Get access: Sign up for $5 in free credits and create an assistant.
2. Check the limits and costs first: $5 free credits. Paid: Pay per minute (Vapi hosting $0.05/min pre-paid, plus provider costs); plan amounts of $29 and $999 per month listed; Enterprise custom. Read from vapi.ai/pricing on 2026-10-25. Provider costs (speech, model, telephony) add to the per-minute rate.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- What the $29 and $999 amounts buy

Sources:
- Vapi: https://vapi.ai/
- Vapi pricing: https://vapi.ai/pricing
Cite as: Agentic Weekly, "Vapi: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/vapi

## Voiceflow (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/voiceflow (markdown: https://theagenticweekly.com/tools/voiceflow.md)
Vendor: Voiceflow

Voiceflow's site describes an AI agent platform for building, launching and scaling chat and voice agents for customer support and CX, with a free trial. Prices were not captured.

Verdict: Platform for chat and voice AI agents for customer experience teams.
Good for: CX teams building chat and voice agents
Skip if: Hobby projects

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Voiceflow
1. Get access: Start the free trial.
2. Check the limits and costs first: Free trial, no card. Pricing not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Pricing

Sources:
- Voiceflow: https://www.voiceflow.com/
Cite as: Agentic Weekly, "Voiceflow: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/voiceflow

## Warp (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/warp (markdown: https://theagenticweekly.com/tools/warp.md)
Vendor: Warp

Warp's site describes factories-as-code: versioned definitions that connect repositories, credentials, work sources and cloud access for agents, with any model or harness. A Warp Agent CLI and a free plan are listed on the pricing page.

Verdict: Warp's terminal and agent platform; it now pitches 'open infrastructure for cloud software factories' defined as code, with any model.
Good for: Developers who want an agent-aware terminal; Teams that want to define agent workflows as code
Skip if: You want an open-source terminal; You only want a classic terminal with no agents

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Warp
1. Get access: Download Warp from warp.dev and sign in; the Free plan needs no payment.
2. Check the limits and costs first: Free: $0 a month. Paid: Build $20/month ($18 billed annually; $20 of agent usage included at API rates); Max $200/month; Business $50 per user per month; Enterprise custom. Read from warp.dev/pricing on 2026-10-25. Usage included in paid plans is priced at API rates.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Agent quality (not tested)
- Supported OS list

Sources:
- Warp: https://www.warp.dev/
- Warp pricing: https://www.warp.dev/pricing
Cite as: Agentic Weekly, "Warp: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/warp

## Writesonic (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/writesonic (markdown: https://theagenticweekly.com/tools/writesonic.md)
Vendor: Writesonic

Writesonic describes itself as an AI search growth engine covering GEO, AEO and AI visibility, tracking 8 AI platforms. Its pricing lists plans with 15, 25 and 50 AI articles per month and Enterprise; prices were not captured.

Verdict: Platform for tracking and improving visibility in AI search with article generation.
Good for: Marketers tracking AI search visibility
Skip if: You only need a basic writer

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Writesonic
1. Get access: Sign up from the site.
2. Check the limits and costs first: Not captured. Paid: Plans listed by article counts: 15, 25 and 50 per month; Enterprise custom. Prices were not captured.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Plan prices

Sources:
- Writesonic: https://writesonic.com/
Cite as: Agentic Weekly, "Writesonic: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/writesonic

## You.com (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/you-com (markdown: https://theagenticweekly.com/tools/you-com.md)
Vendor: You.com

You.com's homepage and pricing page describe search and inference for AI. The pricing page lists a Web Search API with a free tier (no API key, account or credit card for the MCP config) and a $5 per 1,000 calls standard tier, a Contents API at $1 per 1,000 pages, an Answer API at $5 per 1,000 calls and Research API tiers from $12 to $1,200 per 1,000 calls.

Verdict: You.com now presents itself as search and inference infrastructure for AI: web search, answer and research APIs plus an inference router.
Good for: Developers adding web search or research to an agent
Skip if: You want a consumer chat assistant (the site now leads with APIs)

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with You.com
1. Get access: Get an API key from you.com or use the MCP config docs for the free tier.
2. Check the limits and costs first: Web Search API: free tier; no key, account or card needed for the MCP setup. Paid: Web Search $5 per 1k calls; Contents $1 per 1k pages; Answer $5 per 1k calls; Research Lite $12 to Frontier $1,200 per 1k calls. Read from you.com/pricing on 2026-10-25. Enterprise is custom.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Answer quality (not tested)

Sources:
- You.com: https://you.com/
- You.com pricing: https://you.com/pricing
Cite as: Agentic Weekly, "You.com: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/you-com

## Zapier (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/zapier (markdown: https://theagenticweekly.com/tools/zapier.md)
Vendor: Zapier

Zapier's site says it connects 9,000+ apps and is trusted by over 3 million businesses. Its pricing page lists a Free plan with 100 tasks per month, Professional from $19.99 a month billed annually (750 tasks), Team from $69 a month billed annually (2,000 tasks) and custom Enterprise.

Verdict: No-code automation with 9,000+ apps and AI agents; free plan of 100 tasks a month.
Good for: Teams that need the widest app coverage
Skip if: Cost-sensitive high-volume automation

What's new:
- Catalog entry: facts read from the vendor's pages on 2026-10-25. We do not track release notes for this entry.

Tutorial: Get started with Zapier
1. Get access: Create a free account and build a first Zap from a template.
2. Check the limits and costs first: Free: 100 tasks per month, includes Zaps, Tables and Forms. Paid: Professional from $19.99/month billed annually ($29.99 billed monthly) at 750 tasks; Team from $69/month billed annually ($103.50 monthly) at 2,000 tasks; Enterprise custom; task tiers rise to 2M a month; annual billing is 33% off the monthly rate; 14-day Professional trial. Read from zapier.com/pricing on 2026-10-25. Prices are per month for the smallest task tier of each plan; larger tiers cost more.

Not verified:
- Release date (none published on the pages we read; the date shown is the day we first covered this entry)
- Prices above 2,000 tasks a month

Sources:
- Zapier: https://zapier.com/
- Zapier pricing: https://zapier.com/pricing
Cite as: Agentic Weekly, "Zapier: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/zapier

## Zed (2026-10-25, Reported (not run by us))
URL: https://theagenticweekly.com/tools/zed (markdown: https://theagenticweekly.com/tools/zed.md)
Vendor: Zed Industries

Zed's homepage calls it a high-performance multiplayer code editor from the creators of Atom and Tree-sitter, 'open source, AI optional'. The repository shows the editor under GPL-3.0 with other components under other licenses. Its pricing page lists a free Personal plan and Pro at $10 a month.

Verdict: High-performance, multiplayer code editor from the creators of Atom and Tree-sitter, open source with AI optional.
Good for: Developers who want a fast editor and optional AI; Teams that want org-wide AI controls (Business)
Skip if: You want AI built into every action by default; You need SSO, SAML or SCIM: the page says these are planned but not available

What's new:
- Latest release we saw (2026-10-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Zed
1. Get access: Download Zed from zed.dev/download and open a project; AI features can be switched on with your own keys or a Pro plan.
2. Check the limits and costs first: Personal: $0 forever, bring your own keys or no AI. Paid: Pro $10/month with $5 of tokens included then usage-based; Business $30 per seat per month. Read from zed.dev/pricing on 2026-10-25. Pro has a 14-day trial with $5 of usage.

Not verified:
- Windows and Linux support status (not read)
- Agent quality (not tested)

Sources:
- Zed: https://zed.dev/
- Zed pricing: https://zed.dev/pricing
- Zed repository: https://github.com/zed-industries/zed
Cite as: Agentic Weekly, "Zed: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/zed

## LangGraph 1.2.17 (2026-10-24, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/langgraph (markdown: https://theagenticweekly.com/tools/langgraph.md)
Vendor: LangChain

LangGraph 1.2.17 is a small fix release: it applies an explicit recursion limit equal to the default and keeps the caller's namespace in RemoteGraph streaming, plus dependency bumps.

Verdict: A dependable foundation for stateful agents; this release is a routine fix, so only update if the RemoteGraph or recursion fixes affect you.
Good for: Developers who want explicit control over agent state and flow; Teams running long-lived, resumable agents; People already in the LangChain ecosystem
Skip if: You want the shortest path to a simple agent; You are looking for new features in this release

What's new:
- 1.2.17 (Oct 24): Apply an explicit recursion_limit equal to the default.
- 1.2.17: Keep the caller's namespace in RemoteGraph.stream(subgraphs=True).
- 1.2.17: Remove precautionary third-party upper bounds on dependencies and bump certifi.

Tutorial: Install LangGraph
1. Install the package: The README gives one command.
2. Read the quickstart: The README links to the official docs and a quickstart that builds a first graph. Start with a two-node graph before adding tools.
3. Pin your version: Releases ship often. Pin the version in requirements.txt and read the release notes before upgrading.

Not verified:
- Hosted LangGraph platform pricing (not covered)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Installed langgraph 1.2.17 and ran a two-node StateGraph (count words, then label); the input 'agentic weekly tests every tool' returned n=5 and the labelled text. Extended: LangGraph 1.2.17 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 2.5 s.
- Not tested: RemoteGraph, checkpoints, the recursion-limit fixes, any LLM-driven agent beyond the one-tool task, the hosted LangGraph platform; Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- LangGraph 1.2.17 release: https://github.com/langchain-ai/langgraph/releases/tag/1.2.17
- README: https://github.com/langchain-ai/langgraph
Cite as: Agentic Weekly, "LangGraph 1.2.17: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/langgraph

## Vercel AI SDK (2026-10-24, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/vercel-ai-sdk (markdown: https://theagenticweekly.com/tools/vercel-ai-sdk.md)
Vendor: Vercel

The AI SDK site describes a unified TypeScript SDK with streaming, fallbacks and multi-model support. The repository's LICENSE is Apache 2.0 and releases publish multiple times a week (for example @ai-sdk/zai 3.0.40 on Oct 24, 2026).

Verdict: Vercel's TypeScript SDK that gives one API across many model providers, with streaming and agent helpers.
Good for: TypeScript and Next.js teams building chat or agent UIs
Skip if: You work in Python

What's new:
- Latest release we saw (2026-10-24): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Vercel AI SDK
1. Get access: npm install ai and follow the quickstart for your framework.
2. Check the limits and costs first: Free to self-host. Open-source library; costs come from the model APIs you call.

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Vercel AI SDK passed our one-tool task 3 of 3 on Node 20 (ai 6.0.315) and 3 of 3 on Node 24 (ai 7.0.153); the install is 26 to 27 MB.
- Not tested: Any vendor-hosted model: everything ran against one local 3B model; Multi-agent flows, memory, streaming and deployment; Anything beyond one tool and one question; five trials is a smoke test, not a benchmark.

Sources:
- AI SDK: https://ai-sdk.dev/
- AI SDK repository (GitHub): https://github.com/vercel/ai
Cite as: Agentic Weekly, "Vercel AI SDK: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/vercel-ai-sdk

## Activepieces 0.92.3-hotfix.1 (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/activepieces (markdown: https://theagenticweekly.com/tools/activepieces.md)
Vendor: Activepieces

Activepieces' site lists Free ($0, 1,000 credits a month), $20 flat per month, $200 flat per month (10x the free allowance) and Enterprise from $36k a year, with extra credits at $0.007 each. Its repository does not resolve to a single standard license on GitHub, so read the LICENSE before commercial use.

Verdict: Open-source workflow automation with chat, agents, tables and 770+ apps; $20 or $200 flat monthly plans.
Good for: Teams that want self-hostable automation; Zapier users sensitive to per-task costs
Skip if: You need the largest app library (Zapier lists 9,000+)

What's new:
- 0.92.3-hotfix.1 (2026-10-23): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Activepieces
1. Get access: Sign up for the free plan or self-host from the repository.
2. Check the limits and costs first: Free: $0, 1,000 credits per month. Paid: $20 flat per month; $200 flat per month; Enterprise from $36k per year; credits $0.007 each. Read from activepieces.com/pricing on 2026-10-25.

Not verified:
- Self-hosted feature limits

Sources:
- Activepieces: https://www.activepieces.com/
- Activepieces pricing: https://www.activepieces.com/pricing
- Activepieces repository (GitHub): https://github.com/activepieces/activepieces
Cite as: Agentic Weekly, "Activepieces 0.92.3-hotfix.1: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/activepieces

## Agno v3.1.10 (2026-10-23, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/agno (markdown: https://theagenticweekly.com/tools/agno.md)
Vendor: Agno

Agno's site describes an agent platform that runs in your cloud so you can build, deploy and improve customer-facing and internal agents. The repository is Apache-2.0; v3.1.10 was tagged Oct 23, 2026.

Verdict: Open-source agent framework that now sells a self-driving agent platform you run in your own cloud.
Good for: Teams that want to run agents in their own cloud
Skip if: You need a no-code builder

What's new:
- v3.1.10 (2026-10-23): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Agno
1. Get access: Install agno from pip and follow the docs; the platform needs sign-up via agno.com.
2. Check the limits and costs first: Open-source framework free. Paid: Platform pricing not read. No pricing page read.

Not verified:
- Platform pricing

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Agno 3.1.10 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.8 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- Agno: https://www.agno.com/
- Agno repository (GitHub): https://github.com/agno-agi/agno
Cite as: Agentic Weekly, "Agno v3.1.10: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/agno

## Claude Code v2.1.310 (2026-10-23, Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models))
URL: https://theagenticweekly.com/tools/claude-code (markdown: https://theagenticweekly.com/tools/claude-code.md)
Vendor: Anthropic

Claude Code 2.1.309 and 2.1.310 are heavy on sandbox and permission-check fixes, plus the Monitor tool now working on Bedrock, Vertex and Foundry and new Enterprise controls for cloud sessions.

Verdict: Update now if you use the sandbox or run Claude Code on a cloud provider; skip if you wanted flashy new features, this is a hardening release.
Good for: Anyone relying on the Bash sandbox or permission rules (the 2.1.309 fixes close real gaps); Teams running Claude Code through Bedrock, Vertex or Foundry; Enterprise admins who want role-based control over cloud sessions and Remote Control
Skip if: You are looking for new headline capabilities; these two releases are mostly fixes; You script claude ultrareview; check the new --trust-workspace requirement before updating CI

What's new:
- v2.1.310: The Monitor tool and plugin monitors now work on Bedrock, Vertex and Foundry, and when telemetry or nonessential traffic is disabled.
- v2.1.310: Requests no longer fail behind gateways that reject the effort beta value; Claude Code retries without it.
- v2.1.310: Claude Code cloud sessions and Remote Control permissions can be given to Enterprise custom roles.
- v2.1.310: On Linux, /sandbox now names Ubuntu's AppArmor user-namespace restriction when it stops the sandbox starting, and how to lift it.
- v2.1.309: Many security-flavored fixes: sandboxed Bash no longer runs shell startup files outside the sandbox, macOS sandbox denyWrite globs work through symlinks, and sandboxed commands can no longer modify the MCP config passed with --mcp-config.
- v2.1.309: Several Bash permission-check fixes for compound commands, redirects and worktree-isolated sessions.
- v2.1.309: claude ultrareview now requires workspace trust; scripts pass --trust-workspace.

Tutorial: Install Claude Code and check the sandbox
1. Install Claude Code: The README recommends the native installer. npm installation is marked deprecated.
2. Start it in your project: Navigate to your project directory and run claude, then sign in when prompted.
3. Update to the latest version: These fixes shipped in 2.1.309 and 2.1.310. Update using the method you installed with and check the version afterwards.
4. Review your sandbox setup: The release notes mention /sandbox and a Linux warning that names Ubuntu's AppArmor restriction on user namespaces when it stops bubblewrap from starting. Run /sandbox inside Claude Code to see your status. (Not tested by us.)

Not verified:
- Pricing and plan limits (not fetched)
- Whether the sandbox fixes affect your specific setup (reported from release notes, not tested)
- Sandbox fixes, ultrareview and behavior with Claude models (not tested; they need a signed-in session)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (4-bit), a small local open-weight model served by Ollama 0.40.11; no vendor-hosted model was used; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Installed with Anthropic's native installer (claude --version printed 2.1.310) and ran one headless edit task (change a-b to a+b in calc.py) through a local model; the first, vaguer prompt failed and the second, with the exact path and strings, succeeded.
- Not tested: Claude models, subscription login, the sandbox and permission fixes, claude ultrareview and --trust-workspace, Bedrock/Vertex/Foundry.

Sources:
- GitHub release v2.1.310: https://github.com/anthropics/claude-code/releases/tag/v2.1.310
- GitHub release v2.1.309: https://github.com/anthropics/claude-code/releases/tag/v2.1.309
- Claude Code README (install instructions): https://github.com/anthropics/claude-code
Cite as: Agentic Weekly, "Claude Code v2.1.310: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/claude-code

## Codex CLI 0.164.0 (2026-10-23, Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models))
URL: https://theagenticweekly.com/tools/codex-cli (markdown: https://theagenticweekly.com/tools/codex-cli.md)
Vendor: OpenAI

Codex CLI 0.164.0 adds numbered /presets slots for reusable prompts, inline arguments for /model and /permissions, image previews in supporting terminals, enterprise MCP sign-in and credential masking.

Verdict: A quality-of-life release: worth updating if you re-type the same prompts or switch models a lot, not a reason to switch tools.
Good for: People who already use Codex and repeat the same setup prompts (presets save the typing); Teams that connect Codex to enterprise MCP servers and want fewer sign-in hurdles; Anyone who wants an open-source (Apache-2.0) agent they can run from a terminal
Skip if: You are happy with another coding agent; nothing here changes the core capability; You need the image preview feature but your terminal does not support Kitty or Sixel graphics

What's new:
- Save reusable prompts and slash commands in ten numbered /presets slots, with a keyboard menu for quick access.
- Set model, reasoning effort, speed and permission profiles directly with inline /model and /permissions arguments.
- Preview viewed and generated images in the fullscreen transcript on terminals that support Kitty or Sixel graphics.
- Authenticate enterprise MCP connections using eligible Codex account grants or dedicated sign-in, including OAuth clients that require secrets.
- Mask credentials during supported remote Bash commands and Zsh snapshot sessions, including credentials from custom providers.
- The Amazon Bedrock setup wizard is now enabled by default.
- Bug fixes: failed compaction no longer leaves partial summaries in history, and resumed sessions restore saved thread settings instead of global defaults.

Tutorial: Install Codex CLI and try the new presets
1. Install Codex CLI: On Mac or Linux, run the official installer. On Windows use the PowerShell installer in the README. npm and Homebrew are also supported.
2. Start it and sign in: Run codex and choose Sign in with ChatGPT to use it with your ChatGPT plan (the README lists Plus, Pro, Business, Edu and Enterprise).
3. Check you are on 0.164.0 or newer: The features below shipped in 0.164.0. If you installed from a package manager, update it first.
4. Open the presets menu: The release notes say reusable prompts and slash commands can be saved in ten numbered /presets slots with a keyboard menu. Type /presets inside Codex and follow the on-screen prompts. (Not tested by us; the exact save flow is not in the release notes.)
5. Use inline arguments on /model and /permissions: Per the release notes, /model and /permissions now accept arguments inline so you can set model, reasoning effort, speed and permission profile in one command. The argument syntax is not in the notes; run /model and /permissions to see the options. (Not tested by us.)

Not verified:
- How to save a preset and the exact /model and /permissions argument syntax (not documented in the release notes)
- Pricing and plan limits for using Codex (not fetched)
- Presets, /model and /permissions arguments, MCP sign-in and image previews (not tested; they need an interactive, signed-in session)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (4-bit), a small local open-weight model served by Ollama 0.40.11; no vendor-hosted model was used; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Installed @openai/codex from npm (codex-cli 0.164.0) and ran 'codex exec --oss' against a local model; Codex executed a sed command in its sandbox and changed calc.py. A plain-language edit request ran no edit; we kept two runs.
- Not tested: ChatGPT sign-in, presets, /model and /permissions arguments, MCP sign-in, image previews, OpenAI-hosted models.

Sources:
- GitHub release rust-v0.164.0: https://github.com/openai/codex/releases/tag/rust-v0.164.0
- Codex README (install instructions): https://github.com/openai/codex
Cite as: Agentic Weekly, "Codex CLI 0.164.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/codex-cli

## Composio (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/composio (markdown: https://theagenticweekly.com/tools/composio.md)
Vendor: Composio

Composio's pricing page lists a free plan with 100,000 tool calls and 50,000 triggers a month, a $29 a month plan with $29 of monthly usage credit, and Enterprise; beyond free it lists $0.0003 per tool call. The SDK repository is MIT-licensed.

Verdict: Managed tool-calling and authentication for agents across 1,500+ apps, with a free tier of 100,000 tool calls a month.
Good for: Agent builders who need many app integrations with managed auth
Skip if: You need only one or two APIs

What's new:
- Latest release we saw (2026-10-23): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Composio
1. Get access: Install the SDK or CLI from composio.dev and connect an app.
2. Check the limits and costs first: Free: 100,000 tool calls and 50,000 triggers per month, no card. Paid: $29 per month with $29 usage credit; $0.0003 per tool call beyond free; Enterprise custom. Read from composio.dev/pricing on 2026-10-25.

Not verified:
- App count (vendor claim)

Sources:
- Composio: https://composio.dev/
- Composio pricing: https://composio.dev/pricing
- Composio repository (GitHub): https://github.com/ComposioHQ/composio
Cite as: Agentic Weekly, "Composio: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/composio

## Context7 @upstash/context7-sdk 0.6.0 (2026-10-23, Tested (smoke test, 2026-10-25, no language model involved))
URL: https://theagenticweekly.com/tools/context7 (markdown: https://theagenticweekly.com/tools/context7.md)
Vendor: Upstash

Context7 serves current, version-specific documentation and code examples to agents such as Claude, Codex and Cursor. The repository is MIT-licensed and an SDK 0.6.0 release was tagged Oct 23, 2026.

Verdict: Up-to-date, version-specific library docs served to coding agents.
Good for: Agents that hallucinate outdated library APIs
Skip if: Your stack is internal-only code

What's new:
- @upstash/context7-sdk 0.6.0 (2026-10-23): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Context7
1. Get access: Follow the setup instructions on context7.com for your MCP client.
2. Check the limits and costs first: Pricing not read. Pricing page was not read.

Test scope:
- smoke test, 2026-10-25; model: None; no language model was involved; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Context7 MCP 4.4.4 started locally in 462 ms but its tool calls go to a hosted API: resolve-library-id took about 1.2 s per call and returned React documentation matches.
- Not tested: Remote (HTTP) transport, authentication and concurrent clients: only local stdio with one client was measured; Large payloads: calls used tiny inputs, so these latencies are protocol and process overhead, not throughput; Anything about answer quality: no model was involved in the latency numbers; These are single smoke-test runs; it is not a benchmark. The machine was shared with other jobs; The second tool (fetching documentation by library ID) and API-key behaviour.

Sources:
- Context7 repository (GitHub): https://github.com/upstash/context7
- Context7: https://context7.com/
Cite as: Agentic Weekly, "Context7 @upstash/context7-sdk 0.6.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/context7

## CrewAI 1.15.30 (2026-10-23, Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models))
URL: https://theagenticweekly.com/tools/crewai (markdown: https://theagenticweekly.com/tools/crewai.md)
Vendor: CrewAI

CrewAI 1.15.29 promotes AgentExecutor out of experimental status and adds LiteLLM-compatible kwargs, Azure DocumentDB vector search and new CLI evaluation features; 1.15.30 followed with a small fix on Oct 24.

Verdict: Reasonable if you like the crew-of-agents mental model; the memory and tool-loop fixes in this release are worth taking.
Good for: Developers who want role-based multi-agent flows; Teams that want a built-in eval command; Users who need vector search on Azure DocumentDB
Skip if: You prefer an explicit graph model; You need a stable API with few releases; versions move quickly

What's new:
- 1.15.29 (Oct 23): AgentExecutor is no longer experimental.
- 1.15.29: Support for LiteLLM-compatible kwargs; new Azure DocumentDB vector search integration.
- 1.15.29: crewAI eval can run the crew itself for a verdict from one command and export evaluations as files.
- 1.15.29: Fixes include private memories leaking into every recall, repeated identical tool calls, and LanceDB filter escaping.
- 1.15.30 (Oct 24): Fix to the AGENTS.md login flow while evaluating.

Tutorial: Install CrewAI with uv
1. Install uv: The README installs CrewAI with the uv tool installer, so install uv first.
2. Install CrewAI: Install it as a uv tool and refresh your shell path.
3. Check the install: List your uv tools to confirm CrewAI is present, then follow the README's project scaffolding steps.

Not verified:
- Pricing of CrewAI's hosted platform (not covered)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (4-bit), a small local open-weight model served by Ollama 0.40.11; no vendor-hosted model was used; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Installed crewai 1.15.30 in a fresh Python 3.13 venv and ran one Agent, one Task and one Crew with a custom tool; the agent used the tool and reported 9 words (correct). An earlier unpinned install resolved to 1.6.1. Extended: CrewAI 1.15.30 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 8.4 s.
- Not tested: The uv tool install / crewai CLI route, the eval command, Azure DocumentDB, the hosted platform, hosted models; Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- CrewAI 1.15.29 release: https://github.com/crewAIInc/crewAI/releases/tag/1.15.29
- CrewAI 1.15.30 release: https://github.com/crewAIInc/crewAI/releases/tag/1.15.30
- README: https://github.com/crewAIInc/crewAI
Cite as: Agentic Weekly, "CrewAI 1.15.30: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/crewai

## Dify v1.18.0 (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/dify (markdown: https://theagenticweekly.com/tools/dify.md)
Vendor: LangGenius

Dify's LICENSE file says it is a modified Apache License 2.0 with additional conditions, including a requirement for a commercial license for certain multi-tenant service use. v1.18.0 was tagged Oct 23, 2026.

Verdict: Open-source platform for building LLM apps and agent workflows with a visual canvas; licensed under a modified Apache 2.0 with extra conditions.
Good for: Teams that want a visual builder they can self-host; Internal RAG and agent apps
Skip if: You need a standard OSI-approved license for redistribution

What's new:
- v1.18.0 (2026-10-23): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Dify
1. Get access: Self-host with Docker Compose per the repository's README or sign up for the cloud.
2. Check the limits and costs first: Open-source self-hosted version; hosted plans exist. Paid: Cloud plan prices not read. Pricing page not read.

Not verified:
- Cloud pricing

Sources:
- Dify: https://dify.ai/
- Dify license: https://github.com/langgenius/dify/blob/main/LICENSE
- Dify repository (GitHub): https://github.com/langgenius/dify
Cite as: Agentic Weekly, "Dify v1.18.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/dify

## Gemini Notebook (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/gemini-notebook (markdown: https://theagenticweekly.com/tools/gemini-notebook.md)
Vendor: Google

Google added guided creation for Slide Decks and Video Overviews in Gemini Notebook (Oct 19-21) and, on Oct 23, a Connected App so you can manage notebooks from the Gemini app. The help center now calls the product Gemini Notebook.

Verdict: A good fit for turning a pile of documents into drafts and decks you can check against sources; availability of new features is rolling out.
Good for: Students and researchers working from their own sources; Writers who want a briefing doc or slide deck from notes; Google Workspace users
Skip if: You want a general chatbot with no source upload step; You are in the EEA and want the Gemini app connection on by default

What's new:
- Oct 19-21: Guided creation for Slide Decks and Video Overviews, with suggested topics and styles from your sources and a Review toggle to check an outline or producer notes before rendering.
- Oct 23: Gemini Notebook is a Connected App in the Gemini app; type @Gemini Notebook to create notebooks, add sources, save notes or retrieve information. On by default except in the EEA.
- Oct 16: Study features: 60-second Short Video Overviews, Interactive Learning Overviews and more quiz formats (short-answer, multi-select, fill-in-the-blank).
- Notebooks can generate reports, data tables (export to Sheets), flashcards, quizzes, slide decks, infographics and mind maps from your sources.

Tutorial: Create your first notebook
1. Create a notebook: Open Gemini Notebook and choose Create new notebook. A notebook is a collection of sources for one project.
2. Add sources: Upload a source or discover new ones. The Chat panel shows a summary of all sources and answers questions about them.
3. Generate and review an output: In the Studio panel create a report, slide deck or video overview. Turn on the Review toggle to check the outline or producer notes first.

Not verified:
- Pricing and usage limits
- Output quality (not tested)

Sources:
- Gemini Notebook co-creation announcement (Oct 19, 2026): https://blog.google/innovation-and-ai/products/gemini-notebook/co-create-slides-videos/
- Connected App (Oct 23, 2026): https://workspaceupdates.googleblog.com/2026/10/take-actions-on-your-notebooks-from-Gemini-Notebook-directly-in-the-Gemini-app.html
- Study features (Oct 16, 2026): https://blog.google/innovation-and-ai/products/gemini-notebook/three-tips-study-gemini-notebook/
- Google blog: NotebookLM is now Gemini Notebook: https://blog.google/innovation-and-ai/products/gemini-notebook/notebooklm-gemini-notebook/
- Google Workspace Updates: NotebookLM is now Gemini Notebook: https://workspaceupdates.googleblog.com/2026/07/notebooklm-now-gemini-notebook.html
Cite as: Agentic Weekly, "Gemini Notebook: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/gemini-notebook

## goose v1.56.0 (2026-10-23, Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models))
URL: https://theagenticweekly.com/tools/goose (markdown: https://theagenticweekly.com/tools/goose.md)
Vendor: aaif-goose (GitHub organization, previously block/goose)

goose v1.56.0 focuses on subagents: you can resume failed or cancelled delegated subagents, view subagent sessions in the UI and cap concurrency. It also adds a Windows installer for the desktop app.

Verdict: Pick goose if you want an open-source agent with real subagent controls; v1.56.0 is a solid, subagent-heavy update.
Good for: People who want an Apache-2.0 agent with a desktop app and CLI; Users who delegate work to subagents and need resume and limits; Windows desktop users who were waiting for an installer
Skip if: You want a tiny tool with few moving parts; You need a managed, supported product

What's new:
- v1.56.0 (Oct 23): Resume failed or cancelled delegate subagents.
- v1.56.0: Limit concurrent subagents with GOOSE_MAX_CONCURRENT_SUBAGENTS, and open view-only subagent sessions from the UI.
- v1.56.0: Automatic thinking level operation for improved model reasoning.
- v1.56.0: Windows NSIS installer for the desktop app, and a new IO Intelligence provider.
- v1.56.0: Replaces PCTX code mode with a Boa JavaScript runtime.

Tutorial: Install goose and run your first session
1. Install the CLI: The README gives a one-line installer. Read the script first if you prefer.
2. Pick a provider: Run goose and follow the prompts to connect a model provider. goose works with any LLM according to its description.
3. Try subagents: If you use delegated subagents, v1.56.0 lets you cap how many run at once with an environment variable.

Not verified:
- Any pricing from hosted providers
- Real-world quality (not tested)
- Whether the thinking-level automation helps in practice
- Subagent controls, resume and the desktop app (not tested)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (4-bit), a small local open-weight model served by Ollama 0.40.11; no vendor-hosted model was used; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Unpacked the official release archive (the install script stopped on a missing bzip2), goose --version printed 1.56.0, and ran 'goose run' with a local Ollama provider; attempt 1 failed to change the file, attempt 2 with exact strings changed a-b to a+b.
- Not tested: Subagents, resume and limits, the desktop app.

Sources:
- goose v1.56.0 release: https://github.com/aaif-goose/goose/releases/tag/v1.56.0
- goose README: https://github.com/aaif-goose/goose
Cite as: Agentic Weekly, "goose v1.56.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/goose

## Kilo Code v7.8.17 (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/kilo-code (markdown: https://theagenticweekly.com/tools/kilo-code.md)
Vendor: Kilo Code (Kilo-Org)

Kilo's site describes an open-source coding agent for VS Code, JetBrains, a CLI and the cloud, with 500+ models at zero markup, your own keys or local models. The repository shows the MIT license and release v7.8.17 on Oct 23, 2026.

Verdict: Open-source AI coding agent for VS Code, JetBrains, CLI and Cloud, with 500+ models at zero markup or your own keys.
Good for: Developers who want an open-source agent with many models; Teams that want shared agents and centralized billing
Skip if: You want a single-vendor model bundle with a flat monthly price

What's new:
- v7.8.17 (2026-10-23): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Kilo Code
1. Get access: Install the Kilo extension for VS Code or JetBrains, or the CLI, from kilo.ai.
2. Check the limits and costs first: Free for individuals; AI usage billed separately (free models, local models, BYOK or Kilo Gateway) Paid: Teams $15 per user per month; Enterprise custom. Read from kilo.ai/pricing on 2026-10-25. Gateway usage is billed at provider rates.

Not verified:
- Agent quality (not tested)

Sources:
- Kilo: https://kilo.ai/
- Kilo pricing: https://kilo.ai/pricing
- Kilo Code repository: https://github.com/Kilo-Org/kilocode
Cite as: Agentic Weekly, "Kilo Code v7.8.17: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/kilo-code

## Luma (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/luma (markdown: https://theagenticweekly.com/tools/luma.md)
Vendor: Luma AI

Luma added Kling 4.0 for every customer on Oct 23, next to its own Ray 3.2 model, which offers up to 16 keyframes per clip, 16-bit EXR export and a new API.

Verdict: A good pick for creators who want several video models in one workspace; Ray 3.2 launched on June 9, 2026, so check which models are live on your plan.
Good for: Video creators who want to switch models per shot; Studios that need HDR and EXR outputs for post production; Developers who want a video-generation API
Skip if: You want pay-as-you-go pricing; plans are credit-based; You need unlimited generations; plans cap monthly credits

What's new:
- Oct 23: Kling 4.0 is now in Luma for every customer, aimed at fast action and big camera moves, with text-to-video and image-to-video in 16:9, 9:16, 1:1 and 21:9.
- Ray 3.2 (announced June 9, 2026): frame-level control with up to 16 keyframes in one clip, up to 20-second 1080p clips, native HDR and 16-bit EXR export.
- Ray 3.2 adds expressive facial performance tracking for up to eight faces and a Reframe tool to change aspect ratio, extend the frame or replace a background.
- Ray 3.2 is also available as an API for the first time, according to Luma.

Tutorial: Try Kling 4.0 inside Luma
1. Open a board: Per Luma's post, open the board for your project in Luma, or start a new one.
2. Choose the model: Pick Kling 4.0 for the shot that needs more movement and describe the shot, or start from a still of your talent or product.
3. Judge and refine: Review the take and refine it in the same board next to the rest of the cut.

Not verified:
- Video quality (not tested)

Sources:
- Kling 4.0 in Luma (Oct 23, 2026): https://lumalabs.ai/news/kling-4-in-luma-ai-video
- Introducing Ray 3.2: https://lumalabs.ai/news/introducing-ray-3-2
- Luma pricing: https://lumalabs.ai/pricing
Cite as: Agentic Weekly, "Luma: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/luma

## OpenHands v1.33.0 (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/openhands (markdown: https://theagenticweekly.com/tools/openhands.md)
Vendor: OpenHands (open-source project)

OpenHands describes itself as an open-source AI coding agent platform: start locally, automate engineering workflows, use any model and scale with enterprise controls. The repository shows the MIT license and release v1.33.0 on Oct 23, 2026.

Verdict: Open-source AI coding agent platform you can start locally, automate with workflows and run with any model.
Good for: Teams that want an open-source alternative to cloud coding agents; Developers who want to bring their own model
Skip if: You want a flat-fee hosted product with published pricing we can quote

What's new:
- v1.33.0 (2026-10-23): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with OpenHands
1. Get access: Follow the quick-start on openhands.dev to run it locally and connect a model.
2. Check the limits and costs first: Open-source software is free; model usage costs depend on your provider. Paid: Cloud and enterprise pricing not read. Pricing page not read; only the open-source license is verified.

Not verified:
- Cloud pricing
- Anything beyond the attempt in /tests/openhands-cli: we ran it once on a local 3B model and it did not complete our task, so it stays Reported

Sources:
- OpenHands: https://www.openhands.dev/
- OpenHands repository: https://github.com/OpenHands/OpenHands
- OpenHands v1.33.0 release: https://github.com/OpenHands/OpenHands/releases/tag/v1.33.0
- GitHub API data for All-Hands-AI/OpenHands (stars, license, last push): https://github.com/All-Hands-AI/OpenHands
Cite as: Agentic Weekly, "OpenHands v1.33.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/openhands

## Replit (2026-10-23, Reported (not run by us))
URL: https://theagenticweekly.com/tools/replit (markdown: https://theagenticweekly.com/tools/replit.md)
Vendor: Replit

Replit's Oct 23 update adds a Monitoring Agent that investigates production issues from logs and metrics and can attempt fixes, plus reusable workspace skills, chat and asset sharing, and Stripe Atlas incorporation help.

Verdict: Strong pick if you want build, hosting and monitoring in one place; read the credit and overage rules first.
Good for: Builders who want an agent that also watches the app after launch; Solo founders who want incorporation help inside the tool; Teams that want shared chats and reusable skills
Skip if: You want to keep code and hosting on your own infrastructure; You need predictable spend; credits and overage limits apply

What's new:
- Monitoring Agent: investigates production issues using your app's logs and metrics, then attempts and validates fixes. Turn on Monitoring alerts and Auto-fix; publishing fixes automatically needs the experimental Auto-republish setting.
- Agent can create and update reusable workspace skills from your conversations and projects, which you review before saving.
- Share a whole chat or a single asset with selected teammates.
- Stripe Atlas on Replit: ask Agent to help incorporate through Stripe Atlas, with 20% off when you incorporate through Replit.
- Eligible top-up customers can keep published apps running at zero credits up to an overage limit; paid Agent pauses and Free mode stays available.

Tutorial: Start building on Replit
1. Read the get-started guide: Replit's docs walk through a coffee-shop launch from idea to a published opening site. It is the clearest way to see the flow.
2. Start a chat or Project: Use a chat for questions and research, or a Project for substantial build and design work. The docs explain when to continue in a chat and when to use a Project.
3. Turn on Monitoring (optional): Monitoring alerts and Auto-fix are settings you enable per app. Review fixes yourself unless you deliberately enable the experimental Auto-republish setting.

Not verified:
- Core's monthly sticker price and annual discount (the docs state credits, not the price; the pricing page returned 403)
- Monitoring Agent quality (not tested)
- Whether the Stripe Atlas discount applies to your situation

Sources:
- Replit updates (Oct 23, 2026): https://docs.replit.com/updates
- Replit docs: Get started: https://docs.replit.com/home/get-started.md
- Replit pricing and plans (docs): https://docs.replit.com/help/pricing-and-plans
Cite as: Agentic Weekly, "Replit: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/replit

## Chrome DevTools MCP v1.12.0 (2026-10-22, Tested (smoke test, 2026-10-25, no language model involved))
URL: https://theagenticweekly.com/tools/chrome-devtools-mcp (markdown: https://theagenticweekly.com/tools/chrome-devtools-mcp.md)
Vendor: Chrome DevTools team (Google)

Chrome DevTools MCP exposes DevTools to coding agents. It is Apache-2.0 licensed and v1.12.0 was released Oct 22, 2026.

Verdict: Chrome DevTools for coding agents.
Good for: Debugging and performance work in a real Chrome
Skip if: You want cross-browser test automation

What's new:
- v1.12.0 (2026-10-22): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Chrome DevTools MCP
1. Get access: Follow the README to register the server with your MCP client.
2. Check the limits and costs first: Free. Open-source server; the service it connects to may have its own costs.

Test scope:
- smoke test, 2026-10-25; model: None; no language model was involved; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Chrome DevTools MCP 1.12.0 exposed 32 tools and opened a page in headless Chrome in 136.6 ms median; our first navigate_page call was rejected for a missing pageId.
- Not tested: Remote (HTTP) transport, authentication and concurrent clients: only local stdio with one client was measured; Large payloads: calls used tiny inputs, so these latencies are protocol and process overhead, not throughput; Anything about answer quality: no model was involved in the latency numbers; These are single smoke-test runs; it is not a benchmark. The machine was shared with other jobs; Performance traces, network and console inspection, emulation: only page creation was timed.

Sources:
- Chrome DevTools MCP repository (GitHub): https://github.com/ChromeDevTools/chrome-devtools-mcp
Cite as: Agentic Weekly, "Chrome DevTools MCP v1.12.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/chrome-devtools-mcp

## Google ADK v2.13.0 (2026-10-22, Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models))
URL: https://theagenticweekly.com/tools/google-adk (markdown: https://theagenticweekly.com/tools/google-adk.md)
Vendor: Google

ADK for Python 2.13.0 promotes 12 experimental features to stable, adds a classifier node for fast workflow routing, live audio with barge-in, and MongoDB session and memory services.

Verdict: Update with care: a big, useful release for Gemini users, but it includes a telemetry default change.
Good for: Python teams building on Gemini or Google Cloud; Developers who need live voice agents; Teams who want MongoDB-backed sessions and memory
Skip if: You need identical telemetry across upgrades; defaults changed; You are not using Python

What's new:
- 2.13.0 (Oct 22): 12 mature experimental features are now stable.
- New TypesafeClassifier and ClassifierNode route workflows with calibrated probabilities in one forward pass instead of an LLM call.
- Live audio interactions with barge-in alongside Gemini speech-to-text and text-to-speech.
- MongoDbSessionService and MongoDbMemoryService (experimental) persist sessions and long-term memory.
- Up to 8x faster execution turns for string-model agents. Breaking: telemetry now defaults to OpenTelemetry GenAI v2.

Tutorial: Install ADK
1. Install the package: The README gives pip. If you hit dependency conflicts it also documents a constraints file.
2. Optional extras: Install the extensions bundle if you need integrations.
3. Read the 2.13.0 breaking changes: Telemetry defaults to OpenTelemetry GenAI v2 and removes call_llm. Check the release notes before upgrading production agents.

Not verified:
- Support for non-Google models
- Real-world performance claims (the 8x figure is the vendor's)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (4-bit), a small local open-weight model served by Ollama 0.40.11; no vendor-hosted model was used; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Installed google-adk 2.13.0 (plus litellm for a non-Gemini model) and ran an Agent with one Python function tool on InMemoryRunner; the agent called the tool and answered 6 (correct). Extended: Google ADK (Python) 2.13.0 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.7 s.
- Not tested: Gemini models, live voice, MongoDB sessions, the 2.13.0 telemetry default change, the vendor's 8x claim; Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- ADK 2.13.0 release: https://github.com/google/adk-python/releases/tag/v2.13.0
- README: https://github.com/google/adk-python
Cite as: Agentic Weekly, "Google ADK v2.13.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/google-adk

## LangChain langchain-core 1.7.0 (2026-10-22, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/langchain (markdown: https://theagenticweekly.com/tools/langchain.md)
Vendor: LangChain

LangChain's repository is MIT-licensed; the newest release we saw was langchain-core 1.7.0 on Oct 22, 2026. LangChain is the base library and integration layer; LangGraph (already covered here) is the lower-level orchestration runtime for stateful agents.

Verdict: Open-source framework for building LLM applications and agents in Python and JavaScript, from the team behind LangGraph and LangSmith.
Good for: Developers who want the widest set of integrations; Teams that will pair it with LangGraph for stateful agents
Skip if: You want a minimal library with few abstractions; You only need one model provider's SDK

What's new:
- langchain-core 1.7.0 (2026-10-22): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with LangChain
1. Get access: Install the package (pip install langchain or npm install langchain) and follow the quickstart in the official docs.
2. Check the limits and costs first: Free to self-host. Open-source library; costs come from the model APIs you call.

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: LangChain.js 1.5.19 passed our one-tool task 3 of 3 on Node 20 and 3 of 3 on Node 24; about 101 MB in 36 packages. LangChain 1.4.7 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 0.8 s.
- Not tested: Any vendor-hosted model: everything ran against one local 3B model; Multi-agent flows, memory, streaming and deployment; Anything beyond one tool and one question; five trials is a smoke test, not a benchmark; Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- LangChain: https://www.langchain.com/
- LangChain repository (GitHub): https://github.com/langchain-ai/langchain
Cite as: Agentic Weekly, "LangChain langchain-core 1.7.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/langchain

## opencode v1.19.4 (2026-10-22, Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models))
URL: https://theagenticweekly.com/tools/opencode (markdown: https://theagenticweekly.com/tools/opencode.md)
Vendor: anomalyco (GitHub organization)

opencode v1.19.4 is a small maintenance release: it upgrades the bundled Bun runtime to 1.4.4, which the release notes say includes security fixes. The project itself is very active and widely starred.

Verdict: Update when convenient for the runtime security fixes; pick opencode if you want an open-source agent that is not tied to one model vendor.
Good for: Developers who want an MIT-licensed terminal agent; People who switch between model providers; Anyone who wants a desktop app as well as a CLI
Skip if: You need a vendor-backed support contract; You expected feature news; this release is mostly a runtime upgrade

What's new:
- v1.19.4 (Oct 22): Upgraded the bundled Bun runtime to 1.4.4, which the notes say includes security fixes for Brotli, TLS, HTTP/1.1, HTTP/2, HTTP/3, WebSocket and fetch, and updates root certificates (NSS 3.131.1).
- The README lists several install routes: an install script, npm, Homebrew, and a desktop app cask.

Tutorial: Install opencode and start it in a repo
1. Install it: The README recommends the install script, and also lists npm and Homebrew. Check the README for the current command before running a script from the internet.
2. Or use a package manager: Homebrew on macOS and Linux, or npm globally.
3. Run it in your project: Open a terminal in your project folder and start the agent, then connect a model provider when prompted.

Not verified:
- Pricing for any hosted opencode services
- Real-world quality versus other agents (not tested)
- The full feature list of the v2 CLI (README only skimmed)
- The v2 CLI install routes and desktop app (not tested; we ran opencode-ai 1.19.4 from npm)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (4-bit), a small local open-weight model served by Ollama 0.40.11; no vendor-hosted model was used; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Installed opencode-ai 1.19.4 from npm and ran 'opencode run' headless with a custom OpenAI-compatible provider pointing at a local model; attempt 1 edited the wrong path and failed, attempt 2 with the path stated edited calc.py.
- Not tested: The README's v2 installer or @opencode/cli 2.x, the desktop app, the Bun security fixes.

Sources:
- opencode v1.19.4 release: https://github.com/anomalyco/opencode/releases/tag/v1.19.4
- opencode README: https://github.com/anomalyco/opencode
Cite as: Agentic Weekly, "opencode v1.19.4: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/opencode

## Claude (2026-10-21, Reported (not run by us))
URL: https://theagenticweekly.com/tools/claude (markdown: https://theagenticweekly.com/tools/claude.md)
Vendor: Anthropic

Anthropic's Oct 21 release notes announce Claude Duet, a research preview of a proactive agent app for Pro and Max users; earlier in October came Claude Haiku 5.5 and monthly API credits for Max and Team plans.

Verdict: A strong writing partner; try Claude Duet only if you are on Pro or Max and comfortable with a research-preview product.
Good for: Writers who want a careful, long-form drafting assistant; Pro and Max users curious about proactive agents; Max and Team users who also build on the API
Skip if: You need Duet and are on Free (it is Pro and Max only); You want published benchmarks; we cite none

What's new:
- Oct 21: Claude Duet research preview for Pro and Max: a new app where Claude works as a proactive agent.
- Oct 7: Claude Haiku 5.5, described as the cheapest, fastest and most capable small model Anthropic has released, for high-volume, cost-sensitive tasks.
- Oct 7: Max and Team plans now include monthly API credits for running your own apps and agents on the Claude Platform; claim them by linking a Console organization in billing settings.
- Claude Duet details (official announcement): research preview by waitlist, starting with US Pro and Max subscribers; standalone app on iOS and Android and on claude.ai; included in Pro and Max with the same usage limits. It sends a morning brief, watches connected apps, completes tasks in the background and checks with you before acting. Reuters and CNBC covered it on Oct 21.

Tutorial: Get started with Claude
1. Open the app: Go to claude.ai and sign in or create an account.
2. Paste your draft and say who it is for: Give Claude the draft, the audience and the tone you want, and ask for specific edits rather than a full rewrite.
3. Check release notes for Duet: Claude Duet is a research preview for Pro and Max. Read the linked blog post from the release notes before trusting it with sensitive tasks.

Not verified:
- How well Claude Duet works (preview by waitlist; not tested)
- Writing quality (not tested)

Sources:
- Claude release notes: https://support.claude.com/en/articles/12138966-release-notes
- Introducing Claude Duet: https://claude.com/resources/articles/claude-duet
- Claude pricing: https://claude.com/pricing
Cite as: Agentic Weekly, "Claude: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/claude

## GitHub MCP Server 2.0.3 (2026-10-21, Tested (smoke test, 2026-10-26, no LLM involved))
URL: https://theagenticweekly.com/tools/github-mcp-server (markdown: https://theagenticweekly.com/tools/github-mcp-server.md)
Vendor: GitHub

GitHub's official MCP server. The repository is MIT-licensed, maintained by GitHub, and release 2.0.3 was published Oct 21, 2026.

Verdict: GitHub's official MCP server for repos, issues and pull requests.
Good for: Agents that need to read and act on GitHub repositories
Skip if: You do not use GitHub

What's new:
- 2.0.3 (2026-10-21): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with GitHub MCP Server
1. Get access: Follow the install section of the repository README for your MCP client.
2. Check the limits and costs first: Free. Server is free; GitHub account limits and plans apply.

Not verified:
- Behavior in practice (we have not run this tool)

Test scope:
- smoke test, 2026-10-26; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: GitHub MCP Server 2.0.3 started over stdio in a median 266 ms and listed 46 tools in a median 15 ms; we did not make any authenticated call.
- Not tested: Any authenticated call: we used a placeholder token, so no request reached the vendor's API and we did not check that real accounts, permissions or rate limits work; Remote (HTTP) transport and concurrent clients: only local stdio with one client was run; Anything about answer quality: no model was involved in the handshake and tool-list check; A single smoke-test run set (three starts); it is not a benchmark.

Sources:
- GitHub MCP Server repository (GitHub): https://github.com/github/github-mcp-server
Cite as: Agentic Weekly, "GitHub MCP Server 2.0.3: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/github-mcp-server

## Haystack v3.4.1 (2026-10-21, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/haystack (markdown: https://theagenticweekly.com/tools/haystack.md)
Vendor: deepset

Haystack's repository is Apache-2.0 licensed with v3.4.1 tagged Oct 21, 2026 and a push on Oct 24, 2026.

Verdict: Apache-2.0 Python framework for building search, RAG and agent pipelines, maintained by deepset.
Good for: Production RAG pipelines; Teams that like explicit pipeline components
Skip if: You want a visual builder

What's new:
- v3.4.1 (2026-10-21): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Haystack
1. Get access: Install haystack-ai with pip and follow the quickstart in the docs.
2. Check the limits and costs first: Free to self-host. Open-source library; costs come from the model APIs you call.

Not verified:
- Hosted platform pricing

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Haystack 3.4.1 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 2.3 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark; Document stores, retrieval pipelines and evaluation.

Sources:
- Haystack: https://haystack.deepset.ai/
- Haystack repository (GitHub): https://github.com/deepset-ai/haystack
Cite as: Agentic Weekly, "Haystack v3.4.1: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/haystack

## Mastra @mastra/core@1.79.0 (2026-10-21, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/mastra (markdown: https://theagenticweekly.com/tools/mastra.md)
Vendor: Mastra

@mastra/core 1.79.0 adds experimental MCP Events, richer observability (span aggregation and time-to-first-chunk), session-ID based agent controller sessions (a breaking change) and a native filesystem API for sandboxes.

Verdict: A strong TypeScript option if you are building agents in a Node stack; read the breaking-change note on sessions before upgrading.
Good for: TypeScript and Next.js teams building agents; Teams that need observability and tracing out of the box; Developers experimenting with MCP events
Skip if: You work in Python; You need a stable session API; this release changes it

What's new:
- @mastra/core@1.79.0 (Oct 21): Experimental MCP Events let MCP servers publish event streams that clients can poll, stream or receive by signed webhook, with Studio inspection under mastra dev.
- Observability: new aggregateSpans() for per-span analytics such as p95 latency per tool, plus timeToFirstChunkMs fields and richer querySpans filters.
- Agent controller sessions are now addressed by sessionId, allowing many sessions per resource (breaking change).
- Sandboxes can expose a native filesystem API for faster file operations.
- @mastra/connect adds per-end-user connection scoping and a Microsoft Teams channel.

Tutorial: Create a Mastra project
1. Scaffold a project: The README starts with the create command.
2. Run the local Studio: Mastra has a dev command that starts Studio so you can inspect agents. The release notes mention mastra dev for inspecting MCP events.
3. Read the breaking changes: Before upgrading an existing app to @mastra/core 1.79.0, read the release notes on controller sessions now keyed by sessionId.

Not verified:
- License terms (GitHub reports NOASSERTION)

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Mastra passed 5 of 5 on Node 20 (0.24.9) and 5 of 5 on Node 24 (1.79.0) with temperature pinned to 0, after failing 5 of 6 earlier trials at default sampling.
- Not tested: Any vendor-hosted model: everything ran against one local 3B model; Multi-agent flows, memory, streaming and deployment; Anything beyond one tool and one question; five trials is a smoke test, not a benchmark.

Sources:
- @mastra/core@1.79.0 release: https://github.com/mastra-ai/mastra/releases/tag/%40mastra/core%401.79.0
- README: https://github.com/mastra-ai/mastra
Cite as: Agentic Weekly, "Mastra @mastra/core@1.79.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/mastra

## MCP Inspector 2.12.0 (2026-10-21, Tested (smoke test, 2026-10-25, no LLM involved))
URL: https://theagenticweekly.com/tools/mcp-inspector (markdown: https://theagenticweekly.com/tools/mcp-inspector.md)
Vendor: Model Context Protocol project

MCP Inspector is the official tool for connecting to an MCP server and calling its tools. The GitHub license field reads NOASSERTION; 2.12.0 was tagged on 2026-10-21 and needs a recent Node (declares >=22.19).

Verdict: The quickest way to check what an MCP server exposes; run it on a current Node or npm may quietly install version 0.15.0.
Good for: Anyone building or evaluating an MCP server
Skip if: You only use servers through a client and never inspect them

What's new:
- 2.12.0 (2026-10-21): the newest release on the repository when we read it on 2026-10-25.

Tutorial: Get started with MCP Inspector
1. Run the CLI against a server: Use absolute paths. On Node 24 npm installed 2.12.0; on Node 20 it installed 0.15.0.
2. Check the version: Always print the version you got before following a tutorial.

Not verified:
- Web UI, OAuth flows, remote transports

Test scope:
- smoke test, 2026-10-25; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: MCP Inspector listed the filesystem server's 14 tools from the command line on both 0.15.0 (Node 20) and 2.12.0 (Node 24) in under a second; the 2.x install is smaller.
- Not tested: The browser UI, OAuth flows, SSE and streamable HTTP transports; Calling tools through the Inspector: only tools/list was run; One run each; not a benchmark.

Sources:
- MCP Inspector documentation: https://modelcontextprotocol.io/docs/2026-07-28/tools/inspector
- MCP Inspector repository (GitHub): https://github.com/modelcontextprotocol/inspector
Cite as: Agentic Weekly, "MCP Inspector 2.12.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/mcp-inspector

## ChatGPT (2026-10-20, Reported (not run by us))
URL: https://theagenticweekly.com/tools/chatgpt (markdown: https://theagenticweekly.com/tools/chatgpt.md)
Vendor: OpenAI

OpenAI's Oct 20 release notes add locked chats, ChatGPT Slides for Pro, Business and Enterprise users in Space, and updates to GPT-6 Pro; earlier in October came GPT-6 with Intelligent UI (Oct 7) and audio uploads (Oct 6).

Verdict: Still the broadest all-round assistant; the October updates matter most if you write documents, decks or work with audio.
Good for: Writers who need drafts, decks and meeting notes in one place; Teams already on ChatGPT Business or Enterprise; People who want image generation (Images 2.5) in the same app
Skip if: You want strict data isolation by default (check settings); You need features only on Pro plans and are on a lower tier

What's new:
- Oct 20: Locked chats hide selected conversations from history and search behind a 4-digit PIN or device authentication; rolling out to Free, Go, Plus and Pro.
- Oct 20: ChatGPT Slides lets you create decks and collaborate in Space, with export to PowerPoint, PDF or Google Slides on the web; for eligible Pro, Business and Enterprise users.
- Oct 20: GPT-6 Pro in chat gets faster responses, better tool use and parallel agents for Pro users.
- Oct 7: GPT-6 with Intelligent UI can mix text, visuals and interactive elements such as calculators; Plus, Pro, Business and Enterprise use GPT-6 Sol; Free and Go use GPT-6 Luna.
- Oct 6: Upload audio files for transcripts and summaries on paid plans.
- Sep 29: A new Pro 500 plan at $500/month; Pro 200 at $200/month is available again. OpenAI's pricing docs now list Pro plans at $100, $200 or $500 per month.

Tutorial: Use ChatGPT for a first draft and a deck
1. Draft in chat: Describe the document, audience and tone. The June release notes describe full-screen writing blocks for essays, reports and blog posts, which you can save to your Library.
2. Add a source: Upload a file or audio recording (paid plans for audio) and ask for a summary or structured notes.
3. Turn it into slides: If you have access to Space, ask ChatGPT to create a presentation, edit it, and export to PowerPoint, PDF or Google Slides.

Not verified:
- Writing quality compared with other assistants (not tested)
- Whether new features are available on your account

Sources:
- ChatGPT release notes: https://help.openai.com/en/articles/6825453-chatgpt-release-notes
- ChatGPT pricing (OpenAI docs): https://learn.chatgpt.com/docs/pricing.md
Cite as: Agentic Weekly, "ChatGPT: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/chatgpt

## FastMCP v4.1.3 (2026-10-20, Tested (smoke test, 2026-10-25, no language model involved))
URL: https://theagenticweekly.com/tools/fastmcp (markdown: https://theagenticweekly.com/tools/fastmcp.md)
Vendor: Prefect

FastMCP 4.1.2 and 4.1.3 (both Oct 20) are security releases: component authorization now applies after transforms, OAuth client credentials can bind to an issuer, and 4.1.3 updates certificate handling after a CA incident.

Verdict: If you run FastMCP servers, upgrade and read the security notices; the OAuth issuer step is not automatic.
Good for: Python developers building MCP servers or clients; Teams exposing internal tools to agents; Anyone running OAuth-protected MCP servers
Skip if: You do not use MCP; You cannot update trust stores in your deployment environment

What's new:
- 4.1.2 (Oct 20): Component authorization now applies after transforms and denied resources cannot fall through matching templates.
- 4.1.2: OAuth client credentials providers support issuer binding, but you must set the issuer to enable the protection.
- 4.1.2: Foundation for the SEP-2640 Skills extension and experimental skills discovery.
- 4.1.3: Updates certifi-backed dependencies after the eMudhra CA compromise; FastMCP 4 uses system trust by default, so you must update OS, container or custom trust stores too.

Tutorial: Add FastMCP to a project
1. Install it: The README recommends uv.
2. Set the issuer if you use client credentials OAuth: Per the 4.1.2 notes, ClientCredentialsOAuthProvider and PrivateKeyJWTOAuthProvider users must set issuer to enable the new protection.
3. Update system trust stores: The 4.1.3 notes say upgrading FastMCP alone is not enough; follow the repository's SECURITY.md certificate-authority incident notice.

Not verified:
- Whether your deployment is affected by the CA incident

Test scope:
- smoke test, 2026-10-25; model: None; no language model was involved; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: First run (2026-10-25, 17:35 to 17:55 CDT; raw log not published on a /tests page): installed fastmcp 4.1.3 with uv, wrote a two-tool server and called it through an in-memory client, a stdio subprocess and HTTP on 127.0.0.1; fastmcp inspect listed 2 tools. Published log (/tests/mcp-python-servers, stdio only): the same three-tool server written with FastMCP 4.1.3 and with the official MCP Python SDK 2.3.0 answered calls equally fast (about 1 ms); FastMCP started in 643 to 735 ms against 461 to 516 ms for the SDK on repeats.
- Not tested: OAuth issuer settings and the CA trust-store fix this release is about; Remote (HTTP) transport, authentication and concurrent clients: only local stdio with one client was measured; Large payloads: calls used tiny inputs, so these latencies are protocol and process overhead, not throughput; Anything about answer quality: no model was involved in the latency numbers; These are single smoke-test runs; it is not a benchmark. The machine was shared with other jobs.

Sources:
- FastMCP v4.1.3 release: https://github.com/PrefectHQ/fastmcp/releases/tag/v4.1.3
- FastMCP v4.1.2 release: https://github.com/PrefectHQ/fastmcp/releases/tag/v4.1.2
- README: https://github.com/PrefectHQ/fastmcp
Cite as: Agentic Weekly, "FastMCP v4.1.3: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/fastmcp

## Gemini CLI v0.65.0 (2026-10-20, Reported (not run by us))
URL: https://theagenticweekly.com/tools/gemini-cli (markdown: https://theagenticweekly.com/tools/gemini-cli.md)
Vendor: Google

Gemini CLI v0.65.0 is a fix-heavy stable release (session resume, OAuth, workspace trust) with custom OTLP headers for telemetry. The bigger story: Google says Gemini CLI stopped serving consumer plans in June 2026 in favor of Antigravity CLI.

Verdict: Skip it as a new starter on a consumer plan; keep it only if you are on an enterprise license or a paid API key.
Good for: Organizations on Gemini Code Assist Standard or Enterprise (Google says their access is unchanged); People using Gemini CLI with a paid Gemini API key; Developers who want an open-source Apache-2.0 terminal agent they can read and extend
Skip if: You are on a Google AI Pro/Ultra plan or the free individual tier: Google says Gemini CLI stopped serving those requests on June 18, 2026; You want the best coding experience with the latest Gemini models: the README itself points to Antigravity CLI

What's new:
- Custom OTLP headers are now supported in telemetry configuration.
- Workspace settings are enforced as read-only in untrusted folders.
- Fixes for session resume: duplicate tool-response turns and history deletion on a quick exit.
- OAuth fixes: callback iss validation aligned with RFC 9207 and fixes for infinite verification and retry loops.
- Fix for a hang on the Enter key in interactive mode, plus several terminal UI fixes.
- Performance: ignore filtering optimized with subtree pruning.

Tutorial: Try Gemini CLI (paid API key or enterprise) and know your alternative
1. Check which plan you are on: Per Google's May 19, 2026 announcement, Gemini CLI stopped serving requests for Google AI Pro and Ultra and for free individual use on June 18, 2026. Enterprise licenses and paid API keys keep access.
2. Run it without installing: If you qualify, you can try Gemini CLI instantly with npx.
3. Or install globally: The README lists npm and Homebrew among other options.
4. If you are a consumer user, look at Antigravity CLI: Google says Antigravity CLI is available to everyone and keeps Agent Skills, Hooks, Subagents and Extensions (as Antigravity plugins), though not with 1:1 feature parity. See the sources below.

Not verified:
- What Antigravity CLI is like in practice (not tested; only Google's own description was read)
- Whether your specific account tier can still use Gemini CLI (check your plan with Google)

Sources:
- GitHub release v0.65.0: https://github.com/google-gemini/gemini-cli/releases/tag/v0.65.0
- Gemini CLI README: https://github.com/google-gemini/gemini-cli
- Google Developers Blog: Transitioning Gemini CLI to Antigravity CLI (May 19, 2026): https://developers.googleblog.com/an-important-update-transitioning-gemini-cli-to-antigravity-cli
Cite as: Agentic Weekly, "Gemini CLI v0.65.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/gemini-cli

## Jan 0.8.7 (2026-10-20, Reported (not run by us))
URL: https://theagenticweekly.com/tools/jan (markdown: https://theagenticweekly.com/tools/jan.md)
Vendor: Menlo Research

Jan's site describes an open-source ChatGPT alternative that runs open models locally or connects to online models and remembers your context. The repository's license file is Apache-2.0 and release 0.8.7 was tagged Oct 20, 2026.

Verdict: Open-source alternative to ChatGPT that runs open models locally or connects to cloud models like GPT and Claude.
Good for: People who want a private local chat app; Users who want one app for local and cloud models
Skip if: You need a hosted multi-user team product

What's new:
- 0.8.7 (2026-10-20): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Jan
1. Get access: Download Jan from jan.ai and choose an open model or connect a cloud provider.
2. Check the limits and costs first: Free; cloud model use is paid to the provider you connect. No pricing page read.

Not verified:
- Performance by model

Sources:
- Jan: https://jan.ai/
- Jan repository (GitHub): https://github.com/janhq/jan
Cite as: Agentic Weekly, "Jan 0.8.7: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/jan

## n8n 2.45 (2026-10-20, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/n8n (markdown: https://theagenticweekly.com/tools/n8n.md)
Vendor: n8n

n8n 2.45 (Oct 20) brings the Monday.com node V2 to full parity, a Kafka v2 default, an AI log that shows which model answered, and agent-builder improvements, among 11 features.

Verdict: Great fit for small teams that want to wire apps and AI together without code; the license is fair-code, not open source.
Good for: Small businesses automating lead capture and follow-up; Teams that want to self-host automations; Builders adding AI agents to existing workflows
Skip if: You need a strictly open-source license (n8n is fair-code); You want a no-setup consumer app

What's new:
- n8n 2.45 (Oct 20): Monday.com node V2 reaches full parity with V1 (39 operations) and is the default for new nodes.
- New Kafka and Kafka Trigger nodes default to version 2, which runs on librdkafka and supports Snappy, LZ4 and Zstd.
- The AI log now shows which model actually answered, useful when a model router picks a different model.
- Agent builder: monthly budget usage updates after each turn, a publish confirmation modal, and a warning when an agent uses unpublished workflows as tools.
- Admins can notify workflow owners from migration reports.

Tutorial: Try n8n
1. Pick cloud or self-host: n8n offers both. The README and docs cover cloud sign-up and self-hosting.
2. Build a first workflow: Start with a trigger such as a form submission, add an action node such as email or a spreadsheet, and run it manually before activating.
3. Add an AI agent step: The 2.45 notes describe an agent builder with budget tracking and sub-agents. Set a budget before letting an agent run on a schedule.

Not verified:
- Pricing and cloud plan limits
- Exact license terms for your use

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: n8n 2.44.6 installed from npm on Node 24 in 1 min 54 s (2,343 packages), imported a three-node workflow from the CLI and executed it: a Code node returned 42 and a request to a local model returned "42".
- Not tested: The editor UI, webhooks, credentials, scheduled triggers, community nodes, Python Code nodes and queue mode; Docker deployment, which n8n now recommends.

Sources:
- n8n release notes: https://docs.n8n.io/release-notes/
- n8n GitHub: https://github.com/n8n-io/n8n
Cite as: Agentic Weekly, "n8n 2.45: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/n8n

## ElevenLabs (2026-10-19, Reported (not run by us))
URL: https://theagenticweekly.com/tools/elevenlabs (markdown: https://theagenticweekly.com/tools/elevenlabs.md)
Vendor: ElevenLabs

ElevenLabs' Oct 19 changelog adds Eleven Realtime models for voice agents, a Realtime Translation WebSocket API for enterprise workspaces, and a Video to Music API that scores your video.

Verdict: Best for builders adding voice or music to products; the Video to Music API is the interesting new piece for creators.
Good for: Developers building voice agents or translation features; Creators who want an API that scores videos; Teams moving from OpenAI Realtime clients
Skip if: You want a simple consumer music app (see Suno); You need Realtime Translation outside enterprise workspaces

What's new:
- Eleven Realtime: a model family for ElevenAgents voice agents, with v1 Mini (default, fastest and cheapest) and v1 Max (reasons before answering). Existing OpenAI Realtime clients can connect by changing the endpoint and session configuration.
- Realtime Translation: speech in one language to speech in another while the speaker is still talking, via WebSocket, for enterprise workspaces.
- Video to Music: upload up to 10 videos (200 MB combined) and get a soundtrack; options include using the video's audio as reference, adding sound effects and returning separate tracks in a ZIP.
- ElevenAgents: alerts for error-rate thresholds, guardrails out of alpha, and new LLM options for agents.

Tutorial: Score a video with the API
1. Read the endpoint docs: The Video to Music API reference lists request parameters and the response format.
2. Send your videos: Upload up to 10 video files (200 MB combined); they are joined in order and scored as one video. Add a text description and up to 10 style tags to steer the result.
3. Choose options: use_audio_as_reference takes genre, BPM, instruments and mood from the video's audio; add_sfx adds sound effects; separate_tracks returns music and effects as separate files.

Not verified:
- Pricing and limits
- Audio quality (not tested)
- Availability of Eleven Realtime in your workspace

Sources:
- ElevenLabs changelog (Oct 19, 2026): https://elevenlabs.io/docs/changelog
Cite as: Agentic Weekly, "ElevenLabs: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/elevenlabs

## v0 (2026-10-19, Reported (not run by us))
URL: https://theagenticweekly.com/tools/v0 (markdown: https://theagenticweekly.com/tools/v0.md)
Vendor: Vercel

v0's October changelog adds a Slack MCP integration (Oct 19), download support in the Code tab plus iOS pickers (Oct 16), and turns personal accounts into team workspaces (Oct 6).

Verdict: Good for UI-heavy web work, especially if you deploy on Vercel; the October changes are about teamwork and polish.
Good for: Front-end builders who want generated UI and a code view; Teams that want v0 connected to Slack; Vercel users who want a short path to deployment
Skip if: You want a general-purpose coding agent for non-web projects; You need published pricing details (not verified here)

What's new:
- Oct 19: Connect Slack to v0 through MCP so it can find relevant messages and post updates when you ask. Available on all v0 plans.
- Oct 16: Large files that cannot be previewed can be downloaded from the Code tab; the iOS app adds a searchable skills picker, a design system picker and video attachments.
- Oct 16: Fixes include Gemini 4 requests failing at higher effort levels and several Design Mode issues.
- Oct 6: Personal v0 accounts became team workspaces; chats, projects and settings moved automatically.

Tutorial: Connect Slack to v0
1. Open a chat: Go to v0.app and open or start a chat.
2. Add the Slack MCP: In the chat composer open + then MCPs, choose Slack, and connect your workspace.
3. Ask v0 to use it: Ask v0 to use a Slack discussion as context for a change, or to share an update with your team.

Not verified:
- Current pricing and plan limits
- Output quality compared with other builders (not tested)

Sources:
- v0 changelog: https://v0.app/changelog
- v0 docs: https://v0.app/docs
Cite as: Agentic Weekly, "v0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/v0

## ComfyUI v0.42.1 (2026-10-18, Tested (smoke test, 2026-10-26, no LLM involved))
URL: https://theagenticweekly.com/tools/comfyui (markdown: https://theagenticweekly.com/tools/comfyui.md)
Vendor: Comfy Org

ComfyUI v0.42.0 adds a Krea 3 Image partner node, MOSS-image support and Qwen performance work; v0.42.1 the next day fixes crashes in the MOSS image template workflow.

Verdict: The tool to learn if you want maximum control over open models on your own GPU; it has a learning curve.
Good for: Power users who want to run open models locally; Artists building repeatable image or video workflows; People who want to mix local models with partner nodes
Skip if: You want a one-click consumer app; You do not have a capable GPU and do not want to use partner (paid) nodes

What's new:
- v0.42.0 (Oct 18): Partner node for Krea 3 Image; node labels now say FLUX 3 Image.
- v0.42.0: Support for MOSS-image and performance improvements for Qwen 3.x.
- v0.42.0: Workflow templates updated several times, and retired Quiver Arrow 1.x models removed.
- v0.42.1 (Oct 19): Fixes crashes some users hit running the MOSS image template workflow.

Tutorial: Install ComfyUI with comfy-cli
1. Install the CLI: The README documents comfy-cli.
2. Install ComfyUI: Run the install command and follow the prompts.
3. Start from a template: Use the built-in workflow templates to generate your first image before you build graphs from scratch.

Not verified:
- Hardware requirements in practice (README claims only)
- Partner node pricing
- Output quality (not tested)

Test scope:
- smoke test, 2026-10-26; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: ComfyUI 0.42.0 installed in 17.42 s and its CPU-only server was ready in 30.1 s with 1014 node types, but we did not generate an image.
- Not tested: Generating any image: no checkpoint was downloaded (the SD-Turbo file we considered is 5.2 GB, and the machine's free memory could not hold it); GPU use, custom nodes, the ComfyUI Manager, workflows from the web interface, and the browser UI; Speed of image generation on CPU or on a GPU.

Sources:
- ComfyUI v0.42.0 release: https://github.com/Comfy-Org/ComfyUI/releases/tag/v0.42.0
- ComfyUI v0.42.1 release: https://github.com/Comfy-Org/ComfyUI/releases/tag/v0.42.1
- README: https://github.com/Comfy-Org/ComfyUI
Cite as: Agentic Weekly, "ComfyUI v0.42.1: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/comfyui

## Cline SDK v0.0.95 (2026-10-17, Reported (not run by us))
URL: https://theagenticweekly.com/tools/cline (markdown: https://theagenticweekly.com/tools/cline.md)
Vendor: Cline

Cline shipped a batch of releases on Oct 16-17: CLI 3.0.73, desktop 0.0.48, SDK 0.0.95 and extension 4.1.25. The SDK notes cover GPT-6 request fixes and a refreshed model catalog.

Verdict: A good open-source pick if you want an agent inside your IDE or a CLI and an SDK to embed it; the October notes are incremental.
Good for: IDE users who want an open-source agent; Teams that want to embed an agent through the SDK; People who bring their own model API keys
Skip if: You want one polished all-in-one app with bundled models; You only want headline features; these releases are mostly fixes and plumbing

What's new:
- SDK v0.0.95 (Oct 17): OpenAI-compatible requests for GPT-6 models now send max_completion_tokens instead of max_tokens so OpenAI and Azure no longer reject them.
- SDK v0.0.95: The model catalog was refreshed from 6,952 to 7,041 models and adds the CommandCode and Reflection providers.
- SDK v0.0.95: The hub daemon records which client started it and publishes that as OpenTelemetry resource attributes.
- The same day Cline published CLI v3.0.73 and desktop v0.0.48; the IDE extension was at v4.1.25 on Oct 16.

Tutorial: Install Cline
1. Install the CLI: The README gives an npm global install. Node.js is required.
2. Or install the SDK to embed it: The README also documents an SDK package for building your own tools on top of Cline.
3. Connect a model: Open Cline and choose a model provider, then paste your own API key. Provider options change often, so check the in-app list.

Not verified:
- Feature-level changes in extension 4.1.25 (not read)
- Any hosted Cline pricing
- Anything beyond the attempt in /tests/cline-cli: we ran it once on a local 3B model and it did not complete our task, so it stays Reported

Sources:
- Cline SDK v0.0.95 release: https://github.com/cline/cline/releases/tag/sdk/sdk/v0.0.95
- Cline README: https://github.com/cline/cline
Cite as: Agentic Weekly, "Cline SDK v0.0.95: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/cline

## Cursor (2026-10-16, Reported (not run by us))
URL: https://theagenticweekly.com/tools/cursor (markdown: https://theagenticweekly.com/tools/cursor.md)
Vendor: Anysphere

Cursor's October changelog adds canvases built by cloud agents, a beta Environments API, much larger cloud environments, and a public app marketplace for Origin, its Git hosting product.

Verdict: Worth a look if you run agents on several repos or want one place for editor, cloud agents and Git hosting; the October changes are mostly for teams.
Good for: Teams that want cloud agents with shared, scriptable environments; Developers who like an editor-first workflow with an agent built in; Platform teams evaluating Origin and its app marketplace
Skip if: You work only in a terminal and want a lightweight open-source agent; You want all third-party model usage included in a flat fee; Cursor says it is billed separately by usage

What's new:
- Oct 16: Cloud agents can build interactive canvases as part of their work, and you can share a finished canvas with your team by link.
- Oct 16: The Environments API is in beta, so you can create and manage the cloud environments your agents run in from code.
- Oct 16: Cloud agent environments now support up to 300 repositories, and large environments load 5x faster.
- Oct 13: Origin now integrates with over 20 apps for deployments, CI, code review, security, monitoring and project management (including Cloudflare, Linear, Sentry and Socket Security), and the marketplace is open to public submissions.

Tutorial: Get started with Cursor
1. Download the editor: Cursor's download page offers the app for your operating system. Install it like any desktop app and sign in.
2. Open a project and start an agent chat: Open a folder or repository in Cursor and describe the change you want in the Agent panel. Review each proposed edit before accepting it.
3. Try a cloud agent (optional): Cursor's docs have a Cloud Agents section with setup, environments and best practices. Cloud agents run remotely, so read the setup page before connecting a repository.

Not verified:
- Whether the 4x, 15x and 50x usage multipliers fit your workload (vendor figures, not tested)
- Whether the cloud agent features perform well in practice (reported from the changelog, not tested)
- Exact install steps beyond the download page

Sources:
- Cursor changelog: https://cursor.com/changelog
- Cursor download page: https://cursor.com/download
- Cursor docs: https://cursor.com/docs
- Cursor pricing: https://cursor.com/pricing
Cite as: Agentic Weekly, "Cursor: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/cursor

## Browser Use 0.13.13 (2026-10-15, Reported (not run by us))
URL: https://theagenticweekly.com/tools/browser-use (markdown: https://theagenticweekly.com/tools/browser-use.md)
Vendor: Browser Use

Browser Use's site describes a cloud browser for AI agents with browsers priced at $0.02 an hour and a free credit to start; there is no subscription and credits never expire. The open-source repository is MIT-licensed with release 0.13.13 on Oct 15, 2026.

Verdict: Open-source library that lets AI agents drive a browser, with a pay-as-you-go cloud browser starting at $0.02 an hour.
Good for: Developers building browser-driving agents
Skip if: You need a simple scraper (consider Firecrawl)

What's new:
- 0.13.13 (2026-10-15): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Browser Use
1. Get access: pip install browser-use and run the quickstart, or create a cloud key.
2. Check the limits and costs first: Free starting credit; no subscription. Paid: Cloud browsers from $0.02 an hour; proxy data about $5 per GB; model tokens billed per model. Read from browser-use.com on 2026-10-25.

Not verified:
- Task success rates (not tested)

Sources:
- Browser Use: https://browser-use.com/
- Browser Use repository (GitHub): https://github.com/browser-use/browser-use
Cite as: Agentic Weekly, "Browser Use 0.13.13: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/browser-use

## GitHub Copilot (2026-10-15, Reported (not run by us))
URL: https://theagenticweekly.com/tools/github-copilot (markdown: https://theagenticweekly.com/tools/github-copilot.md)
Vendor: GitHub (Microsoft)

GitHub's October Copilot changelog includes general availability of Gemini 4 Argon, agentic autofix for code scanning alerts reaching GA, and dynamic workflows in Copilot CLI and the Copilot app.

Verdict: A sensible default if your team already lives on GitHub; check which plan you need for new models.
Good for: Teams already on GitHub who want AI in editor, CLI and code scanning fixes; Enterprise admins who need model policy controls; Developers who want to pick from several vendors' models
Skip if: You want a fully open-source stack; You are on a plan without access to the newest models (Gemini 4 Argon is Pro+ and above)

What's new:
- Oct 15: Agentic autofix for code scanning alerts is generally available.
- Oct 13: Gemini 4 Argon is generally available in Copilot for Pro+, Max, Business and Enterprise users, selectable in VS Code, Visual Studio, Copilot CLI, the coding agent and the Copilot app. It is billed at provider list pricing under usage-based billing.
- Oct 1: Dynamic workflows arrive in Copilot CLI and the Copilot app.
- Oct 15: GitHub announced upcoming deprecation of selected Copilot models in mid-November (and deprecated others on Oct 19).

Tutorial: Start using GitHub Copilot
1. Use it on GitHub.com: GitHub says Copilot is built into GitHub, so you can use it there without installing anything.
2. Install it in your editor: The docs have an install page for the Copilot extension in supported IDEs such as VS Code. Sign in with your GitHub account.
3. Try the CLI: The docs also cover Copilot CLI. Follow the install section of the quickstart for your platform.

Not verified:
- Pricing and plan limits
- Which model deprecations apply to your plan
- Real-world quality (not tested)

Sources:
- GitHub Copilot changelog: https://github.blog/changelog/label/copilot/
- Gemini 4 Argon in GitHub Copilot (Oct 13): https://github.blog/changelog/2026-10-13-gemini-4-argon-in-github-copilot
- Agentic autofix GA (Oct 15): https://github.blog/changelog/2026-10-15-agentic-autofix-for-code-scanning-alerts-general-availability
- Copilot quickstart: https://docs.github.com/en/copilot/get-started/quickstart
Cite as: Agentic Weekly, "GitHub Copilot: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/github-copilot

## Kling AI 4.0 (2026-10-15, Reported (not run by us))
URL: https://theagenticweekly.com/tools/kling (markdown: https://theagenticweekly.com/tools/kling.md)
Vendor: Kuaishou (Kling AI)

Kling 4.0 officially launched on Oct 15, 2026, according to Kuaishou's Oct 16 press release: steadier large-scale motion, native 30-second clips, up to 15 reference assets and 10 keyframes, and 10-bit HDR. On Oct 23 Luma added it inside Luma. Pricing figures here come from a third-party guide.

Verdict: Worth testing for action and camera-heavy shots; our information is secondhand from Luma plus the Kling homepage.
Good for: Creators who need strong motion and camera moves; People who want to try Kling 4.0 through Luma; Short social and ad clips
Skip if: You need API access today; Kuaishou says it is coming soon; You want an independent benchmark; none is cited here

What's new:
- Oct 15 (announced Oct 16): Kling 4.0 launched. Kuaishou says it handles intense motion and complex camera work more steadily, generates up to 30 seconds in one pass, accepts up to 15 reference assets (10 images, 5 videos, 7 subjects) and up to 10 keyframes, and outputs up to 10-bit HDR.
- The Kling homepage promotes 'All-New Kling 4.0'.
- Oct 23: Luma announced Kling 4.0 is available to every Luma customer.
- Luma describes Kling 4.0 as built for fast action, directed camera work (tracking, orbits, push-ins), more expressive faces and on-screen text that stays put, with text-to-video and image-to-video in 16:9, 9:16, 1:1 and 21:9.
- Kuaishou says 4K, 1080p and 21:9 output, longer video extension and API access for enterprises and developers are coming soon, and that Kling 4.0 is integrated with Kling MCP and CLI.

Tutorial: Try Kling 4.0
1. Use Kling directly: The Kling homepage offers a Create Now button. We did not test sign-up or credits.
2. Or use it inside Luma: Per Luma, open a board, choose Kling 4.0 for the shot that needs movement, and describe the shot.
3. Judge the take: Check whether the subject holds shape through the move and refine your prompt rather than regenerating blindly.

Not verified:
- Pricing and credits (third-party figures; Kling's membership page did not render for our fetch)
- Video quality and limits (not tested)
- Which Kling 4.0 features are live today versus marked coming soon

Sources:
- Kling AI homepage: https://klingai.com/
- Kling 4.0 in Luma (Oct 23, 2026): https://lumalabs.ai/news/kling-4-in-luma-ai-video
- Kuaishou press release: Kling 4.0 launch (Oct 16, 2026): https://www.globenewswire.com/news-release/2026/10/16/3382768/0/en/kling-ai-officially-launches-next-generation-kling-4-0-raising-the-bar-for-professional-grade-video-models.html
- Kling AI pricing (third-party guide): https://voyager.so/blog/kling-ai-pricing
- Kling membership page: https://klingai.com/membership/membership-plan
Cite as: Agentic Weekly, "Kling AI 4.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/kling

## OpenAI Agents SDK v0.24.0 (2026-10-15, Tested (smoke test, 2026-10-25, local Qwen2.5-3B-Instruct, not vendor models))
URL: https://theagenticweekly.com/tools/openai-agents-sdk (markdown: https://theagenticweekly.com/tools/openai-agents-sdk.md)
Vendor: OpenAI

openai-agents-python v0.24.0 exposes MCP annotations on converted tools and adds run-wide tool exception formatting, alongside session-write and compaction fixes.

Verdict: Pick it if you want a small, official framework for multi-agent flows on the OpenAI API; the update is incremental.
Good for: Python developers building agents on OpenAI models; Teams that want MCP tools plus approvals and sessions; Prototypers who want few abstractions
Skip if: You need a TypeScript-first framework; You want a framework tied to no model vendor

What's new:
- v0.24.0 (Oct 15): Expose MCP annotations on converted tools.
- v0.24.0: Support run-wide tool exception formatting.
- v0.24.0: Batch OpenAI conversation session writes within the API limit, and defer automatic compaction for pending approvals.
- v0.24.0: Many dependency bumps, including temporalio and vercel-sandbox.

Tutorial: Install the SDK and run a first agent
1. Create a virtual environment and install: The README shows pip, or uv if you prefer.
2. Set your API key: The SDK calls the OpenAI API, so export your key as an environment variable. Keep it out of source control.
3. Copy the hello-world agent from the README: The README has a short example that creates an agent and runs it. Start there before adding tools or handoffs.

Not verified:
- Support for non-OpenAI models

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (4-bit), a small local open-weight model served by Ollama 0.40.11; no vendor-hosted model was used; environment: Linux x86_64 box, 8 CPUs, no GPU; no vendor account, login or API key
- Ran: Installed openai-agents 0.24.0 and ran an Agent with one function tool through Runner.run using the Chat Completions model class against a local model; it produced a tool call, a tool output and a correct final message (9 words). Extended: OpenAI Agents SDK for JS (0.21.0) passed 2 of 5 (Node 20) and 1 of 5 (Node 24) at default sampling but 5 of 5 on both Node versions once temperature was pinned to 0. OpenAI Agents SDK (Python) 0.24.0 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.7 s.
- Not tested: OpenAI-hosted models, the default Responses API, tracing, MCP, approvals, sessions; Any vendor-hosted model: everything ran against one local 3B model; Multi-agent flows, memory, streaming and deployment; Anything beyond one tool and one question; five trials is a smoke test, not a benchmark; Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- openai-agents-python v0.24.0 release: https://github.com/openai/openai-agents-python/releases/tag/v0.24.0
- README: https://github.com/openai/openai-agents-python
Cite as: Agentic Weekly, "OpenAI Agents SDK v0.24.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/openai-agents-sdk

## Runway (2026-10-15, Reported (not run by us))
URL: https://theagenticweekly.com/tools/runway (markdown: https://theagenticweekly.com/tools/runway.md)
Vendor: Runway

Runway's Oct 15 changelog adds Mirelo SFX 2.0 and BiRefNet v2 background removal to Runway MCP, following Fish Audio Drama 3, Seedream 5.0 Flash and Seedance 2.5 Draft Mode in the previous two weeks.

Verdict: A strong choice if you want one place to drive many video, image and audio models from an agent; many items are paid-plan only.
Good for: Editors who want Runway inside DaVinci Resolve; Teams that script video work through MCP; Creators who want draft-then-polish video generation
Skip if: You need a free tier (many items are Paid Plans); You want a single in-house model; Runway aggregates partners

What's new:
- Oct 15: Mirelo SFX 2.0 in Runway MCP: give your agent a clip and describe the sound to get a separate audio track.
- Oct 15: BiRefNet v2 background removal in Runway MCP.
- Oct 14 and 12: Fish Audio Drama 3 voiceovers and Seedream 5.0 Flash image generation arrive in Runway MCP.
- Oct 2: Seedance 2.5 Draft Mode in Agent makes fast 480p drafts with post-generation enhancement.
- Oct 9: Enterprise admins can set credit caps per project.
- Sep 23: Runway is available inside DaVinci Resolve and on Cursor's Marketplace via Runway MCP.

Tutorial: Try Runway through MCP
1. Check your plan: The changelog marks most recent additions as Paid Plans, a few as All Plans. Confirm your plan before testing.
2. Add Runway MCP to your agent tool: Runway MCP is on Cursor's Marketplace. Other agents need their own MCP setup; follow Runway's current docs.
3. Give it a clip and a request: For example, ask the agent to add footsteps and ambience to a clip with Mirelo SFX 2.0 and return a separate audio track.

Not verified:
- Pricing and credit amounts
- Output quality (not tested)
- Exact steps for MCP setup outside Cursor

Sources:
- Runway changelog: https://runwayml.com/changelog
Cite as: Agentic Weekly, "Runway: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/runway

## Lovable (2026-10-14, Reported (not run by us))
URL: https://theagenticweekly.com/tools/lovable (markdown: https://theagenticweekly.com/tools/lovable.md)
Vendor: Lovable

Lovable's changelog adds Growth tools, which turn a goal like 'my first 100 signups' into a plan built from your app and traffic, plus new Discord and Google Business Profile connectors for the apps you build.

Verdict: A friendly way for non-developers to get a working web app and a first-customers plan; check credit costs before you commit.
Good for: Founders and small-business owners who want a site or app without writing code; Anyone who wants growth suggestions in the same chat as the build; People who want to start free and publish to a lovable.app address
Skip if: You need full control of the code and hosting from day one; You want predictable pricing; credits are consumed by task complexity

What's new:
- Growth tools: set a goal such as 'my first 100 signups' and Lovable turns it into a plan from your app, audience and traffic, and says when ads are premature. Results show in a redesigned Analytics view with visits by channel and campaign results.
- Lovable adds no extra fees for Growth tools; the standard costs of chatting with Lovable apply.
- New connectors for Discord (post messages, read channel history, respond to slash commands) and Google Business Profile.
- The quick start loop is describe, refine, publish; apps publish to a lovable.app URL, free on every plan according to the docs.

Tutorial: Build and publish a simple app with Lovable
1. Sign up: The quick start says you can sign up free with email, Google, GitHub or Apple. Your account comes with a personal workspace on the free plan.
2. Describe your app and refine one change at a time: Describe what you want in plain words, then ask for one change at a time and check the preview after each. For small visual changes, select the element in the preview toolbar and say what should change.
3. Test, then publish: Click every button and link, check mobile view, then click Publish in the top right. Lovable suggests a lovable.app address you can edit before deploying.

Not verified:
- Paid plan prices and credit tiers (from a search summary of Lovable's docs; not read from the pricing page)
- Whether Growth plans produce results (not tested)

Sources:
- Lovable changelog: https://docs.lovable.dev/changelog
- Lovable quick start: https://docs.lovable.dev/introduction/getting-started
- Lovable pricing: https://lovable.dev/pricing
- Lovable subscription plans: https://docs.lovable.dev/introduction/subscription-plans
Cite as: Agentic Weekly, "Lovable: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/lovable

## Supabase MCP (2026-10-14, Tested (smoke test, 2026-10-26, no LLM involved))
URL: https://theagenticweekly.com/tools/supabase-mcp (markdown: https://theagenticweekly.com/tools/supabase-mcp.md)
Vendor: Supabase

Supabase's MCP repository connects Supabase to AI assistants. It is Apache-2.0 licensed; a mcp-utils v0.9.0 release was tagged Oct 14, 2026.

Verdict: Connect Supabase projects to AI assistants.
Good for: Agents managing Supabase projects and databases
Skip if: You use a different backend

What's new:
- Latest release we saw (2026-10-14): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Supabase MCP
1. Get access: Follow the repository README to configure it with your MCP client.
2. Check the limits and costs first: Free. Server is free; Supabase plan limits apply.

Not verified:
- Behavior in practice (we have not run this tool)

Test scope:
- smoke test, 2026-10-26; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Supabase MCP server 0.14.0 (@supabase/mcp-server-supabase) started over stdio in a median 468 ms and listed 19 tools in a median 363 ms; we did not make any authenticated call.
- Not tested: Any authenticated call: we used a placeholder token, so no request reached the vendor's API and we did not check that real accounts, permissions or rate limits work; Remote (HTTP) transport and concurrent clients: only local stdio with one client was run; Anything about answer quality: no model was involved in the handshake and tool-list check; A single smoke-test run set (three starts); it is not a benchmark.

Sources:
- Supabase MCP repository (GitHub): https://github.com/supabase-community/supabase-mcp
Cite as: Agentic Weekly, "Supabase MCP: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/supabase-mcp

## DSPy 3.4.1 (2026-10-12, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/dspy (markdown: https://theagenticweekly.com/tools/dspy.md)
Vendor: Stanford NLP

DSPy is a Python framework from Stanford NLP for composing language-model programs as modules with typed signatures. The repository is MIT-licensed; 3.4.1 was tagged on 2026-10-12.

Verdict: Worth a look if you want structured, testable LM pipelines; start with a plain Predict module before optimisers.
Good for: Developers who want typed, composable LM programs; Teams that want to optimise prompts against a metric
Skip if: You only need a quick one-off script calling a model

What's new:
- 3.4.1 (2026-10-12): the newest release on the repository when we read it on 2026-10-25.

Tutorial: Get started with DSPy
1. Install: One package; we measured the install on a cold cache.
2. Turn off caching while testing: DSPy caches LM calls by default, which made our first timing look much faster than a real call.

Not verified:
- DSPy optimisers (MIPRO, GEPA and similar): not run
- Behaviour with hosted models

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: DSPy 3.4.1 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 3.7 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark; Optimizers (MIPRO, GEPA, BootstrapFewShot) and evaluation, which are the reason to use DSPy.

Sources:
- DSPy documentation: https://dspy.ai
- DSPy repository (GitHub): https://github.com/stanfordnlp/dspy
Cite as: Agentic Weekly, "DSPy 3.4.1: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/dspy

## Midjourney (2026-10-08, Reported (not run by us))
URL: https://theagenticweekly.com/tools/midjourney (markdown: https://theagenticweekly.com/tools/midjourney.md)
Vendor: Midjourney

Midjourney's alpha changelog titled 10/7/26 (published Oct 8) adds a 'thinking' rerun for images made with version 8.2 and shared folders with collaborator and viewer roles.

Verdict: Use the thinking rerun if prompt-following misses are your pain; be careful with privacy when sharing folders.
Good for: Designers who iterate on images with a team; Users of 8.2 who want better prompt adherence; People who want a web app rather than Discord
Skip if: You need private-by-default collaboration; You want open weights or local control

What's new:
- Thinking mode: rerun jobs made with 8.2 (standard and edit) with 'thinking', where Midjourney finds where the image missed your prompt (objects, layout, anatomy, text) and generates again. Early version.
- Shared folders: invite people as Collaborators (can generate into the folder) or Viewers, or share by link.
- Warning: sharing a folder does not change image privacy; non-stealth images can still appear on Explore and profiles.
- Fixes for stalled edits, keyboard shortcuts under dialogs, and long history catch-up; pin folders in the sidebar.

Tutorial: Try the thinking rerun
1. Open the web app: The changelog refers to alpha.midjourney.com. Sign in with your Midjourney account.
2. Generate with 8.2: Make an image with version 8.2 (standard or edit mode).
3. Rerun with thinking: Open the job and choose to rerun with thinking so Midjourney can check and correct misses against your prompt.

Not verified:
- Plan prices and limits (from third-party guides; Midjourney's own docs returned 403 to our fetch)
- Image quality (not tested)
- Whether thinking improves results in practice

Sources:
- Alpha Changelog 10/7/26: https://updates.midjourney.com/alpha-changelog-10-7-26/
- Midjourney plans (official docs, not readable by our fetch): https://docs.midjourney.com/hc/en-us/articles/27870484040333-Comparing-Midjourney-Plans
- Midjourney pricing 2026 (third-party guide): https://www.renderahouse.com/blog/midjourney-pricing
Cite as: Agentic Weekly, "Midjourney: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/midjourney

## Semantic Kernel python-1.45.0 (2026-10-06, Reported (not run by us))
URL: https://theagenticweekly.com/tools/semantic-kernel (markdown: https://theagenticweekly.com/tools/semantic-kernel.md)
Vendor: Microsoft

The Semantic Kernel README says Microsoft Agent Framework (MAF) is the enterprise-ready successor and is at version 1.0 with stable APIs. The repository is still MIT-licensed and active, with python-1.45.0 released Oct 6, 2026.

Verdict: Microsoft's model-agnostic SDK for building AI agents; its README calls Microsoft Agent Framework the enterprise-ready successor.
Good for: Existing .NET and Azure shops already on Semantic Kernel
Skip if: You are starting new agent work and want Microsoft's recommended path (Agent Framework)

What's new:
- python-1.45.0 (2026-10-06): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Semantic Kernel
1. Get access: Install the Semantic Kernel package for your language and follow the README; check the Agent Framework page for new projects.
2. Check the limits and costs first: Free to self-host. Open-source library; costs come from the model APIs you call.

Not verified:
- Migration timeline
- Anything beyond the attempt in /tests/semantic-kernel: we ran it once on a local 3B model and it did not complete our task, so it stays Reported

Sources:
- Semantic Kernel README: https://github.com/microsoft/semantic-kernel
Cite as: Agentic Weekly, "Semantic Kernel python-1.45.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/semantic-kernel

## Sentry MCP cli 0.47.0 (2026-10-05, Tested (smoke test, 2026-10-26, no LLM involved))
URL: https://theagenticweekly.com/tools/sentry-mcp (markdown: https://theagenticweekly.com/tools/sentry-mcp.md)
Vendor: Sentry

Sentry's agent tooling repository (formerly sentry-mcp) is released under FSL-1.1-Apache-2.0, a source-available license that converts to Apache 2.0 later. A CLI 0.47.0 release was tagged Oct 5, 2026.

Verdict: Sentry's agentic tooling and MCP server for errors and traces.
Good for: Agents that debug production errors with Sentry data
Skip if: You need a permissive OSI license today

What's new:
- cli 0.47.0 (2026-10-05): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Sentry MCP
1. Get access: Follow the repository README to configure the MCP server with your Sentry account.
2. Check the limits and costs first: Free. Server is free; Sentry plan limits apply.

Not verified:
- Behavior in practice (we have not run this tool)

Test scope:
- smoke test, 2026-10-26; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Sentry MCP server 0.42.0 (@sentry/mcp-server) started over stdio in a median 1134 ms and listed 9 tools in a median 13 ms; we did not make any authenticated call.
- Not tested: Any authenticated call: we used a placeholder token, so no request reached the vendor's API and we did not check that real accounts, permissions or rate limits work; Remote (HTTP) transport and concurrent clients: only local stdio with one client was run; Anything about answer quality: no model was involved in the handshake and tool-list check; A single smoke-test run set (three starts); it is not a benchmark.

Sources:
- Sentry MCP repository (GitHub): https://github.com/getsentry/sentry-mcp
Cite as: Agentic Weekly, "Sentry MCP cli 0.47.0: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/sentry-mcp

## Playwright MCP v0.0.83 (2026-09-28, Tested (smoke test, 2026-10-25, no language model involved))
URL: https://theagenticweekly.com/tools/playwright-mcp (markdown: https://theagenticweekly.com/tools/playwright-mcp.md)
Vendor: Microsoft

Playwright MCP is Microsoft's MCP server for browser automation. It is Apache-2.0 licensed and v0.0.83 was released Sep 28, 2026.

Verdict: Microsoft's MCP server that lets agents drive a browser with Playwright.
Good for: Agents that need to test or navigate web pages
Skip if: You need DevTools-level performance traces (see Chrome DevTools MCP)

What's new:
- v0.0.83 (2026-09-28): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Playwright MCP
1. Get access: Add the server to your MCP client following the repository README.
2. Check the limits and costs first: Free. Open-source server; the service it connects to may have its own costs.

Test scope:
- smoke test, 2026-10-25; model: None; no language model was involved; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Playwright MCP 0.0.83 exposed 25 tools and navigated a headless Chrome to a local page in 40.5 ms median (cold start 383 ms).
- Not tested: Remote (HTTP) transport, authentication and concurrent clients: only local stdio with one client was measured; Large payloads: calls used tiny inputs, so these latencies are protocol and process overhead, not throughput; Anything about answer quality: no model was involved in the latency numbers; These are single smoke-test runs; it is not a benchmark. The machine was shared with other jobs; Clicking, typing, screenshots and snapshot-based element selection: only navigation was timed.

Sources:
- Playwright MCP repository (GitHub): https://github.com/microsoft/playwright-mcp
Cite as: Agentic Weekly, "Playwright MCP v0.0.83: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/playwright-mcp

## Mem0 Python SDK v2.2.1 (2026-09-25, Tested (smoke test, 2026-10-26, local 3B model plus a local embedder))
URL: https://theagenticweekly.com/tools/mem0 (markdown: https://theagenticweekly.com/tools/mem0.md)
Vendor: Mem0

Mem0's pricing page lists Free (10,000 add requests and 1,000 retrieval requests a month), $19 a month (50,000 add and 5,000 retrieval) and $249 a month (500,000 add and 50,000 retrieval), plus Enterprise. The repository is Apache-2.0 and the Python SDK v2.2.1 was tagged Sep 25, 2026.

Verdict: Open-source and hosted memory layer that lets agents remember users across sessions, with a free tier and published request quotas.
Good for: Apps that need long-term per-user memory
Skip if: You only need a vector database

What's new:
- Python SDK v2.2.1 (2026-09-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Mem0
1. Get access: Create an API key at mem0.ai or install the open-source package and call add and search.
2. Check the limits and costs first: Free: 10,000 add requests and 1,000 retrieval requests per month. Paid: $19 per month (50,000 add / 5,000 retrieval); $249 per month (500,000 add / 50,000 retrieval); Enterprise custom. Read from mem0.ai/pricing on 2026-10-25.

Not verified:
- Benchmark claims (vendor)

Test scope:
- smoke test, 2026-10-26; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11 for memory extraction; nomic-embed-text v1.5 (Q4_K_M, through Ollama) for embeddings; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Mem0's open-source SDK 2.2.1 stored a fact from one chat message with a local 3B model and retrieved it with a semantic search (2.04 s) on the second attempt; the first two runs failed on a missing package and an API change.
- Not tested: Mem0's hosted platform, pricing tiers and quotas: only the open-source SDK was run; Graph memory, multi-user isolation, updates and deletes, and the optional extras (spaCy, fastembed); Retrieval quality over more than one fact: a single message and a single query is a smoke test, not a benchmark; Timing: add() ran while another model job loaded the machine.

Sources:
- Mem0: https://mem0.ai/
- Mem0 pricing: https://mem0.ai/pricing
- Mem0 repository (GitHub): https://github.com/mem0ai/mem0
Cite as: Agentic Weekly, "Mem0 Python SDK v2.2.1: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/mem0

## LlamaIndex v0.14.25 (2026-09-21, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/llamaindex (markdown: https://theagenticweekly.com/tools/llamaindex.md)
Vendor: LlamaIndex

LlamaIndex's repository is MIT-licensed with release v0.14.25 tagged Sep 21, 2026 and recent pushes in October 2026. It focuses on connecting LLMs to your data: loaders, indexes, query engines and agent workflows.

Verdict: Open-source data framework for retrieval-augmented generation and document-centred agents, with a hosted document platform alongside.
Good for: RAG and document-question-answering projects; Teams with lots of PDFs and unstructured data
Skip if: You need a general multi-agent orchestrator rather than data tooling

What's new:
- v0.14.25 (2026-09-21): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with LlamaIndex
1. Get access: Install llama-index with pip and follow the starter tutorial in the docs.
2. Check the limits and costs first: Free to self-host. Open-source library is free; the hosted LlamaCloud product has its own pricing that we did not read.

Not verified:
- Hosted product pricing

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: LlamaIndex 0.14.25 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 3.4 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark; Indexing, retrieval, embeddings and the data connectors that LlamaIndex is mostly used for.

Sources:
- LlamaIndex: https://www.llamaindex.ai/
- LlamaIndex repository (GitHub): https://github.com/run-llama/llama_index
Cite as: Agentic Weekly, "LlamaIndex v0.14.25: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/llamaindex

## Notion MCP Server v2.5.2 (2026-09-20, Tested (smoke test, 2026-10-26, no LLM involved))
URL: https://theagenticweekly.com/tools/notion-mcp (markdown: https://theagenticweekly.com/tools/notion-mcp.md)
Vendor: Notion

Notion's official MCP server repository is MIT-licensed; v2.5.2 was released Sep 20, 2026.

Verdict: Official Notion MCP server for reading and editing workspace content.
Good for: Agents that need Notion pages and databases
Skip if: You do not use Notion

What's new:
- v2.5.2 (2026-09-20): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Notion MCP Server
1. Get access: Follow the README to create an integration token and connect your MCP client.
2. Check the limits and costs first: Free. Server is free; Notion plan and API limits apply.

Not verified:
- Behavior in practice (we have not run this tool)

Test scope:
- smoke test, 2026-10-26; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Notion MCP server 2.5.2 (@notionhq/notion-mcp-server) started over stdio in a median 316 ms and listed 24 tools in a median 16 ms; we did not make any authenticated call.
- Not tested: Any authenticated call: we used a placeholder token, so no request reached the vendor's API and we did not check that real accounts, permissions or rate limits work; Remote (HTTP) transport and concurrent clients: only local stdio with one client was run; Anything about answer quality: no model was involved in the handshake and tool-list check; A single smoke-test run set (three starts); it is not a benchmark.

Sources:
- Notion MCP Server repository (GitHub): https://github.com/makenotion/notion-mcp-server
Cite as: Agentic Weekly, "Notion MCP Server v2.5.2: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/notion-mcp

## Instructor v1.17.0 (2026-09-09, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/instructor (markdown: https://theagenticweekly.com/tools/instructor.md)
Vendor: 567 Labs

Instructor is a Python library that turns model responses into validated Pydantic objects. It is MIT-licensed; v1.17.0 was tagged on 2026-09-09.

Verdict: A small, light way to get typed output from a model; good first step before reaching for a full framework.
Good for: Developers who need typed, validated model output; Teams already using Pydantic
Skip if: You need an agent runtime with tools and memory

What's new:
- v1.17.0 (2026-09-09): the newest release on the repository when we read it on 2026-10-25.

Tutorial: Get started with Instructor
1. Install: A light install in our measurement.
2. Wrap your client: Patch an OpenAI-compatible client and pass a Pydantic response_model.

Not verified:
- Retries, validation failures and streaming: not run
- Behaviour with hosted models

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Instructor 1.17.0 returned a validated Pydantic object from a local 3B model in 3 of 3 trials (structured extraction, not a tool call); cold uv install 1.4 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- Instructor documentation: https://python.useinstructor.com/
- Instructor repository (GitHub): https://github.com/567-labs/instructor
Cite as: Agentic Weekly, "Instructor v1.17.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/instructor

## Suno v6 (2026-09-09, Reported (not run by us))
URL: https://theagenticweekly.com/tools/suno (markdown: https://theagenticweekly.com/tools/suno.md)
Vendor: Suno

Suno introduced v6 on Sep 9 as three models: v6 (Pro and Premier), v6-wild for exploration (Pro and Premier) and v6-mini for everyone, with plain-language editing of parts of a song.

Verdict: The most complete song-making tool we track; the v6 editing features are the draw. Rights and licensing questions are yours to check.
Good for: Anyone who wants finished songs from a prompt or a voice memo; Producers who want to edit sections in plain language; Free users who want to try v6-mini
Skip if: You need clearly licensed commercial music (check Suno's terms); You want open weights

What's new:
- v6 is the flagship model for Pro and Premier subscribers; v6-wild is less predictable and more varied; v6-mini is faster and available to everyone.
- Edit part of an existing song in plain language, such as changing one chorus while keeping the rest.
- Build mashups from multiple sources in one request, and sample, isolate and build a new beat in a single workflow.
- Create from text, audio, images and video, and change a single lyric without rebuilding the song.
- Suno says v6 was developed with industry partners including Warner Music Group, BMG and Believe.

Tutorial: Make a song with Suno v6
1. Open Suno: Go to suno.com and sign in. v6-mini is available to everyone; v6 and v6-wild need Pro or Premier.
2. Describe the song: Use a vibe, genre or reference, or upload audio, an image or a video as the starting point.
3. Edit instead of regenerating: Ask for a change to one section, such as the chorus, or one lyric line, rather than rebuilding the whole song.

Not verified:
- Pricing and commercial license terms
- Audio quality (not tested)
- Anything after Sep 9, 2026

Sources:
- Introducing v6 (Sep 9, 2026): https://suno.com/blog/introducing-v6
Cite as: Agentic Weekly, "Suno v6: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/suno

## Piper v1.8.0 (2026-09-04, Tested (smoke test, 2026-10-25, no LLM involved))
URL: https://theagenticweekly.com/tools/piper (markdown: https://theagenticweekly.com/tools/piper.md)
Vendor: Open Home Foundation

Piper is a local text-to-speech engine. The repository piper1-gpl is GPL-3.0 licensed; v1.8.0 was tagged on 2026-09-04.

Verdict: Fast, private speech synthesis on a CPU; mind the GPL-3.0 license if you embed it.
Good for: Local voice output without a GPU; Home and offline projects
Skip if: You need very natural or expressive voices

What's new:
- v1.8.0 (2026-09-04): the newest release on the repository when we read it on 2026-10-25.

Tutorial: Get started with Piper
1. Install: Package name on PyPI is piper-tts.
2. Download a voice and speak: Voices are separate downloads (about 60 MB for the one we used).

Not verified:
- Other voices and languages
- Perceived naturalness

Test scope:
- smoke test, 2026-10-25; model: none (no language model was involved); environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Piper 1.8.0 synthesised 10 sentences (about 41 s of 22.05 kHz audio) on CPU in 1.4 to 1.5 s each including process start; 3.5 s uv install, 167 MB.
- Not tested: Voice quality as judged by a person (we only measured machine round-trip accuracy), other voices and languages, streaming, and GPU use; One voice only; a smoke test.

Sources:
- Piper repository (GitHub): https://github.com/OHF-Voice/piper1-gpl
Cite as: Agentic Weekly, "Piper v1.8.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/piper

## Flowise flowise-ui 3.1.4 (2026-07-29, Reported (not run by us))
URL: https://theagenticweekly.com/tools/flowise (markdown: https://theagenticweekly.com/tools/flowise.md)
Vendor: FlowiseAI

The GitHub API reports FlowiseAI/Flowise as archived (read-only) with the last push on Aug 13, 2026. Its license file says portions under the enterprise directory use a commercial license and other code uses Apache 2.0.

Verdict: Open-source drag-and-drop builder for LLM flows and agents; GitHub marks the main repository as archived with its last push in Aug 2026.
Good for: Existing Flowise users maintaining flows
Skip if: You need an actively developed builder: the repository is archived

What's new:
- flowise-ui 3.1.4 (2026-07-29): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Flowise
1. Get access: Existing users can self-host the final release; for new projects compare Dify and Langflow.
2. Check the limits and costs first: Free to self-host. Repository archived; check the vendor site for any hosted product.

Not verified:
- Reasons for archiving and any successor
- Anything beyond the attempt in /tests/flowise: we ran it once on a local 3B model and it did not complete our task, so it stays Reported

Sources:
- Flowise: https://flowiseai.com/
- Flowise repository (GitHub): https://github.com/FlowiseAI/Flowise
Cite as: Agentic Weekly, "Flowise flowise-ui 3.1.4: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/flowise

## Continue v2.0.0-vscode (2026-06-19, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/continue (markdown: https://theagenticweekly.com/tools/continue.md)
Vendor: Continue

Continue is an Apache-2.0 open-source coding assistant with IDE extensions and a command-line interface. The latest GitHub release is v2.0.0-vscode (2026-06-19); we tested the CLI package @continuedev/cli 1.5.47.

Verdict: A configurable open-source assistant; the CLI worked with a local model through a YAML config.
Good for: Developers who want an open-source assistant with local-model support
Skip if: You want a zero-config experience

What's new:
- v2.0.0-vscode (2026-06-19): the newest GitHub release when we read it on 2026-10-25. We tested the separately versioned CLI, 1.5.47.

Tutorial: Get started with Continue
1. Install the CLI: We installed the CLI with npm.
2. Configure a local model: Point a YAML config at an OpenAI-compatible server; our test log has the file.

Not verified:
- IDE extensions, hosted models, agent mode on larger tasks

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Continue CLI 1.5.47 answered a question and fixed a one-line bug with a local 3B model through a YAML config; a 1.6 s, 81 MB install.
- Not tested: The IDE extensions, hub-hosted configs and any hosted model; Reliability: the edit task ran once, so this is an existence proof, not a pass rate.

Sources:
- Continue: https://continue.dev
- Continue repository (GitHub): https://github.com/continuedev/continue
Cite as: Agentic Weekly, "Continue v2.0.0-vscode: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/continue

## Windsurf (now Devin Desktop) (2026-06-02, Reported (not run by us))
URL: https://theagenticweekly.com/tools/windsurf (markdown: https://theagenticweekly.com/tools/windsurf.md)
Vendor: Cognition (Devin)

On June 2, 2026 Cognition renamed Windsurf as Devin Desktop, delivered as an over-the-air update. Per Cognition's FAQ it is the same IDE with the same features, now with an Agent Command Center (Spaces and a Kanban view for managing local and cloud agents) as the default view. windsurf.com now redirects to devin.ai/desktop.

Verdict: Windsurf was renamed Devin Desktop on June 2, 2026: a full IDE with an Agent Command Center for managing local and cloud coding agents.
Good for: Developers already using Windsurf who want the same editor under the Devin brand; People who want one place to manage local and cloud coding agents; Teams that want centralized billing and an admin dashboard (Teams plan)
Skip if: You want a fully open-source editor (it is proprietary; see Zed or Kilo); You need predictable flat pricing: extra usage is billed at API pricing

What's new:
- Jun 2, 2026: Windsurf renamed Devin Desktop; Agent Command Center becomes the default view.

Tutorial: Get started with Windsurf (now Devin Desktop)
1. Get access: Download Devin Desktop from devin.ai/desktop and sign in; existing Windsurf users receive it as an over-the-air update.
2. Check the limits and costs first: Free: $0, light agent quota, limited models, unlimited inline edits and Tab completions. Paid: Pro $20/month, Max $200/month, Teams $80/month plus $40/month per full dev seat, Enterprise custom. Plan prices read from devin.ai/pricing on 2026-10-25. The rename FAQ says existing Windsurf plans and pricing did not change. Extra usage is billed at API pricing.

Not verified:
- Quota sizes per plan (vendor gives none)
- Real-world agent quality (not tested)
- Whether Devin Cloud availability matches the plan page in your region

Sources:
- Windsurf is now Devin Desktop (Cognition blog, June 2, 2026): https://devin.ai/blog/windsurf-is-now-devin-desktop
- Devin Desktop FAQ: https://docs.devin.ai/desktop/devin-desktop-faq
- Devin plans and pricing: https://devin.ai/pricing
- Devin Desktop: https://devin.ai/desktop
Cite as: Agentic Weekly, "Windsurf (now Devin Desktop): what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/windsurf

## smolagents v1.26.0 (2026-05-29, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/smolagents (markdown: https://theagenticweekly.com/tools/smolagents.md)
Vendor: Hugging Face

The smolagents docs describe an open-source Python library for building and running agents in a few lines of code, with abstractions kept minimal. The repository is Apache-2.0 and the latest release we saw is v1.26.0 on May 29, 2026.

Verdict: Hugging Face's small Python library for agents that write and run code; the docs say the core logic fits in roughly a thousand lines.
Good for: People who want a small hackable agent library; Code-as-action agent experiments
Skip if: You need a full production orchestration platform

What's new:
- v1.26.0 (2026-05-29): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with smolagents
1. Get access: pip install smolagents and follow the docs' first agent example.
2. Check the limits and costs first: Free to self-host. Open-source library; costs come from the model APIs you call.

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: smolagents 1.26.0 passed our one-tool task on a local 3B model: 6 of 6 trials; cold uv install 1.5 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark; The CodeAgent sandboxes (Docker, E2B, Modal) and the hub tools.

Sources:
- smolagents docs: https://huggingface.co/docs/smolagents
- smolagents repository (GitHub): https://github.com/huggingface/smolagents
Cite as: Agentic Weekly, "smolagents v1.26.0: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/smolagents

## Letta v0.16.8 (2026-05-14, Reported (not run by us))
URL: https://theagenticweekly.com/tools/letta (markdown: https://theagenticweekly.com/tools/letta.md)
Vendor: Letta

Letta's pricing page lists a Free plan at $0, a $20 per month plan, and usage-based items such as $0.10 per active agent per month and $0.00015 per second of tool execution, plus a Teams plan at $20 per seat per month and custom Enterprise. The repository is Apache-2.0.

Verdict: Platform and open-source framework for stateful agents that keep and improve memory across sessions.
Good for: Agents that must remember across sessions
Skip if: You want a stateless simple chatbot

What's new:
- v0.16.8 (2026-05-14): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Letta
1. Get access: Sign up at letta.com or run the open-source server and create an agent from the docs.
2. Check the limits and costs first: Free: $0; includes limited agents and LLM requests with rotating free models. Paid: $20 per month plan; Teams $20 per seat per month; usage items such as $0.10 per active agent per month; Enterprise custom. Read from letta.com/pricing on 2026-10-25.

Not verified:
- Behavior in practice (we have not run this tool)

Sources:
- Letta: https://www.letta.com/
- Letta pricing: https://www.letta.com/pricing
- Letta repository (GitHub): https://github.com/letta-ai/letta
Cite as: Agentic Weekly, "Letta v0.16.8: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/letta

## Aider (2026-02-12, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/aider (markdown: https://theagenticweekly.com/tools/aider.md)
Vendor: Aider-AI (open-source project)

Aider's site says it lets you pair program with LLMs from the terminal, builds a map of your entire codebase to help on larger projects, and can connect to almost any LLM including local models. The GitHub repository lists the Apache-2.0 license. The newest release tag we saw is v0.86.3.dev (Feb 12, 2026).

Verdict: Open-source AI pair programming in your terminal that maps your codebase and connects to almost any LLM, including local models.
Good for: Developers who live in the terminal and want git-friendly AI edits; People who want to pick their own model or run local models
Skip if: You want a GUI editor with an integrated agent; You want a vendor-supported product with an SLA

What's new:
- Latest release we saw (2026-02-12): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with Aider
1. Get access: Follow the install instructions on aider.chat and connect an API key for your chosen model.
2. Check the limits and costs first: Free software; you pay your LLM provider. Paid: No paid plan listed on the pages we read. No pricing page. The tool is free; model usage costs depend on the provider you connect.

Not verified:
- Current maintenance pace

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: Aider 0.86.2 implemented a missing function with a local 3B model in 3 of 3 trials (27 to 32 s each); it did not install on Python 3.13 without a C++ compiler, and it left a stray placeholder comment in the file.
- Not tested: Any hosted model (aider is mostly used with them); Repo-map, git auto-commit, multi-file edits, voice and the browser UI; A single toy task; three trials is a smoke test, not a benchmark.

Sources:
- Aider: https://aider.chat/
- Aider repository (GitHub): https://github.com/Aider-AI/aider
- Aider releases: https://github.com/Aider-AI/aider/releases
Cite as: Agentic Weekly, "Aider: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/aider

## ACE-Step v1.5 (2026-01-28, Reported (not run by us))
URL: https://theagenticweekly.com/tools/ace-step (markdown: https://theagenticweekly.com/tools/ace-step.md)
Vendor: ACE-Step

ACE-Step v1.5 was released on Jan 28, 2026 per the README. The repository's last push we saw was in February 2026, so treat it as a stable open option rather than a fast-moving one.

Verdict: The pick if you want open weights and local control; expect a manual setup and no recent updates.
Good for: Tinkerers who want to run music generation locally; Developers who want to fine-tune with LoRA; People who need Apache-2.0 licensing
Skip if: You want a polished app with no setup; You need active weekly releases

What's new:
- 2026-01-28: README announces ACE-Step v1.5 as the latest and most advanced model.
- The model targets many styles and 19 languages, with ComfyUI nodes and LoRA training instructions listed in the README.
- The README notes a memory optimization that reduced maximum VRAM to 8GB for consumer devices.

Tutorial: Run ACE-Step locally
1. Clone the repo: The README starts by cloning the repository.
2. Create an environment and install: Use a virtual environment and an editable install. Install the PyTorch build that matches your GPU first, as the README describes.
3. Launch the UI: Start the Gradio UI. The README recommends extra flags on lower-VRAM GPUs.

Not verified:
- Audio quality compared with Suno (not tested)
- Whether newer versions exist outside the repo
- Hardware needs for v1.5 specifically

Sources:
- ACE-Step README: https://github.com/ace-step/ACE-Step
- ACE-Step repository metadata (GitHub API): https://api.github.com/repos/ace-step/ACE-Step
Cite as: Agentic Weekly, "ACE-Step v1.5: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/ace-step

## faster-whisper v1.2.1 (2025-10-31, Tested (smoke test, 2026-10-25, no language model involved))
URL: https://theagenticweekly.com/tools/faster-whisper (markdown: https://theagenticweekly.com/tools/faster-whisper.md)
Vendor: SYSTRAN

faster-whisper reimplements Whisper inference with CTranslate2 for faster CPU and GPU transcription. It is MIT-licensed; v1.2.1 was tagged on 2025-10-31.

Verdict: Fast local transcription; on CPU use float32 and check int8 against a reference before trusting it.
Good for: Local, private transcription; Developers building voice pipelines on CPU
Skip if: You need diarisation or a managed service

What's new:
- v1.2.1 (2025-10-31): the newest release on the repository when we read it on 2026-10-25.

Tutorial: Get started with faster-whisper
1. Install and pin av: In our run an unpinned av release broke first use; pinning fixed it.
2. Transcribe: Use float32 on CPU if you need repeatable output.

Not verified:
- GPU speed
- Non-English audio, noisy audio, long recordings

Test scope:
- smoke test, 2026-10-25; model: None; no language model was involved; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: faster-whisper 1.2.1 transcribed 41 s of speech at 2.4 percent word error with the base.en model in float32 (deterministic), but the int8 setting gave 2 to 60 percent on identical audio across runs.
- Not tested: Other audio: one synthetic (Piper) voice, English only, no noise, no accents, no long recordings; GPU, other compute types (float16, int8_float16) and the larger models; Not a benchmark; this is a smoke test on 41 seconds of audio.

Sources:
- faster-whisper repository (GitHub): https://github.com/SYSTRAN/faster-whisper
Cite as: Agentic Weekly, "faster-whisper v1.2.1: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/faster-whisper

## AutoGen python-v0.7.5 (2025-09-30, Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/autogen (markdown: https://theagenticweekly.com/tools/autogen.md)
Vendor: Microsoft

AutoGen's README says the project is no longer actively developed with new features, warns that without ongoing maintenance and security fixes vulnerabilities may remain unaddressed, and links an AutoGen to Microsoft Agent Framework migration guide. The latest release tag we saw is python-v0.7.5 from Sep 30, 2025.

Verdict: Microsoft's open-source multi-agent framework; its README now says the project is in maintenance mode and points to Microsoft Agent Framework.
Good for: Existing AutoGen projects that need reference material; Researchers studying multi-agent patterns
Skip if: You are starting a new production project: Microsoft points to Agent Framework instead

What's new:
- python-v0.7.5 (2025-09-30): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with AutoGen
1. Get access: For new work, follow Microsoft's migration guide to Microsoft Agent Framework; for existing code, pin python-v0.7.5.
2. Check the limits and costs first: Free to self-host. Open-source library; costs come from the model APIs you call.

Not verified:
- Exact end-of-support dates

Test scope:
- smoke test, 2026-10-25; model: Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: AutoGen 0.7.5 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.9 s.
- Not tested: Any vendor-hosted model: everything ran on a local 3B model, so tool-call reliability with frontier models is not measured here; Multi-agent flows, memory, long conversations, streaming, tracing and cloud deployment; Anything beyond one tool and one question; three trials is a smoke test, not a benchmark.

Sources:
- AutoGen README: https://github.com/microsoft/autogen
- Microsoft Agent Framework migration guide: https://github.com/microsoft/autogen
Cite as: Agentic Weekly, "AutoGen python-v0.7.5: what's new, verdict & setup", last updated 2026-10-25, https://theagenticweekly.com/tools/autogen

## GPT4All v3.10.0 (2025-02-25, Tested (smoke test, 2026-10-26, local 3B model (Qwen2.5-3B)))
URL: https://theagenticweekly.com/tools/gpt4all (markdown: https://theagenticweekly.com/tools/gpt4all.md)
Vendor: Nomic

GPT4All's page says it runs open-source models locally on your device with no cloud required. The repository is MIT-licensed and the newest release tag we saw is v3.10.0 (Feb 25, 2025).

Verdict: Local chatbot from Nomic that runs open-source models on your device; the latest GitHub release we saw is v3.10.0 from Feb 2025.
Good for: People who want a simple private desktop chatbot
Skip if: You want a project with recent releases (last release we saw is Feb 2025)

What's new:
- v3.10.0 (2025-02-25): the newest release listed on the official page or repository when we read it on 2026-10-25.

Tutorial: Get started with GPT4All
1. Get access: Download GPT4All from gpt4all.io and pick a model.
2. Check the limits and costs first: Free. No pricing page read.

Not verified:
- Current maintenance status

Test scope:
- smoke test, 2026-10-26; model: Qwen2.5-3B-Instruct (Q4_K_M) loaded directly from a GGUF file by the GPT4All Python SDK; environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key
- Ran: GPT4All's Python SDK 2.8.2 loaded a local 3B GGUF in 4.01 s and answered 17 + 25 with 42 in 3 of 3 trials, but kept writing past the answer.
- Not tested: The GPT4All desktop app and its LocalDocs feature: only the Python SDK was run; GPT4All's own downloadable model catalogue and its download speed; we used a file already on disk; Speed: the machine was heavily shared during the run, so the timings are not a measure of the SDK; A single prompt and three trials is a smoke test, not a benchmark.

Sources:
- GPT4All: https://gpt4all.io/
- GPT4All repository (GitHub): https://github.com/nomic-ai/gpt4all
Cite as: Agentic Weekly, "GPT4All v3.10.0: what's new, verdict & setup", last updated 2026-10-26, https://theagenticweekly.com/tools/gpt4all

# AI models (sourced facts)

# Newest and best AI models, October 2026

Newest: GPT-6.1 Terra (OpenAI), released Oct 26, 2026. Updated 2026-10-26 (88 models verified against vendor sources) · Canonical: https://theagenticweekly.com/models · JSON: https://theagenticweekly.com/models.json · Answers: https://theagenticweekly.com/answers.json

## Answers

- **What is the newest AI model release in this dataset?** GPT-6.1 Terra (OpenAI), released Oct 26, 2026. Derived from the release dates in our data; check /models.json for anything newer than the as-of date. (Criterion: Latest release date among models we verified against vendor sources; models sharing the date are all listed. As of 2026-10-26.)
- **What is the latest OpenAI language model?** GPT-6.1 Terra, released Oct 26, 2026. (Criterion: Newest release date among OpenAI language models we verified. As of 2026-10-26.)
- **What is the latest Moonshot AI language model?** Kimi K3.1 Pro, released Oct 23, 2026. (Criterion: Newest release date among Moonshot AI language models we verified. As of 2026-10-26.)
- **What is the latest NVIDIA language model?** Nemotron 3.5 Bolt E4B (4GB), released Oct 23, 2026. (Criterion: Newest release date among NVIDIA language models we verified. As of 2026-10-26.)
- **What is the latest Inception language model?** Mercury 3, released Oct 19, 2026. (Criterion: Newest release date among Inception language models we verified. As of 2026-10-26.)
- **What is the latest MiniMax language model?** MiniMax M3.1 Flash, released Oct 19, 2026. (Criterion: Newest release date among MiniMax language models we verified. As of 2026-10-26.)
- **What is the latest DeepSeek language model?** DeepSeek-V4.1-Flash-1017, released Oct 17, 2026. (Criterion: Newest release date among DeepSeek language models we verified. As of 2026-10-26.)
- **What is the latest Meta language model?** Muse Spark Open, released Oct 15, 2026. (Criterion: Newest release date among Meta language models we verified. As of 2026-10-26.)
- **What is the latest Google language model?** Gemini 4 Argon, released Oct 13, 2026. (Criterion: Newest release date among Google language models we verified. As of 2026-10-26.)
- **What is the latest Anthropic language model?** Claude Haiku 5.5, released Oct 7, 2026. (Criterion: Newest release date among Anthropic language models we verified. As of 2026-10-26.)
- **What is the latest Mistral AI language model?** Mistral Large 4, released Oct 6, 2026 (public preview). (Criterion: Newest release date among Mistral AI language models we verified. As of 2026-10-26.)
- **What is the latest xAI language model?** Grok 4.7, released Sep 21, 2026. (Criterion: Newest release date among xAI language models we verified. As of 2026-10-26.)
- **What is the latest Cohere language model?** North Small Translate, released Sep 9, 2026. (Criterion: Newest release date among Cohere language models we verified. As of 2026-10-26.)
- **What is the latest Z.ai language model?** GLM-5.3-Flash, released Aug 26, 2026. (Criterion: Newest release date among Z.ai language models we verified. As of 2026-10-26.)
- **What is the latest Alibaba language model?** Qwen3.8-27B, released Aug 14, 2026. (Criterion: Newest release date among Alibaba language models we verified. As of 2026-10-26.)
- **What is the newest image AI model in this dataset?** Nano Banana 2.1 (Google), released Oct 6, 2026. (Criterion: Newest release date among image models we verified. As of 2026-10-26.)
- **What is the newest voice AI model in this dataset?** Voxtral 3 (Mistral AI), released Oct 20, 2026. (Criterion: Newest release date among voice models we verified. As of 2026-10-26.)
- **What is the newest video AI model in this dataset?** Gemini Omni 1.1 Flash (Google), released Aug 27, 2026. (Criterion: Newest release date among video models we verified. As of 2026-10-26.)
- **What is the best AI model for coding right now?** Claude Opus 5.5; next: GPT-6 Astra, Claude Sonnet 5.5. (Criterion: Highest-ranked model on the LMArena WebDev leaderboard (snapshot 2026-10-22) that is also in our model list. As of 2026-10-22.)
- **Which AI model is ranked first for general chat preference right now?** Gemini 4 Argon; next: Claude Opus 5.5. (Criterion: Rank 1 on the LMArena Text Arena (snapshot 2026-10-08). As of 2026-10-08.)
- **What is the best open-weights AI model right now?** DeepSeek-V4.1-Flash-1017; next: Kimi K3, GLM-5.3. (Criterion: Highest-ranked open-weights model in our list on the LMArena WebDev leaderboard (snapshot 2026-10-22). As of 2026-10-22.)
- **What is the best open-weights model small enough to run on one machine?** Qwen3.8-27B; next: Gemma 4. (Criterion: Highest-ranked open-weights model of 31B parameters or fewer in our list on the LMArena WebDev leaderboard (snapshot 2026-10-22). As of 2026-10-22.)
- **What is the best AI image generation model right now?** GPT-Image-2.5 Sunburst; next: Nano Banana 2.1. (Criterion: Rank 1 on the LMArena Text-to-Image leaderboard (snapshot 2026-10-07). As of 2026-10-07.)
- **What is the best AI model for agents right now?** No single answer. We found no neutral public agent leaderboard we could verify this week. Candidates: Claude Opus 5.5, GPT-6.1 Sol, Kimi K3. (Criterion: We do not name one. We found no neutral public agent leaderboard we could verify this week. As of 2026-10-26.)
- **What is the best AI voice model right now?** No single answer. We found no neutral public voice leaderboard we could verify this week. Candidates: GPT-Live 1, Gemini 3.8 Live. (Criterion: We do not name one. We found no neutral public voice leaderboard we could verify this week. As of 2026-10-26.)
- **What is the best AI video generation model right now?** No single answer. We verified only one video model against vendor sources. Candidates: Gemini Omni 1.1 Flash. (Criterion: We do not name one. We verified only one video model against vendor sources. As of 2026-10-26.)
- **What is the cheapest AI language model API in this dataset?** Ministral 3 14B at $0.2 in / $0.2 out per 1M tokens. DeepSeek-V4.1-Flash is priced in CNY and is not compared here. (Criterion: Lowest sum of USD list price per 1M input plus 1M output tokens among language models with a USD price on the vendor's page. Cached rates, tiers and free tiers are ignored. As of 2026-10-26.)
- **Which AI models have a context window of 1M tokens or more?** GPT-6.1 Terra, Kimi K3.1 Pro, Kimi K3.1 Pro Highspeed, Mercury 3, MiniMax M3.1 Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Gemini 4 Argon, Muse Wonder, Claude Haiku 5.5, Mistral Large 4, GPT-6.1 Sol, Claude Sonnet 5.5, Claude Opus 5.5, GPT-6 Luna, Claude Fable 5.1, DeepSeek-V4.1-Flash, GPT-6 Astra, Gemini 3.8 Flash, Claude Mythos 5.1, GLM-5.3, Claude Opus 5, Gemini 3.5 Flash-Lite, Gemini 3.6 Flash, Kimi K3, GPT-5.6 Luna, GPT-5.6 Sol, GPT-5.6 Terra, Claude Sonnet 5, GLM-5.2, Claude Fable 5, Claude Mythos 5, Claude Opus 4.8, Gemini 3.1 Flash-Lite, GPT-5.5, GPT-5.5 Pro, Grok 4.20, GPT-5.4. The largest stated window is 1,050,000 tokens. (Criterion: Context window stated by the vendor, in tokens (Gemini and GPT-6 list slightly above 1M). As of 2026-10-26.)
- **Which models in this dataset have downloadable open weights?** Nemotron 3.5 Bolt E4B (4GB) (OpenMDW-1.1); MiniMax M3.1 Flash (MiniMax Community License); DeepSeek-V4.1-Flash-1017 (MIT); Muse Spark Open (Muse Model License (limited commercial terms)); DeepSeek-V4.1-Flash (MIT); North Small Translate (CC BY-NC 4.0); GLM-5.3 (GLM-5.3 License (MIT-style plus a MaaS condition)); Qwen3.8-27B (Apache-2.0); Qwen3.8-2.4T-A95B (Qwen3.8-Max License (custom)); Kimi K3 (Kimi K3 License (custom)); Cohere Transcribe Arabic (Apache 2.0); North Mini Code (Apache 2.0); Command A+ (Apache 2.0); Mistral Medium 3.5 (Modified MIT); Qwen3.6-27B (Apache-2.0); Gemma 4 (Apache 2.0); Mistral Small 4 (Apache 2.0); Ministral 3 14B (Apache 2.0). (Criterion: Weights published for download; the licence is shown because terms differ and some limit commercial use. As of 2026-10-26.)

## Leaderboards we cite

- [LMArena Text Arena](https://arena.ai/leaderboard/text), snapshot 2026-10-08: Crowd-voted human preference between anonymous model answers (Elo-style score). Top entries then: Gemini 4 Argon (high), score marked preliminary; Claude Opus 5.5 (high); Claude Opus 4.6 (high). We read this snapshot on Oct 8, 2026. The live board changes daily; check it before relying on the order. Third-party mirrors show different numbers, so we cite arena.ai only.
- [LMArena WebDev](https://arena.ai/leaderboard/webdev), snapshot 2026-10-22: Crowd-voted preference on front-end web development tasks, including agentic coding workflows. Top entries then: Claude Opus 5.5 (Max); GPT-6 Astra (Max); Claude Sonnet 5.5 (xHigh). Highest open-weights entries from our model list on this snapshot: DeepSeek-V4.1-Flash-20261017 (#14, MIT), Kimi K3 (#17), GLM-5.3 (#25), Qwen3.8-27B (#32, Apache 2.0), Muse Spark Open (#53), Gemma 4 31B (#113). Page showed 151 models and 923,854 votes.
- [LMArena Text-to-Image](https://arena.ai/leaderboard/text-to-image), snapshot 2026-10-07: Crowd-voted preference between images generated from the same prompt. Top entries then: GPT-Image-2.5 Sunburst (preliminary); GPT-Image-2.5 Flare (preliminary); GPT-Image-2 (Medium). Nano Banana 2.1 ranked #6 (preliminary) on this snapshot.

## All models, newest first

| Model | Vendor | Released | Context | Modalities (in → out) | Licence | Price | Verified |
| --- | --- | --- | --- | --- | --- | --- | --- |
| [GPT-6.1 Terra](https://theagenticweekly.com/models/gpt-6-1-terra) | OpenAI | 2026-10-26 | 1.05M | text, image → text | Proprietary (API) | $1 in / $5 out per 1M tokens | 2026-10-26 |
| [Kimi K3.1 Pro](https://theagenticweekly.com/models/kimi-k3-1-pro) | Moonshot AI | 2026-10-23 | 1.05M | n/a → n/a | Not yet published | $2 in / $10 out per 1M tokens | 2026-10-26 |
| [Kimi K3.1 Pro Highspeed](https://theagenticweekly.com/models/kimi-k3-1-pro-highspeed) | Moonshot AI | 2026-10-23 | 1.05M | n/a → n/a | n/a | $4 in / $20 out per 1M tokens | 2026-10-26 |
| [Nemotron 3.5 Bolt E4B (4GB)](https://theagenticweekly.com/models/nemotron-3-5-bolt-e4b) | NVIDIA | 2026-10-23 | 64K | text → text | OpenMDW-1.1 | not stated | 2026-10-26 |
| [Voxtral 3](https://theagenticweekly.com/models/voxtral-3) | Mistral AI | 2026-10-20 |  | n/a → n/a | n/a | Mini Transcribe 3, per minute of audio $0.0035; Small Transcribe 3, per minute of audio $0.004; Mini Transcribe Realtime 3, per minute of audio $0.008; Mini TTS 3, per 1M characters $13; Voice agent API with Mistral Small 4 for reasoning, per minute $0.04; Voice agent API with a larger model, per minute $0.08 (see each line) | 2026-10-26 |
| [Mercury 3](https://theagenticweekly.com/models/mercury-3) | Inception | 2026-10-19 | 1M | n/a → n/a | Proprietary (API) | $0.1 in / $0.35 out per 1M tokens | 2026-10-26 |
| [MiniMax M3.1 Flash](https://theagenticweekly.com/models/minimax-m3-1-flash) | MiniMax | 2026-10-19 | 1M | text, image, video → text | MiniMax Community License | $0.15 in / $0.6 out per 1M tokens | 2026-10-26 |
| [DeepSeek-V4.1-Flash-1017](https://theagenticweekly.com/models/deepseek-v4-1-flash-1017) | DeepSeek | 2026-10-17 | 1M | text, image → text | MIT | ¥1 in / ¥4 out per 1M tokens; peak: ¥2 / ¥8 | 2026-10-26 |
| [Muse Spark Open](https://theagenticweekly.com/models/muse-spark-open) | Meta | 2026-10-15 | 1M | text, image, video, audio → text | Muse Model License (limited commercial terms) | not stated | 2026-10-26 |
| [Grok Voice 3.0](https://theagenticweekly.com/models/grok-voice-3-0) | xAI | 2026-10-14 |  | n/a → n/a | Proprietary (API) | not stated | 2026-10-26 |
| [Gemini 4 Argon](https://theagenticweekly.com/models/gemini-4-argon) | Google | 2026-10-13 | 1.05M | text, image, video, audio, pdf → text | Proprietary (API) | $2 in / $10 out per 1M tokens; from Dec 1, 2026: $4 / $20 | 2026-10-26 |
| [Muse Wonder](https://theagenticweekly.com/models/muse-wonder) | Meta | 2026-10-12 | 1.05M | text, image, video, audio, pdf → text | Proprietary (API) | $4 in / $20 out per 1M tokens | 2026-10-26 |
| [Muse Wonder Full](https://theagenticweekly.com/models/muse-wonder-full) | Meta | 2026-10-12 |  | n/a → n/a | Proprietary (subscription) | not stated | 2026-10-26 |
| [Claude Haiku 5.5](https://theagenticweekly.com/models/claude-haiku-5-5) | Anthropic | 2026-10-07 | 1M | text, image → text | Proprietary (API) | $0.1 in / $0.5 out per 1M tokens | 2026-10-26 |
| [Mistral Large 4](https://theagenticweekly.com/models/mistral-large-4) | Mistral AI | 2026-10-06 | 1M | n/a → n/a | Not yet published | $1.36 in / $4.18 out per 1M tokens | 2026-10-26 |
| [Nano Banana 2.1](https://theagenticweekly.com/models/nano-banana-2-1) | Google | 2026-10-06 |  | n/a → image | Proprietary (API) | 1K image $0.0336; 2K image $0.0504; 4K image $0.113 (per image) | 2026-10-26 |
| [GPT-6.1 Sol](https://theagenticweekly.com/models/gpt-6-1-sol) | OpenAI | 2026-09-29 | 1.05M | text, image → text | Proprietary (API) | $2 in / $10 out per 1M tokens | 2026-10-26 |
| [Claude Sonnet 5.5](https://theagenticweekly.com/models/claude-sonnet-5-5) | Anthropic | 2026-09-28 | 1M | text, image → text | Proprietary (API) | $2 in / $10 out per 1M tokens | 2026-10-26 |
| [Claude Opus 5.5](https://theagenticweekly.com/models/claude-opus-5-5) | Anthropic | 2026-09-22 | 1M | text, image → text | Proprietary (API) | $4 in / $20 out per 1M tokens | 2026-10-26 |
| [Gemini 3.8 Flash TTS](https://theagenticweekly.com/models/gemini-3-8-flash-tts) | Google | 2026-09-22 |  | n/a → audio | Proprietary (API) | Text input $0.5; Audio output $9 (per 1M tokens) | 2026-10-26 |
| [Gemini 3.8 Flash-Lite TTS](https://theagenticweekly.com/models/gemini-3-8-flash-lite-tts) | Google | 2026-09-22 | 8K | text → audio | Proprietary (API) | not stated | 2026-10-26 |
| [GPT-6 Luna](https://theagenticweekly.com/models/gpt-6-luna) | OpenAI | 2026-09-22 | 1.05M | text, image → text | Proprietary (API) | $0.1 in / $0.5 out per 1M tokens | 2026-10-26 |
| [GPT-6 Sol](https://theagenticweekly.com/models/gpt-6-sol) | OpenAI | 2026-09-22 |  | text, image → text | Proprietary (API) | $2 in / $10 out per 1M tokens | 2026-10-26 |
| [Grok 4.7](https://theagenticweekly.com/models/grok-4-7) | xAI | 2026-09-21 | 500K | text, image → text | Proprietary (API) | $2 in / $6 out per 1M tokens | 2026-10-26 |
| [Grok Voice Transcribe 2.0](https://theagenticweekly.com/models/grok-voice-transcribe-2-0) | xAI | 2026-09-17 |  | audio → text | Proprietary (API) | not stated | 2026-10-26 |
| [Claude Fable 5.1](https://theagenticweekly.com/models/claude-fable-5-1) | Anthropic | 2026-09-15 | 1M | text, image → text | Proprietary (API) | $10 in / $50 out per 1M tokens | 2026-10-26 |
| [Gemini 3.8 Live](https://theagenticweekly.com/models/gemini-3-8-live) | Google | 2026-09-15 |  | text, audio → text, audio | Proprietary (API) | Text input $0.75; Audio input $3; Text output $4.5; Audio output $12 (per 1M tokens) | 2026-10-26 |
| [Gemini 3.8 Live Extended Thinking](https://theagenticweekly.com/models/gemini-3-8-live-extended-thinking) | Google | 2026-09-15 | 131K | text, image, audio → text, audio | Proprietary (API) | not stated | 2026-10-26 |
| [DeepSeek-V4.1-Flash](https://theagenticweekly.com/models/deepseek-v4-1-flash) | DeepSeek | 2026-09-10 | 1M | text, image → text | MIT | ¥1 in / ¥4 out per 1M tokens; peak: ¥2 / ¥8 | 2026-10-26 |
| [GPT-Live 1](https://theagenticweekly.com/models/gpt-live-1) | OpenAI | 2026-09-10 |  | text, audio → text, audio | Proprietary (API) | Voice session, billed per second $0.05 (per minute) | 2026-10-26 |
| [North Small Translate](https://theagenticweekly.com/models/north-small-translate) | Cohere | 2026-09-09 | 16K | n/a → n/a | CC BY-NC 4.0 | not stated | 2026-10-26 |
| [GPT-Image-2.5 Flare](https://theagenticweekly.com/models/gpt-image-2-5-flare) | OpenAI | 2026-09-08 |  | text, image → image | Proprietary (API) | $5 in / $30 out per 1M tokens | 2026-10-26 |
| [GPT-Image-2.5 Sunburst](https://theagenticweekly.com/models/gpt-image-2-5-sunburst) | OpenAI | 2026-09-08 |  | n/a → image | Proprietary (API) | $5 in / $30 out per 1M tokens | 2026-10-26 |
| [GPT-6 Astra](https://theagenticweekly.com/models/gpt-6-astra) | OpenAI | 2026-09-03 | 1.05M | text, image → text | Proprietary (API) | $10 in / $50 out per 1M tokens | 2026-10-26 |
| [Gemini 3.8 Flash](https://theagenticweekly.com/models/gemini-3-8-flash) | Google | 2026-09-02 | 1.05M | text, image, video, audio, pdf → n/a | Proprietary (API) | $0.75 in / $3.75 out per 1M tokens; from Jan 1, 2027: $1.5 / $7.5 | 2026-10-26 |
| [Claude Mythos 5.1](https://theagenticweekly.com/models/claude-mythos-5-1) | Anthropic | 2026-09-01 | 1M | n/a → n/a | Proprietary (API) | $10 in / $50 out per 1M tokens | 2026-10-26 |
| [Gemini Omni 1.1 Flash](https://theagenticweekly.com/models/gemini-omni-1-1-flash) | Google | 2026-08-27 |  | n/a → video | Proprietary (API) | Input per 1M tokens $1.5; Output per second of 720p video (approx.) $0.1 (mixed) | 2026-10-26 |
| [Gemini 3.5 Transcribe](https://theagenticweekly.com/models/gemini-3-5-transcribe) | Google | 2026-08-26 |  | audio → text | Proprietary (API) | not stated | 2026-10-26 |
| [GLM-5.3-Flash](https://theagenticweekly.com/models/glm-5-3-flash) | Z.ai | 2026-08-26 |  | n/a → n/a | n/a | not stated | 2026-10-26 |
| [GLM-5.3](https://theagenticweekly.com/models/glm-5-3) | Z.ai | 2026-08-14 | 1.05M | n/a → text | GLM-5.3 License (MIT-style plus a MaaS condition) | $1.4 in / $4.4 out per 1M tokens | 2026-10-26 |
| [Qwen3.8-27B](https://theagenticweekly.com/models/qwen3-8-27b) | Alibaba | 2026-08-14 | 262K | n/a → n/a | Apache-2.0 | not stated | 2026-10-26 |
| [Grok 4.6](https://theagenticweekly.com/models/grok-4-6) | xAI | 2026-08-12 | 500K | text, image → text | Proprietary (API) | $2 in / $6 out per 1M tokens | 2026-10-26 |
| [Qwen3.8-2.4T-A95B](https://theagenticweekly.com/models/qwen3-8-2-4t-a95b) | Alibaba | 2026-08-12 | 262K | n/a → n/a | Qwen3.8-Max License (custom) | not stated | 2026-10-26 |
| [Gemini Robotics ER 2](https://theagenticweekly.com/models/gemini-robotics-er-2) | Google | 2026-07-30 |  | n/a → n/a | Proprietary (API) | not stated | 2026-10-26 |
| [GPT-Live-Transcribe](https://theagenticweekly.com/models/gpt-live-transcribe) | OpenAI | 2026-07-28 |  | audio, text → text | Proprietary (API) | Realtime transcription $0.017 (per minute) | 2026-10-26 |
| [GPT-Transcribe](https://theagenticweekly.com/models/gpt-transcribe) | OpenAI | 2026-07-28 |  | audio, text → text | Proprietary (API) | Transcription $0.0045 (per minute) | 2026-10-26 |
| [Claude Opus 5](https://theagenticweekly.com/models/claude-opus-5) | Anthropic | 2026-07-24 | 1M | n/a → n/a | Proprietary (API) | $5 in / $25 out per 1M tokens | 2026-10-26 |
| [Gemini 3.5 Flash-Lite](https://theagenticweekly.com/models/gemini-3-5-flash-lite) | Google | 2026-07-21 | 1.05M | text, image, video, audio, pdf → text | Proprietary (API) | not stated | 2026-10-26 |
| [Gemini 3.6 Flash](https://theagenticweekly.com/models/gemini-3-6-flash) | Google | 2026-07-21 | 1.05M | text, image, video, audio, pdf → text | Proprietary (API) | not stated | 2026-10-26 |
| [Kimi K3](https://theagenticweekly.com/models/kimi-k3) | Moonshot AI | 2026-07-16 | 1M | n/a → n/a | Kimi K3 License (custom) | $3 in / $15 out per 1M tokens | 2026-10-26 |
| [GPT-5.6 Luna](https://theagenticweekly.com/models/gpt-5-6-luna) | OpenAI | 2026-07-09 | 1.05M | text, image → text | Proprietary (API) | $0.2 in / $1.2 out per 1M tokens | 2026-10-26 |
| [GPT-5.6 Sol](https://theagenticweekly.com/models/gpt-5-6-sol) | OpenAI | 2026-07-09 | 1.05M | text, image → text | Proprietary (API) | $4 in / $20 out per 1M tokens | 2026-10-26 |
| [GPT-5.6 Terra](https://theagenticweekly.com/models/gpt-5-6-terra) | OpenAI | 2026-07-09 | 1.05M | text, image → text | Proprietary (API) | $2 in / $12 out per 1M tokens | 2026-10-26 |
| [Grok 4.5](https://theagenticweekly.com/models/grok-4-5) | xAI | 2026-07-08 | 500K | text, image → text | Proprietary (API) | $2 in / $6 out per 1M tokens | 2026-10-26 |
| [Cohere Transcribe Arabic](https://theagenticweekly.com/models/cohere-transcribe-arabic) | Cohere | 2026-07-07 |  | n/a → n/a | Apache 2.0 | not stated | 2026-10-26 |
| [GPT-Realtime-2.1](https://theagenticweekly.com/models/gpt-realtime-2-1) | OpenAI | 2026-07-06 | 128K | text, audio, image → text, audio | Proprietary (API) | $4 in / $24 out per 1M tokens | 2026-10-26 |
| [GPT-Realtime-2.1 mini](https://theagenticweekly.com/models/gpt-realtime-2-1-mini) | OpenAI | 2026-07-06 | 128K | text, audio, image → text, audio | Proprietary (API) | $0.6 in / $2.4 out per 1M tokens | 2026-10-26 |
| [Claude Sonnet 5](https://theagenticweekly.com/models/claude-sonnet-5) | Anthropic | 2026-06-30 | 1M | n/a → n/a | Proprietary (API) | $2 in / $10 out per 1M tokens | 2026-10-26 |
| [Nano Banana 2 Lite](https://theagenticweekly.com/models/nano-banana-2-lite) | Google | 2026-06-30 | 66K | text, image, video, pdf → image, text | Proprietary (API) | not stated | 2026-10-26 |
| [GLM-5.2](https://theagenticweekly.com/models/glm-5-2) | Z.ai | 2026-06-16 | 1M | n/a → n/a | n/a | not stated | 2026-10-26 |
| [Claude Fable 5](https://theagenticweekly.com/models/claude-fable-5) | Anthropic | 2026-06-09 | 1M | n/a → n/a | Proprietary (API) | $10 in / $50 out per 1M tokens | 2026-10-26 |
| [Claude Mythos 5](https://theagenticweekly.com/models/claude-mythos-5) | Anthropic | 2026-06-09 | 1M | n/a → n/a | Proprietary (API) | not stated | 2026-10-26 |
| [North Mini Code](https://theagenticweekly.com/models/north-mini-code) | Cohere | 2026-06-09 | 256K | n/a → n/a | Apache 2.0 | not stated | 2026-10-26 |
| [Claude Opus 4.8](https://theagenticweekly.com/models/claude-opus-4-8) | Anthropic | 2026-05-28 | 1M | n/a → n/a | Proprietary (API) | $5 in / $25 out per 1M tokens | 2026-10-26 |
| [Nano Banana Pro](https://theagenticweekly.com/models/nano-banana-pro) | Google | 2026-05-28 | 66K | image, text → image, text | Proprietary (API) | not stated | 2026-10-26 |
| [Command A+](https://theagenticweekly.com/models/command-a-plus) | Cohere | 2026-05-20 | 128K | n/a → n/a | Apache 2.0 | not stated | 2026-10-26 |
| [Grok Build 0.1](https://theagenticweekly.com/models/grok-build-0-1) | xAI | 2026-05-19 | 256K | text, image → text | Proprietary (API) | $1 in / $2 out per 1M tokens | 2026-10-26 |
| [Gemini 3.1 Flash-Lite](https://theagenticweekly.com/models/gemini-3-1-flash-lite) | Google | 2026-05-07 | 1.05M | text, image, video, audio, pdf → text | Proprietary (API) | not stated | 2026-10-26 |
| [Mistral Medium 3.5](https://theagenticweekly.com/models/mistral-medium-3-5) | Mistral AI | 2026-04-28 | 256K | n/a → n/a | Modified MIT | $1.5 in / $7.5 out per 1M tokens | 2026-10-26 |
| [GPT-5.5](https://theagenticweekly.com/models/gpt-5-5) | OpenAI | 2026-04-24 | 1.05M | text, image → text | Proprietary (API) | $5 in / $30 out per 1M tokens | 2026-10-26 |
| [GPT-5.5 Pro](https://theagenticweekly.com/models/gpt-5-5-pro) | OpenAI | 2026-04-24 | 1.05M | text, image → text | Proprietary (API) | $30 in / $180 out per 1M tokens | 2026-10-26 |
| [Qwen3.6-27B](https://theagenticweekly.com/models/qwen3-6-27b) | Alibaba | 2026-04-22 | 262K | n/a → n/a | Apache-2.0 | not stated | 2026-10-26 |
| [GPT-Image-2](https://theagenticweekly.com/models/gpt-image-2) | OpenAI | 2026-04-21 |  | text, image → image | Proprietary (API) | not stated | 2026-10-26 |
| [Claude Opus 4.7](https://theagenticweekly.com/models/claude-opus-4-7) | Anthropic | 2026-04-16 |  | n/a → n/a | Proprietary (API) | $5 in / $25 out per 1M tokens | 2026-10-26 |
| [GLM-5.1](https://theagenticweekly.com/models/glm-5-1) | Z.ai | 2026-04-07 |  | n/a → n/a | n/a | not stated | 2026-10-26 |
| [Gemma 4](https://theagenticweekly.com/models/gemma-4) | Google | 2026-03-31 | 256K | text, image, video, audio → text | Apache 2.0 | not stated | 2026-10-26 |
| [Cohere Transcribe](https://theagenticweekly.com/models/cohere-transcribe) | Cohere | 2026-03-26 |  | audio → text | n/a | not stated | 2026-10-26 |
| [Voxtral TTS](https://theagenticweekly.com/models/voxtral-tts) | Mistral AI | 2026-03-23 |  | n/a → audio | n/a | Speech output $16 (per 1M characters) | 2026-10-26 |
| [GPT-5.4 mini](https://theagenticweekly.com/models/gpt-5-4-mini) | OpenAI | 2026-03-17 | 400K | text, image → text | Proprietary (API) | $0.75 in / $4.5 out per 1M tokens | 2026-10-26 |
| [Mistral Small 4](https://theagenticweekly.com/models/mistral-small-4) | Mistral AI | 2026-03-16 | 256K | n/a → n/a | Apache 2.0 | $0.15 in / $0.6 out per 1M tokens | 2026-10-26 |
| [Grok 4.20](https://theagenticweekly.com/models/grok-4-20) | xAI | 2026-03-10 | 1M | text, image → text | Proprietary (API) | $1.25 in / $2.5 out per 1M tokens | 2026-10-26 |
| [GPT-5.4](https://theagenticweekly.com/models/gpt-5-4) | OpenAI | 2026-03-05 | 1.05M | text, image → text | Proprietary (API) | $2.5 in / $15 out per 1M tokens | 2026-10-26 |
| [Gemini 3.1 Pro](https://theagenticweekly.com/models/gemini-3-1-pro) | Google | 2026-02-19 |  | n/a → n/a | Proprietary (API) | not stated | 2026-10-26 |
| [Claude Sonnet 4.6](https://theagenticweekly.com/models/claude-sonnet-4-6) | Anthropic | 2026-02-17 |  | n/a → n/a | Proprietary (API) | not stated | 2026-10-26 |
| [GLM-5](https://theagenticweekly.com/models/glm-5) | Z.ai | 2026-02-12 |  | n/a → n/a | n/a | not stated | 2026-10-26 |
| [Claude Opus 4.6](https://theagenticweekly.com/models/claude-opus-4-6) | Anthropic | 2026-02-05 |  | n/a → n/a | Proprietary (API) | $5 in / $25 out per 1M tokens | 2026-10-26 |
| [GLM-Image](https://theagenticweekly.com/models/glm-image) | Z.ai | 2026-01-14 |  | n/a → image | n/a | not stated | 2026-10-26 |
| [Ministral 3 14B](https://theagenticweekly.com/models/ministral-3-14b) | Mistral AI | 2025-12-02 | 256K | n/a → n/a | Apache 2.0 | $0.2 in / $0.2 out per 1M tokens | 2026-10-26 |

## Frontier models

The vendors' flagship general-purpose language models, as named on their own release pages. Frontier here means flagship tier from a vendor, not a ranking.

- [Muse Spark Open](https://theagenticweekly.com/models/muse-spark-open) (Meta, 2026-10-15)
- [Gemini 4 Argon](https://theagenticweekly.com/models/gemini-4-argon) (Google, 2026-10-13)
- [Muse Wonder](https://theagenticweekly.com/models/muse-wonder) (Meta, 2026-10-12)
- [Muse Wonder Full](https://theagenticweekly.com/models/muse-wonder-full) (Meta, 2026-10-12)
- [Claude Haiku 5.5](https://theagenticweekly.com/models/claude-haiku-5-5) (Anthropic, 2026-10-07)
- [Mistral Large 4](https://theagenticweekly.com/models/mistral-large-4) (Mistral AI, 2026-10-06)
- [GPT-6.1 Sol](https://theagenticweekly.com/models/gpt-6-1-sol) (OpenAI, 2026-09-29)
- [Claude Sonnet 5.5](https://theagenticweekly.com/models/claude-sonnet-5-5) (Anthropic, 2026-09-28)
- [Claude Opus 5.5](https://theagenticweekly.com/models/claude-opus-5-5) (Anthropic, 2026-09-22)
- [GPT-6 Luna](https://theagenticweekly.com/models/gpt-6-luna) (OpenAI, 2026-09-22)
- [GPT-6 Sol](https://theagenticweekly.com/models/gpt-6-sol) (OpenAI, 2026-09-22)
- [Grok 4.7](https://theagenticweekly.com/models/grok-4-7) (xAI, 2026-09-21)
- [Claude Fable 5.1](https://theagenticweekly.com/models/claude-fable-5-1) (Anthropic, 2026-09-15)
- [GPT-6 Astra](https://theagenticweekly.com/models/gpt-6-astra) (OpenAI, 2026-09-03)
- [Gemini 3.8 Flash](https://theagenticweekly.com/models/gemini-3-8-flash) (Google, 2026-09-02)
- [Claude Mythos 5.1](https://theagenticweekly.com/models/claude-mythos-5-1) (Anthropic, 2026-09-01)
- [Grok 4.6](https://theagenticweekly.com/models/grok-4-6) (xAI, 2026-08-12)
- [Qwen3.8-2.4T-A95B](https://theagenticweekly.com/models/qwen3-8-2-4t-a95b) (Alibaba, 2026-08-12)
- [Claude Opus 5](https://theagenticweekly.com/models/claude-opus-5) (Anthropic, 2026-07-24)
- [GPT-5.6 Sol](https://theagenticweekly.com/models/gpt-5-6-sol) (OpenAI, 2026-07-09)
- [GPT-5.6 Terra](https://theagenticweekly.com/models/gpt-5-6-terra) (OpenAI, 2026-07-09)
- [Claude Sonnet 5](https://theagenticweekly.com/models/claude-sonnet-5) (Anthropic, 2026-06-30)
- [Claude Fable 5](https://theagenticweekly.com/models/claude-fable-5) (Anthropic, 2026-06-09)
- [Claude Mythos 5](https://theagenticweekly.com/models/claude-mythos-5) (Anthropic, 2026-06-09)
- [Claude Opus 4.8](https://theagenticweekly.com/models/claude-opus-4-8) (Anthropic, 2026-05-28)
- [Mistral Medium 3.5](https://theagenticweekly.com/models/mistral-medium-3-5) (Mistral AI, 2026-04-28)
- [GPT-5.5](https://theagenticweekly.com/models/gpt-5-5) (OpenAI, 2026-04-24)
- [Claude Opus 4.7](https://theagenticweekly.com/models/claude-opus-4-7) (Anthropic, 2026-04-16)
- [Grok 4.20](https://theagenticweekly.com/models/grok-4-20) (xAI, 2026-03-10)
- [Gemini 3.1 Pro](https://theagenticweekly.com/models/gemini-3-1-pro) (Google, 2026-02-19)

## Coding

Models vendors and public leaderboards point to for writing and changing code.

- [Kimi K3.1 Pro](https://theagenticweekly.com/models/kimi-k3-1-pro) (Moonshot AI, 2026-10-23)
- [Kimi K3.1 Pro Highspeed](https://theagenticweekly.com/models/kimi-k3-1-pro-highspeed) (Moonshot AI, 2026-10-23)
- [MiniMax M3.1 Flash](https://theagenticweekly.com/models/minimax-m3-1-flash) (MiniMax, 2026-10-19)
- [DeepSeek-V4.1-Flash-1017](https://theagenticweekly.com/models/deepseek-v4-1-flash-1017) (DeepSeek, 2026-10-17)
- [Gemini 4 Argon](https://theagenticweekly.com/models/gemini-4-argon) (Google, 2026-10-13)
- [Muse Wonder](https://theagenticweekly.com/models/muse-wonder) (Meta, 2026-10-12)
- [Muse Wonder Full](https://theagenticweekly.com/models/muse-wonder-full) (Meta, 2026-10-12)
- [GPT-6.1 Sol](https://theagenticweekly.com/models/gpt-6-1-sol) (OpenAI, 2026-09-29)
- [Claude Sonnet 5.5](https://theagenticweekly.com/models/claude-sonnet-5-5) (Anthropic, 2026-09-28)
- [Claude Opus 5.5](https://theagenticweekly.com/models/claude-opus-5-5) (Anthropic, 2026-09-22)
- [GPT-6 Sol](https://theagenticweekly.com/models/gpt-6-sol) (OpenAI, 2026-09-22)
- [Claude Fable 5.1](https://theagenticweekly.com/models/claude-fable-5-1) (Anthropic, 2026-09-15)
- [DeepSeek-V4.1-Flash](https://theagenticweekly.com/models/deepseek-v4-1-flash) (DeepSeek, 2026-09-10)
- [GPT-6 Astra](https://theagenticweekly.com/models/gpt-6-astra) (OpenAI, 2026-09-03)
- [GLM-5.3-Flash](https://theagenticweekly.com/models/glm-5-3-flash) (Z.ai, 2026-08-26)
- [GLM-5.3](https://theagenticweekly.com/models/glm-5-3) (Z.ai, 2026-08-14)
- [Qwen3.8-27B](https://theagenticweekly.com/models/qwen3-8-27b) (Alibaba, 2026-08-14)
- [Grok 4.6](https://theagenticweekly.com/models/grok-4-6) (xAI, 2026-08-12)
- [Claude Opus 5](https://theagenticweekly.com/models/claude-opus-5) (Anthropic, 2026-07-24)
- [Gemini 3.6 Flash](https://theagenticweekly.com/models/gemini-3-6-flash) (Google, 2026-07-21)
- [Kimi K3](https://theagenticweekly.com/models/kimi-k3) (Moonshot AI, 2026-07-16)
- [GPT-5.6 Sol](https://theagenticweekly.com/models/gpt-5-6-sol) (OpenAI, 2026-07-09)
- [GPT-5.6 Terra](https://theagenticweekly.com/models/gpt-5-6-terra) (OpenAI, 2026-07-09)
- [Grok 4.5](https://theagenticweekly.com/models/grok-4-5) (xAI, 2026-07-08)
- [Claude Sonnet 5](https://theagenticweekly.com/models/claude-sonnet-5) (Anthropic, 2026-06-30)
- [GLM-5.2](https://theagenticweekly.com/models/glm-5-2) (Z.ai, 2026-06-16)
- [North Mini Code](https://theagenticweekly.com/models/north-mini-code) (Cohere, 2026-06-09)
- [Claude Opus 4.8](https://theagenticweekly.com/models/claude-opus-4-8) (Anthropic, 2026-05-28)
- [Grok Build 0.1](https://theagenticweekly.com/models/grok-build-0-1) (xAI, 2026-05-19)
- [Mistral Medium 3.5](https://theagenticweekly.com/models/mistral-medium-3-5) (Mistral AI, 2026-04-28)
- [GPT-5.5](https://theagenticweekly.com/models/gpt-5-5) (OpenAI, 2026-04-24)
- [Qwen3.6-27B](https://theagenticweekly.com/models/qwen3-6-27b) (Alibaba, 2026-04-22)
- [Claude Opus 4.7](https://theagenticweekly.com/models/claude-opus-4-7) (Anthropic, 2026-04-16)
- [GLM-5.1](https://theagenticweekly.com/models/glm-5-1) (Z.ai, 2026-04-07)
- [GPT-5.4](https://theagenticweekly.com/models/gpt-5-4) (OpenAI, 2026-03-05)
- [Gemini 3.1 Pro](https://theagenticweekly.com/models/gemini-3-1-pro) (Google, 2026-02-19)
- [Claude Sonnet 4.6](https://theagenticweekly.com/models/claude-sonnet-4-6) (Anthropic, 2026-02-17)
- [Claude Opus 4.6](https://theagenticweekly.com/models/claude-opus-4-6) (Anthropic, 2026-02-05)

## Agents

Models positioned by their vendors for long-running, tool-using work.

- [GPT-6.1 Terra](https://theagenticweekly.com/models/gpt-6-1-terra) (OpenAI, 2026-10-26)
- [Kimi K3.1 Pro](https://theagenticweekly.com/models/kimi-k3-1-pro) (Moonshot AI, 2026-10-23)
- [Kimi K3.1 Pro Highspeed](https://theagenticweekly.com/models/kimi-k3-1-pro-highspeed) (Moonshot AI, 2026-10-23)
- [Mercury 3](https://theagenticweekly.com/models/mercury-3) (Inception, 2026-10-19)
- [MiniMax M3.1 Flash](https://theagenticweekly.com/models/minimax-m3-1-flash) (MiniMax, 2026-10-19)
- [Gemini 4 Argon](https://theagenticweekly.com/models/gemini-4-argon) (Google, 2026-10-13)
- [Muse Wonder](https://theagenticweekly.com/models/muse-wonder) (Meta, 2026-10-12)
- [Muse Wonder Full](https://theagenticweekly.com/models/muse-wonder-full) (Meta, 2026-10-12)
- [Claude Haiku 5.5](https://theagenticweekly.com/models/claude-haiku-5-5) (Anthropic, 2026-10-07)
- [GPT-6.1 Sol](https://theagenticweekly.com/models/gpt-6-1-sol) (OpenAI, 2026-09-29)
- [Claude Sonnet 5.5](https://theagenticweekly.com/models/claude-sonnet-5-5) (Anthropic, 2026-09-28)
- [Claude Opus 5.5](https://theagenticweekly.com/models/claude-opus-5-5) (Anthropic, 2026-09-22)
- [GPT-6 Sol](https://theagenticweekly.com/models/gpt-6-sol) (OpenAI, 2026-09-22)
- [Grok 4.7](https://theagenticweekly.com/models/grok-4-7) (xAI, 2026-09-21)
- [Claude Fable 5.1](https://theagenticweekly.com/models/claude-fable-5-1) (Anthropic, 2026-09-15)
- [GPT-6 Astra](https://theagenticweekly.com/models/gpt-6-astra) (OpenAI, 2026-09-03)
- [Gemini 3.8 Flash](https://theagenticweekly.com/models/gemini-3-8-flash) (Google, 2026-09-02)
- [Claude Mythos 5.1](https://theagenticweekly.com/models/claude-mythos-5-1) (Anthropic, 2026-09-01)
- [GLM-5.3-Flash](https://theagenticweekly.com/models/glm-5-3-flash) (Z.ai, 2026-08-26)
- [GLM-5.3](https://theagenticweekly.com/models/glm-5-3) (Z.ai, 2026-08-14)
- [Grok 4.6](https://theagenticweekly.com/models/grok-4-6) (xAI, 2026-08-12)
- [Gemini Robotics ER 2](https://theagenticweekly.com/models/gemini-robotics-er-2) (Google, 2026-07-30)
- [Claude Opus 5](https://theagenticweekly.com/models/claude-opus-5) (Anthropic, 2026-07-24)
- [Gemini 3.5 Flash-Lite](https://theagenticweekly.com/models/gemini-3-5-flash-lite) (Google, 2026-07-21)
- [Gemini 3.6 Flash](https://theagenticweekly.com/models/gemini-3-6-flash) (Google, 2026-07-21)
- [Kimi K3](https://theagenticweekly.com/models/kimi-k3) (Moonshot AI, 2026-07-16)
- [GPT-5.6 Sol](https://theagenticweekly.com/models/gpt-5-6-sol) (OpenAI, 2026-07-09)
- [Grok 4.5](https://theagenticweekly.com/models/grok-4-5) (xAI, 2026-07-08)
- [GPT-Realtime-2.1](https://theagenticweekly.com/models/gpt-realtime-2-1) (OpenAI, 2026-07-06)
- [Claude Sonnet 5](https://theagenticweekly.com/models/claude-sonnet-5) (Anthropic, 2026-06-30)
- [GLM-5.2](https://theagenticweekly.com/models/glm-5-2) (Z.ai, 2026-06-16)
- [Claude Fable 5](https://theagenticweekly.com/models/claude-fable-5) (Anthropic, 2026-06-09)
- [Claude Mythos 5](https://theagenticweekly.com/models/claude-mythos-5) (Anthropic, 2026-06-09)
- [Claude Opus 4.8](https://theagenticweekly.com/models/claude-opus-4-8) (Anthropic, 2026-05-28)
- [Command A+](https://theagenticweekly.com/models/command-a-plus) (Cohere, 2026-05-20)
- [Grok Build 0.1](https://theagenticweekly.com/models/grok-build-0-1) (xAI, 2026-05-19)
- [Mistral Medium 3.5](https://theagenticweekly.com/models/mistral-medium-3-5) (Mistral AI, 2026-04-28)
- [GPT-5.5](https://theagenticweekly.com/models/gpt-5-5) (OpenAI, 2026-04-24)
- [Claude Opus 4.7](https://theagenticweekly.com/models/claude-opus-4-7) (Anthropic, 2026-04-16)
- [GLM-5.1](https://theagenticweekly.com/models/glm-5-1) (Z.ai, 2026-04-07)
- [GPT-5.4](https://theagenticweekly.com/models/gpt-5-4) (OpenAI, 2026-03-05)
- [Claude Sonnet 4.6](https://theagenticweekly.com/models/claude-sonnet-4-6) (Anthropic, 2026-02-17)
- [GLM-5](https://theagenticweekly.com/models/glm-5) (Z.ai, 2026-02-12)
- [Claude Opus 4.6](https://theagenticweekly.com/models/claude-opus-4-6) (Anthropic, 2026-02-05)

## Reasoning

Models positioned for complex multi-step problems.

- [Kimi K3.1 Pro](https://theagenticweekly.com/models/kimi-k3-1-pro) (Moonshot AI, 2026-10-23)
- [Gemini 4 Argon](https://theagenticweekly.com/models/gemini-4-argon) (Google, 2026-10-13)
- [GPT-6.1 Sol](https://theagenticweekly.com/models/gpt-6-1-sol) (OpenAI, 2026-09-29)
- [Claude Opus 5.5](https://theagenticweekly.com/models/claude-opus-5-5) (Anthropic, 2026-09-22)
- [GPT-6 Sol](https://theagenticweekly.com/models/gpt-6-sol) (OpenAI, 2026-09-22)
- [Grok 4.7](https://theagenticweekly.com/models/grok-4-7) (xAI, 2026-09-21)
- [Claude Fable 5.1](https://theagenticweekly.com/models/claude-fable-5-1) (Anthropic, 2026-09-15)
- [Gemini 3.8 Live Extended Thinking](https://theagenticweekly.com/models/gemini-3-8-live-extended-thinking) (Google, 2026-09-15)
- [GPT-6 Astra](https://theagenticweekly.com/models/gpt-6-astra) (OpenAI, 2026-09-03)
- [Claude Fable 5](https://theagenticweekly.com/models/claude-fable-5) (Anthropic, 2026-06-09)
- [GPT-5.5 Pro](https://theagenticweekly.com/models/gpt-5-5-pro) (OpenAI, 2026-04-24)
- [Grok 4.20](https://theagenticweekly.com/models/grok-4-20) (xAI, 2026-03-10)
- [Gemini 3.1 Pro](https://theagenticweekly.com/models/gemini-3-1-pro) (Google, 2026-02-19)
- [GLM-5](https://theagenticweekly.com/models/glm-5) (Z.ai, 2026-02-12)
- [Claude Opus 4.6](https://theagenticweekly.com/models/claude-opus-4-6) (Anthropic, 2026-02-05)

## Open weights

Weights you can download. Licence terms vary and are shown per model.

- [Nemotron 3.5 Bolt E4B (4GB)](https://theagenticweekly.com/models/nemotron-3-5-bolt-e4b) (NVIDIA, 2026-10-23)
- [MiniMax M3.1 Flash](https://theagenticweekly.com/models/minimax-m3-1-flash) (MiniMax, 2026-10-19)
- [DeepSeek-V4.1-Flash-1017](https://theagenticweekly.com/models/deepseek-v4-1-flash-1017) (DeepSeek, 2026-10-17)
- [Muse Spark Open](https://theagenticweekly.com/models/muse-spark-open) (Meta, 2026-10-15)
- [Mistral Large 4](https://theagenticweekly.com/models/mistral-large-4) (Mistral AI, 2026-10-06)
- [DeepSeek-V4.1-Flash](https://theagenticweekly.com/models/deepseek-v4-1-flash) (DeepSeek, 2026-09-10)
- [North Small Translate](https://theagenticweekly.com/models/north-small-translate) (Cohere, 2026-09-09)
- [GLM-5.3](https://theagenticweekly.com/models/glm-5-3) (Z.ai, 2026-08-14)
- [Qwen3.8-27B](https://theagenticweekly.com/models/qwen3-8-27b) (Alibaba, 2026-08-14)
- [Qwen3.8-2.4T-A95B](https://theagenticweekly.com/models/qwen3-8-2-4t-a95b) (Alibaba, 2026-08-12)
- [Kimi K3](https://theagenticweekly.com/models/kimi-k3) (Moonshot AI, 2026-07-16)
- [Cohere Transcribe Arabic](https://theagenticweekly.com/models/cohere-transcribe-arabic) (Cohere, 2026-07-07)
- [North Mini Code](https://theagenticweekly.com/models/north-mini-code) (Cohere, 2026-06-09)
- [Command A+](https://theagenticweekly.com/models/command-a-plus) (Cohere, 2026-05-20)
- [Mistral Medium 3.5](https://theagenticweekly.com/models/mistral-medium-3-5) (Mistral AI, 2026-04-28)
- [Qwen3.6-27B](https://theagenticweekly.com/models/qwen3-6-27b) (Alibaba, 2026-04-22)
- [Gemma 4](https://theagenticweekly.com/models/gemma-4) (Google, 2026-03-31)
- [Mistral Small 4](https://theagenticweekly.com/models/mistral-small-4) (Mistral AI, 2026-03-16)
- [Ministral 3 14B](https://theagenticweekly.com/models/ministral-3-14b) (Mistral AI, 2025-12-02)

## Runs on your own hardware

Open-weights models small enough for a single workstation or server.

- [Nemotron 3.5 Bolt E4B (4GB)](https://theagenticweekly.com/models/nemotron-3-5-bolt-e4b) (NVIDIA, 2026-10-23)
- [Qwen3.8-27B](https://theagenticweekly.com/models/qwen3-8-27b) (Alibaba, 2026-08-14)
- [North Mini Code](https://theagenticweekly.com/models/north-mini-code) (Cohere, 2026-06-09)
- [Qwen3.6-27B](https://theagenticweekly.com/models/qwen3-6-27b) (Alibaba, 2026-04-22)
- [Gemma 4](https://theagenticweekly.com/models/gemma-4) (Google, 2026-03-31)
- [Mistral Small 4](https://theagenticweekly.com/models/mistral-small-4) (Mistral AI, 2026-03-16)
- [Ministral 3 14B](https://theagenticweekly.com/models/ministral-3-14b) (Mistral AI, 2025-12-02)

## Small and fast

Low-cost, low-latency models for high-volume work.

- [GPT-6.1 Terra](https://theagenticweekly.com/models/gpt-6-1-terra) (OpenAI, 2026-10-26)
- [Nemotron 3.5 Bolt E4B (4GB)](https://theagenticweekly.com/models/nemotron-3-5-bolt-e4b) (NVIDIA, 2026-10-23)
- [Mercury 3](https://theagenticweekly.com/models/mercury-3) (Inception, 2026-10-19)
- [MiniMax M3.1 Flash](https://theagenticweekly.com/models/minimax-m3-1-flash) (MiniMax, 2026-10-19)
- [DeepSeek-V4.1-Flash-1017](https://theagenticweekly.com/models/deepseek-v4-1-flash-1017) (DeepSeek, 2026-10-17)
- [Claude Haiku 5.5](https://theagenticweekly.com/models/claude-haiku-5-5) (Anthropic, 2026-10-07)
- [Nano Banana 2.1](https://theagenticweekly.com/models/nano-banana-2-1) (Google, 2026-10-06)
- [Gemini 3.8 Flash-Lite TTS](https://theagenticweekly.com/models/gemini-3-8-flash-lite-tts) (Google, 2026-09-22)
- [GPT-6 Luna](https://theagenticweekly.com/models/gpt-6-luna) (OpenAI, 2026-09-22)
- [DeepSeek-V4.1-Flash](https://theagenticweekly.com/models/deepseek-v4-1-flash) (DeepSeek, 2026-09-10)
- [GPT-Image-2.5 Flare](https://theagenticweekly.com/models/gpt-image-2-5-flare) (OpenAI, 2026-09-08)
- [Gemini 3.8 Flash](https://theagenticweekly.com/models/gemini-3-8-flash) (Google, 2026-09-02)
- [Gemini 3.5 Flash-Lite](https://theagenticweekly.com/models/gemini-3-5-flash-lite) (Google, 2026-07-21)
- [Gemini 3.6 Flash](https://theagenticweekly.com/models/gemini-3-6-flash) (Google, 2026-07-21)
- [GPT-5.6 Luna](https://theagenticweekly.com/models/gpt-5-6-luna) (OpenAI, 2026-07-09)
- [GPT-Realtime-2.1 mini](https://theagenticweekly.com/models/gpt-realtime-2-1-mini) (OpenAI, 2026-07-06)
- [Nano Banana 2 Lite](https://theagenticweekly.com/models/nano-banana-2-lite) (Google, 2026-06-30)
- [Gemini 3.1 Flash-Lite](https://theagenticweekly.com/models/gemini-3-1-flash-lite) (Google, 2026-05-07)
- [Gemma 4](https://theagenticweekly.com/models/gemma-4) (Google, 2026-03-31)
- [GPT-5.4 mini](https://theagenticweekly.com/models/gpt-5-4-mini) (OpenAI, 2026-03-17)
- [Mistral Small 4](https://theagenticweekly.com/models/mistral-small-4) (Mistral AI, 2026-03-16)
- [Ministral 3 14B](https://theagenticweekly.com/models/ministral-3-14b) (Mistral AI, 2025-12-02)

## Image

Image generation and editing models.

- [Nano Banana 2.1](https://theagenticweekly.com/models/nano-banana-2-1) (Google, 2026-10-06)
- [GPT-Image-2.5 Flare](https://theagenticweekly.com/models/gpt-image-2-5-flare) (OpenAI, 2026-09-08)
- [GPT-Image-2.5 Sunburst](https://theagenticweekly.com/models/gpt-image-2-5-sunburst) (OpenAI, 2026-09-08)
- [Nano Banana 2 Lite](https://theagenticweekly.com/models/nano-banana-2-lite) (Google, 2026-06-30)
- [Nano Banana Pro](https://theagenticweekly.com/models/nano-banana-pro) (Google, 2026-05-28)
- [GPT-Image-2](https://theagenticweekly.com/models/gpt-image-2) (OpenAI, 2026-04-21)
- [GLM-Image](https://theagenticweekly.com/models/glm-image) (Z.ai, 2026-01-14)

## Voice

Real-time voice and text-to-speech models.

- [Voxtral 3](https://theagenticweekly.com/models/voxtral-3) (Mistral AI, 2026-10-20)
- [Grok Voice 3.0](https://theagenticweekly.com/models/grok-voice-3-0) (xAI, 2026-10-14)
- [Gemini 3.8 Flash TTS](https://theagenticweekly.com/models/gemini-3-8-flash-tts) (Google, 2026-09-22)
- [Gemini 3.8 Flash-Lite TTS](https://theagenticweekly.com/models/gemini-3-8-flash-lite-tts) (Google, 2026-09-22)
- [Grok Voice Transcribe 2.0](https://theagenticweekly.com/models/grok-voice-transcribe-2-0) (xAI, 2026-09-17)
- [Gemini 3.8 Live](https://theagenticweekly.com/models/gemini-3-8-live) (Google, 2026-09-15)
- [Gemini 3.8 Live Extended Thinking](https://theagenticweekly.com/models/gemini-3-8-live-extended-thinking) (Google, 2026-09-15)
- [GPT-Live 1](https://theagenticweekly.com/models/gpt-live-1) (OpenAI, 2026-09-10)
- [Gemini 3.5 Transcribe](https://theagenticweekly.com/models/gemini-3-5-transcribe) (Google, 2026-08-26)
- [GPT-Live-Transcribe](https://theagenticweekly.com/models/gpt-live-transcribe) (OpenAI, 2026-07-28)
- [GPT-Transcribe](https://theagenticweekly.com/models/gpt-transcribe) (OpenAI, 2026-07-28)
- [Cohere Transcribe Arabic](https://theagenticweekly.com/models/cohere-transcribe-arabic) (Cohere, 2026-07-07)
- [GPT-Realtime-2.1](https://theagenticweekly.com/models/gpt-realtime-2-1) (OpenAI, 2026-07-06)
- [GPT-Realtime-2.1 mini](https://theagenticweekly.com/models/gpt-realtime-2-1-mini) (OpenAI, 2026-07-06)
- [Cohere Transcribe](https://theagenticweekly.com/models/cohere-transcribe) (Cohere, 2026-03-26)
- [Voxtral TTS](https://theagenticweekly.com/models/voxtral-tts) (Mistral AI, 2026-03-23)

## Video

Video generation and editing models.

- [Gemini Omni 1.1 Flash](https://theagenticweekly.com/models/gemini-omni-1-1-flash) (Google, 2026-08-27)

## How we label facts

Reported means read from the cited vendor page on the verified date; we did not run the model. Tested means we ran it ourselves; no model on this page is Tested yet. Facts older than 14 days are flagged for re-checking. Data licence CC BY 4.0.


# GPT-6.1 Terra

GPT-6.1 Terra is OpenAI's latest language model in our dataset, released Oct 26, 2026. Proprietary, API access only. OpenAI's fast, lower-cost GPT-6.1 model for visual reasoning, computer use and writing; the successor to GPT-5.6 Terra, priced at half of GPT-6.1 Sol. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-6-1-terra · JSON: https://theagenticweekly.com/api/models/gpt-6-1-terra

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Oct 26, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-6.1-terra` | https://developers.openai.com/api/docs/models/gpt-6.1-terra |
| Context window | 1,050,000 tokens (1.05M). Maximum input is 922,000 tokens. | https://developers.openai.com/api/docs/models/gpt-6.1-terra |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-6.1-terra |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-6.1-terra |
| Output | text | https://developers.openai.com/api/docs/models/gpt-6.1-terra |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-6.1-terra |
| Pricing | $1 in / $5 out per 1M tokens. Standard processing. Fast mode is 2x; Batch and Flex are 50% lower. | https://developers.openai.com/api/docs/models/gpt-6.1-terra |

## Vendor claims and neutral signals

- Vendor claim (OpenAI): on Agents' Last Exam at medium effort, 15 percentage points above GPT-5.6 Terra at about one-third of the cost. We have not independently measured this.
- At launch it was available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and in the API.

## Notes

- Knowledge cutoff listed: Jun 01, 2026.
- Reasoning effort supports none, low, medium (default), high, xhigh and max. GPT-6.1 Sol and GPT-6 Astra do not support none.
- Tool calling requires the Responses API. Fine-tuning is not supported. Audio and video input are not supported.
- Listed as meeting the Critical threshold in cybersecurity under OpenAI's Preparedness Framework.

## What we tested

Nothing yet. We have not run GPT-6.1 Terra ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Benchmark scores: OpenAI's launch post shows Agents' Last Exam, OSWorld 2.1 and MMMU-Pro results as charts whose values are not in the page text, so none are copied here. The text gives Terminal-Bench 4.0 at 50.3% (max effort), which is OpenAI's own figure.
- Speed: OpenAI says Terra completes tasks over three times faster than GPT-6 Astra and GPT-6.1 Sol; we found no independent latency measurement, so no speed figure is listed.

## Sources

- [OpenAI model page](https://developers.openai.com/api/docs/models/gpt-6.1-terra)
- [OpenAI API pricing](https://developers.openai.com/api/docs/pricing)
- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI GPT-6 guide](https://developers.openai.com/api/docs/guides/latest-model)
- [Agentic Weekly brief on the launch](https://theagenticweekly.com/news/openai-gpt-6-1-terra-computer-use)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Kimi K3.1 Pro

Kimi K3.1 Pro is Moonshot AI's latest language model in our dataset, released Oct 23, 2026. Open weights are promised but not yet published. Moonshot's post-trained successor to Kimi K3, a 3T-class model with a 1M-token context, live in Kimi apps and the API on Oct 23, 2026; open weights are promised by Nov 4. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/kimi-k3-1-pro · JSON: https://theagenticweekly.com/api/models/kimi-k3-1-pro

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Moonshot AI |  |
| Released | Oct 23, 2026 | https://vercel.com/changelog/kimi-k3-1-pro-and-kimi-k3-1-pro-highspeed-are-available-now-in-ai-gateway |
| API id | `kimi-k3.1-pro` | https://www.kimi.ai/blog/kimi-k3-1-pro |
| Context window | 1,048,576 tokens (1.05M). Moonshot states 1M tokens; the pricing page gives 1,048,576. | https://www.kimi.ai/resources/kimi-k3-1-pro-pricing |
| Parameters | 2.8T total, 104B active, mixture-of-experts | https://artificialanalysis.ai/articles/kimi-k3-1-pro-launch-benchmarks |
| Licence | Not yet published. Moonshot calls it an open-weight model and says the full weights will be released by Nov 4, 2026. Until the weights and license are published we do not label it open-weights. | https://www.kimi.ai/blog/kimi-k3-1-pro |
| Pricing | $2 in / $10 out per 1M tokens. Highspeed variant: $4.00 input, $20.00 output, $0.20 cached input. Rates checked by Moonshot on Oct 24, 2026. | https://www.kimi.ai/resources/kimi-k3-1-pro-pricing |

## Vendor claims and neutral signals

- Moonshot reports 65.7% on Terminal-Bench 4.0 (Claude Code harness), against 66.4% for Claude Opus 5.5 and 58.2% for GPT-6 Astra. This is a vendor-run figure.

## Notes

- Artificial Analysis scored it 55 on its Intelligence Index at launch (4th, behind Claude Opus 5.5 at 58, Claude Sonnet 5.5 at 56 and Muse Wonder Full at 56) and 67% on Terminal-Bench 4.0 in its own run, which differs from Moonshot's 65.7%.
- The faster tier, Kimi K3.1 Pro Highspeed, has its own page: /models/kimi-k3-1-pro-highspeed.

## What we tested

Nothing yet. We have not run Kimi K3.1 Pro ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Input and output modalities were not confirmed on Moonshot's own pages.
- Weights and license are not published yet; Moonshot says they arrive by Nov 4, 2026.

## Sources

- [Kimi K3.1 Pro blog](https://www.kimi.ai/blog/kimi-k3-1-pro)
- [Vercel AI Gateway changelog](https://vercel.com/changelog/kimi-k3-1-pro-and-kimi-k3-1-pro-highspeed-are-available-now-in-ai-gateway)
- [Kimi K3.1 Pro pricing page](https://www.kimi.ai/resources/kimi-k3-1-pro-pricing)
- [Artificial Analysis launch benchmarks](https://artificialanalysis.ai/articles/kimi-k3-1-pro-launch-benchmarks)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Kimi K3.1 Pro Highspeed

Kimi K3.1 Pro Highspeed is a Moonshot AI language model released Oct 23, 2026; Moonshot AI's newest language model in our dataset is Kimi K3.1 Pro. The faster serving tier of Kimi K3.1 Pro: Moonshot says it delivers 2 to 3 times the output speed for interactive coding and agents, at twice the per-token price, launched alongside K3.1 Pro on Oct 23, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/kimi-k3-1-pro-highspeed · JSON: https://theagenticweekly.com/api/models/kimi-k3-1-pro-highspeed

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Moonshot AI |  |
| Released | Oct 23, 2026 | https://vercel.com/changelog/kimi-k3-1-pro-and-kimi-k3-1-pro-highspeed-are-available-now-in-ai-gateway |
| API id | `kimi-k3.1-pro-highspeed` | https://www.kimi.ai/blog/kimi-k3-1-pro |
| Context window | 1,048,576 tokens (1.05M). Both K3.1 Pro models support a 1,048,576-token context window per Moonshot's pricing page. | https://www.kimi.ai/resources/kimi-k3-1-pro-pricing |
| Pricing | $4 in / $20 out per 1M tokens. Twice the standard K3.1 Pro input and output rate; the cached-input rate is the same. Rates checked by Moonshot on Oct 24, 2026. | https://www.kimi.ai/resources/kimi-k3-1-pro-pricing |

## Notes

- Standard Kimi K3.1 Pro: see /models/kimi-k3-1-pro. The weights Moonshot promised by Nov 4, 2026 are for K3.1 Pro; we have not verified a license for Highspeed.

## What we tested

Nothing yet. We have not run Kimi K3.1 Pro Highspeed ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Moonshot has not published separate benchmark results for Highspeed; it describes it as the same model as K3.1 Pro served 2 to 3 times faster.
- Input and output modalities were not confirmed on Moonshot's own pages.

## Sources

- [Kimi K3.1 Pro blog](https://www.kimi.ai/blog/kimi-k3-1-pro)
- [Kimi K3.1 Pro pricing page](https://www.kimi.ai/resources/kimi-k3-1-pro-pricing)
- [Vercel AI Gateway changelog](https://vercel.com/changelog/kimi-k3-1-pro-and-kimi-k3-1-pro-highspeed-are-available-now-in-ai-gateway)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Nemotron 3.5 Bolt E4B (4GB)

Nemotron 3.5 Bolt E4B (4GB) is NVIDIA's latest language model in our dataset, released Oct 23, 2026. Open weights (OpenMDW-1.1). NVIDIA's compact open-weights text model built on Gemma-4-E4B, quantized to roughly 4GB (Q4_K_M) for laptops and edge devices; released Oct 23, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/nemotron-3-5-bolt-e4b · JSON: https://theagenticweekly.com/api/models/nemotron-3-5-bolt-e4b

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | NVIDIA |  |
| Released | Oct 23, 2026 | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB |
| Context window | 64,000 tokens (64K). Stated as 64K tokens. | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB |
| Input | text | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB |
| Output | text | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB |
| Parameters | 8B total, 4B effective active, Gemma-4-E4B base | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB |
| Licence | OpenMDW-1.1 (open weights) | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB |

## Notes

- A BF16 variant also exists on Hugging Face. The 4GB build is the Q4_K_M quantization.

## What we tested

Nothing yet. We have not run Nemotron 3.5 Bolt E4B (4GB) ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Benchmark scores were not captured.
- Larger Nemotron 3.5 models announced by NVIDIA on Oct 15 (Super VL, Bolt 74B-A9B) are not tracked here yet.

## Sources

- [NVIDIA Research model page](https://research.nvidia.com/labs/nemotron/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB/)
- [Hugging Face model card](https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB)
- [NVIDIA blog: Nemotron 3.5](https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Voxtral 3

Voxtral 3 is Mistral AI's latest voice model in our dataset (public preview), released Oct 20, 2026. Mistral's third-generation speech family, in public preview since Oct 20, 2026: Mini Transcribe, Small Transcribe, Mini Transcribe Realtime and Mini TTS, plus a voice agent SDK. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/voxtral-3 · JSON: https://theagenticweekly.com/api/models/voxtral-3

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Mistral AI |  |
| Released | Oct 20, 2026 | https://docs.mistral.ai/models/voxtral-mini-tts-3-0.md |
| API id | `voxtral-mini-tts-3-0, voxtral-mini-transcribe-3-0, voxtral-mini-transcribe-realtime-3-0, voxtral-small-transcribe-3-0` | https://docs.mistral.ai/models/voxtral-mini-tts-3-0.md |
| Pricing | Mini Transcribe 3, per minute of audio $0.0035; Small Transcribe 3, per minute of audio $0.004; Mini Transcribe Realtime 3, per minute of audio $0.008; Mini TTS 3, per 1M characters $13; Voice agent API with Mistral Small 4 for reasoning, per minute $0.04; Voice agent API with a larger model, per minute $0.08 (see each line). Prices are from Mistral's announcement table. | https://mistral.ai/news/voxtral-3/ |

## Vendor claims and neutral signals

- Mistral says Voxtral Mini Transcribe 3 ranks 3rd on the FLEURS multilingual transcription and English diarization benchmarks. This is a vendor claim.

## Notes

- Mistral says Voxtral Mini Transcribe 3 has 4B parameters, and that Small Transcribe 3 outperforms Whisper, Scribe v2 and Gemini 3.5 Transcribe on its own evaluations.

## What we tested

Nothing yet. We have not run Voxtral 3 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Per-model context windows and license terms were not captured; Mistral offers API access (EU, US, global endpoints) and self-deployment, but we did not verify the terms for either.
- Mistral also shipped a Voice Agent SDK (built on LiveKit) with the launch, plus Pro Voice adaptation; we list prices only.

## Sources

- [Mistral Voxtral 3 announcement](https://mistral.ai/news/voxtral-3/)
- [Mistral docs: Voxtral TTS 3](https://docs.mistral.ai/models/voxtral-mini-tts-3-0.md)
- [Mistral docs: Voxtral Mini Transcribe 3](https://docs.mistral.ai/models/voxtral-mini-transcribe-3-0.md)
- [Mistral changelog](https://docs.mistral.ai/resources/changelogs)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Mercury 3

Mercury 3 is Inception's latest language model in our dataset, released Oct 19, 2026. Proprietary, API access only. Inception's diffusion-based language model with a 1M-token context, launched Oct 19, 2026 and billed at launch for half its list price. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/mercury-3 · JSON: https://theagenticweekly.com/api/models/mercury-3

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Inception |  |
| Released | Oct 19, 2026 | https://www.inceptionlabs.ai/blog/introducing-mercury-3 |
| Context window | 1,000,000 tokens (1M). Stated as 1M tokens, up from 260K for Mercury 2.5. | https://www.inceptionlabs.ai/blog/introducing-mercury-3 |
| Licence | Proprietary (API). Available through the Inception API Platform, OpenRouter, Baseten and Vercel; no weights are offered. | https://www.inceptionlabs.ai/blog/introducing-mercury-3 |
| Pricing | $0.1 in / $0.35 out per 1M tokens. Launch price, 50% off the list price of $0.20 input and $0.70 output. Cached tokens are 90% off the input price. | https://www.inceptionlabs.ai/blog/introducing-mercury-3 |

## Vendor claims and neutral signals

- Inception says Mercury 3 outperforms GPT-6 Luna and Claude Haiku 5.5 on its benchmarks while running 4-8x faster at 1,068 tokens per second on NVIDIA GPUs. These are vendor claims.

## What we tested

Nothing yet. We have not run Mercury 3 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Parameter count was not disclosed.
- Launch pricing has no stated end date.
- OpenRouter's listing shows Oct 16, 2026 (early access); we use Inception's own Oct 19 launch date.

## Sources

- [Inception blog: Introducing Mercury 3](https://www.inceptionlabs.ai/blog/introducing-mercury-3)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# MiniMax M3.1 Flash

MiniMax M3.1 Flash is MiniMax's latest language model in our dataset, released Oct 19, 2026. Open weights (MiniMax Community License). MiniMax's low-cost agentic and coding model, announced Oct 19, 2026 at $0.15 input and $0.60 output per 1M tokens, with a public Hugging Face repository under a custom community license. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/minimax-m3-1-flash · JSON: https://theagenticweekly.com/api/models/minimax-m3-1-flash

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | MiniMax |  |
| Released | Oct 19, 2026 | https://www.minimax.io/blog/minimax-m31f |
| API id | `MiniMax-M3.1-Flash` | https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash |
| Context window | 1,000,000 tokens (1M). Stated as 1M tokens. | https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash |
| Input | text, image, video | https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash |
| Output | text | https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash |
| Parameters | ~428B total, 23B active, mixture-of-experts | https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash |
| Licence | MiniMax Community License (open weights). Free for non-commercial use. Commercial use needs a visible "Built with MiniMax M3.1-Flash" credit and a notice to MiniMax; products above US$20M yearly revenue need prior written authorization. | https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash/blob/main/LICENSE |
| Pricing | $0.15 in / $0.6 out per 1M tokens. Flat pricing. | https://www.minimax.io/blog/minimax-m31f |

## Vendor claims and neutral signals

- MiniMax reports 73.7% on DeepSWE v1.1, 76.2 on BankerToolBench, 61.7 on OSWorld V2 and 82% on CyberGym. These are vendor-run figures.

## What we tested

Nothing yet. We have not run MiniMax M3.1 Flash ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window and modalities come from the Hugging Face card and were not on the blog post.
- MiniMax's blog says weights "will be released later", yet a Hugging Face repository with weights is public; we follow the repository.

## Sources

- [MiniMax blog](https://www.minimax.io/blog/minimax-m31f)
- [Hugging Face model card](https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash)
- [Hugging Face LICENSE](https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash/blob/main/LICENSE)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# DeepSeek-V4.1-Flash-1017

DeepSeek-V4.1-Flash-1017 is DeepSeek's latest language model in our dataset, released Oct 17, 2026. Open weights (MIT). DeepSeek's Oct 17, 2026 update to V4.1 Flash: MIT-licensed open weights, now served behind the deepseek-flash API name, with shorter outputs and faster task completion than the Sep 10 version. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/deepseek-v4-1-flash-1017 · JSON: https://theagenticweekly.com/api/models/deepseek-v4-1-flash-1017

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | DeepSeek |  |
| Released | Oct 17, 2026 | https://api-docs.deepseek.com/news/news261017/ |
| API id | `deepseek-flash` | https://api-docs.deepseek.com/updates/ |
| Context window | 1,000,000 tokens (1M) | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Max output | 384,000 tokens | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Input | text, image | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Output | text | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Licence | MIT (open weights). License as listed on the Hugging Face model repository. | https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017 |
| Pricing | ¥1 in / ¥4 out per 1M tokens; peak: ¥2 / ¥8. Official price list is in CNY. Off-peak shown; peak (Beijing time Mon-Fri 09-12 and 14-18) is double. DeepSeek's Oct 17 release post lists no price change. We do not convert currencies. | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |

## Vendor claims and neutral signals

- Vendor-reported (DeepSeek, Hugging Face card): Terminal-Bench 4.0 42.3 and DeepSWE v1.1 77.0 for 1017, against 31.2 and 74.2 for the Sep 10 V4.1-Flash. These are vendor-run figures we did not verify.

## Notes

- This is the same product line as DeepSeek-V4.1-Flash (Sep 10, 2026), which DeepSeek has replaced behind the deepseek-flash API name; we keep the earlier page for history.

## What we tested

Nothing yet. We have not run DeepSeek-V4.1-Flash-1017 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- DeepSeek says the 1017 update did not change API pricing; we reuse the price list rather than a separate 1017 price page. The official list is in CNY and we do not convert currencies.

## Sources

- [DeepSeek-V4.1-Flash-1017 release post](https://api-docs.deepseek.com/news/news261017/)
- [DeepSeek API updates](https://api-docs.deepseek.com/updates/)
- [DeepSeek pricing (zh-CN)](https://api-docs.deepseek.com/zh-cn/quick_start/pricing)
- [Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Muse Spark Open

Muse Spark Open is Meta's latest language model in our dataset, released Oct 15, 2026. Open weights (Muse Model License (limited commercial terms)). Meta's open-weights 816B-parameter mixture-of-experts model (42B active), released Oct 15, 2026 under the Muse Model License, which has limited commercial terms. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/muse-spark-open · JSON: https://theagenticweekly.com/api/models/muse-spark-open

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Meta |  |
| Released | Oct 15, 2026 | https://research.meta.ai/blog/muse-spark-open |
| Context window | 1,000,000 tokens (1M) | https://research.meta.ai/blog/muse-spark-open |
| Input | text, image, video, audio | https://research.meta.ai/blog/muse-spark-open |
| Output | text | https://research.meta.ai/blog/muse-spark-open |
| Parameters | 816B total, 42B active, mixture-of-experts, 256 experts, 8 active | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| Licence | Muse Model License (limited commercial terms) (open weights). Not an OSI-approved open-source license; read the terms before commercial use. | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |

## Vendor claims and neutral signals

- Needs about 700 GB of VRAM for the NVFP4 checkpoint and about 2 TB for BF16, per Meta; it is not a laptop model.

## Notes

- Meta lists hosted access at Baseten, Fireworks, Groq, Modal, Nebius, OpenRouter and Together; prices are set by those providers.

## What we tested

Nothing yet. We have not run Muse Spark Open ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Meta's own API price: not stated by Meta. Meta's proprietary Muse Spark 1.x models appear on public leaderboards but we did not verify their specifications; Muse Wonder is listed separately.

## Sources

- [Meta research blog](https://research.meta.ai/blog/muse-spark-open)
- [Hugging Face model card](https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Grok Voice 3.0

Grok Voice 3.0 is xAI's latest voice model in our dataset, released Oct 14, 2026. Proprietary, API access only. xAI's voice model, released Oct 14, 2026 at the same price as Grok Voice 2.0. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/grok-voice-3-0 · JSON: https://theagenticweekly.com/api/models/grok-voice-3-0

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | xAI |  |
| Released | Oct 14, 2026 | https://docs.x.ai/developers/release-notes |
| API id | `grok-voice-3.0` | https://docs.x.ai/developers/release-notes |
| Licence | Proprietary (API) | https://docs.x.ai/developers/release-notes |

## What we tested

Nothing yet. We have not run Grok Voice 3.0 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price: xAI states only "same price as Voice 2.0", and we did not capture that figure. Modalities were not captured.

## Sources

- [xAI release notes](https://docs.x.ai/developers/release-notes)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 4 Argon

Gemini 4 Argon is Google's latest language model in our dataset, released Oct 13, 2026. Proprietary, API access only. Google's newest Gemini model: a multimodal frontier model that Google claims is #1 on the Vals Index, and that sits at #1 on the LMArena Text Arena snapshot we cite. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-4-argon · JSON: https://theagenticweekly.com/api/models/gemini-4-argon

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Oct 13, 2026 | https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/ |
| API id | `gemini-4-argon` | https://ai.google.dev/gemini-api/docs/models/gemini-4-argon |
| Context window | 1,048,576 tokens (1.05M). Unified input + output token limit. | https://ai.google.dev/gemini-api/docs/models/gemini-4-argon |
| Max output | 65,536 tokens. Default; Google lists up to 1,048,576 with long decode. | https://ai.google.dev/gemini-api/docs/models/gemini-4-argon |
| Input | text, image, video, audio, pdf | https://ai.google.dev/gemini-api/docs/models/gemini-4-argon |
| Output | text | https://ai.google.dev/gemini-api/docs/models/gemini-4-argon |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/pricing |
| Pricing | $2 in / $10 out per 1M tokens; from Dec 1, 2026: $4 / $20. Launch pricing runs through Nov 30, 2026, then $4 / $20. Not on the free tier. | https://ai.google.dev/gemini-api/docs/pricing |

## Vendor claims and neutral signals

- Vendor claim (Google): #1 on the Vals Index. We did not verify this.
- Neutral signal: #1 on LMArena Text Arena in the Oct 8, 2026 snapshot (score marked preliminary).

## What we tested

Nothing yet. We have not run Gemini 4 Argon ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API model page](https://ai.google.dev/gemini-api/docs/models/gemini-4-argon)
- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Google launch blog](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/)
- [LMArena Text Arena](https://arena.ai/leaderboard/text)
- [Google announcement (Sep 30, limited access)](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Muse Wonder

Muse Wonder is a Meta language model (public preview) released Oct 12, 2026; Meta's newest language model in our dataset is Muse Spark Open. Proprietary, API access only. Meta's most capable hosted model for long-running agent and coding work, launched Oct 12, 2026 in preview on the Meta Model API and in Muse Code. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/muse-wonder · JSON: https://theagenticweekly.com/api/models/muse-wonder

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Meta |  |
| Released | Oct 12, 2026 | https://about.fb.com/news/2026/10/introducing-muse-wonder-models/ |
| API id | `muse-wonder-1.0` | https://dev.meta.ai/docs/pricing-rate-limits |
| Context window | 1,048,576 tokens (1.05M) | https://dev.meta.ai/docs/models |
| Input | text, image, video, audio, pdf | https://dev.meta.ai/docs/models |
| Output | text | https://dev.meta.ai/docs/models |
| Licence | Proprietary (API). Hosted on the Meta Model API and in Muse Code; no weights. | https://dev.meta.ai/docs/models |
| Pricing | $4 in / $20 out per 1M tokens. Standard mode only; no Contributor-mode discount for Muse Wonder. | https://dev.meta.ai/docs/pricing-rate-limits |

## Vendor claims and neutral signals

- Meta calls Muse Wonder and Muse Wonder Full its most capable models to date, aimed at agentic and coding work.

## Notes

- Meta's docs describe Muse Wonder as "in preview". It is not available in South Korea.

## What we tested

Nothing yet. We have not run Muse Wonder ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Parameter count and benchmark scores were not captured.
- Meta says a full safety report "will be available soon".

## Sources

- [Meta announcement](https://about.fb.com/news/2026/10/introducing-muse-wonder-models/)
- [Meta Model API: models](https://dev.meta.ai/docs/models)
- [Meta Model API: pricing and rate limits](https://dev.meta.ai/docs/pricing-rate-limits)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Muse Wonder Full

Muse Wonder Full is a Meta language model (limited access) released Oct 12, 2026; Meta's newest language model in our dataset is Muse Spark Open. Proprietary, API access only. The higher-capability Muse Wonder variant, available since Oct 12, 2026 only inside Muse Code on the Power Usage and Maximum Usage subscription plans. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/muse-wonder-full · JSON: https://theagenticweekly.com/api/models/muse-wonder-full

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Meta |  |
| Released | Oct 12, 2026 | https://about.fb.com/news/2026/10/introducing-muse-wonder-models/ |
| Licence | Proprietary (subscription). Not on the Model API; Muse Code subscribers only. | https://dev.meta.ai/docs/models |

## Notes

- Setting reasoning effort to "max" on Muse Wonder does not select Muse Wonder Full, per Meta's docs.
- Artificial Analysis placed Muse Wonder Full at 56 on its Intelligence Index on Oct 23, 2026 (https://artificialanalysis.ai/articles/kimi-k3-1-pro-launch-benchmarks).

## What we tested

Nothing yet. We have not run Muse Wonder Full ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window, API pricing and benchmark scores were not published; the variant is not on the Model API.

## Sources

- [Meta announcement](https://about.fb.com/news/2026/10/introducing-muse-wonder-models/)
- [Meta Model API: models](https://dev.meta.ai/docs/models)
- [Artificial Analysis launch benchmarks](https://artificialanalysis.ai/articles/kimi-k3-1-pro-launch-benchmarks)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Haiku 5.5

Claude Haiku 5.5 is Anthropic's latest language model in our dataset, released Oct 7, 2026. Proprietary, API access only. Anthropic's fastest, smallest current Claude model, built for high-volume work like summarization, subagents and browser use. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-haiku-5-5 · JSON: https://theagenticweekly.com/api/models/claude-haiku-5-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Oct 7, 2026 | https://www.anthropic.com/news |
| API id | `claude-haiku-5-5` | https://platform.claude.com/docs/en/about-claude/models/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Input | text, image | https://platform.claude.com/docs/en/about-claude/models/overview |
| Output | text | https://platform.claude.com/docs/en/about-claude/models/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $0.1 in / $0.5 out per 1M tokens. Listed by Anthropic as "from" $0.10 / $0.50. | https://platform.claude.com/docs/en/about-claude/models/overview |

## What we tested

Nothing yet. We have not run Claude Haiku 5.5 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Maximum output tokens for Haiku 5.5 were not captured.

## Sources

- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)
- [Anthropic newsroom](https://www.anthropic.com/news)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Mistral Large 4

Mistral Large 4 is Mistral AI's latest language model in our dataset (public preview), released Oct 6, 2026. Open weights are promised but not yet published. Mistral's 1.05T-parameter multimodal mixture-of-experts model, in public preview via API, with open weights promised by the end of October 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/mistral-large-4 · JSON: https://theagenticweekly.com/api/models/mistral-large-4

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Mistral AI |  |
| Released | Oct 6, 2026 | https://mistral.ai/news/mistral-large-4/ |
| API id | `mistral-large-4` | https://docs.mistral.ai/models/mistral-large-4 |
| Context window | 1,000,000 tokens (1M) | https://docs.mistral.ai/models/mistral-large-4 |
| Parameters | 1.05T total, 52B active, mixture-of-experts | https://docs.mistral.ai/models/mistral-large-4 |
| Licence | Not yet published. Mistral says weights "drop end of this month" (October 2026). Until the weights and license are published we do not label it open-weights. | https://mistral.ai/news/mistral-large-4/ |
| Pricing | $1.36 in / $4.18 out per 1M tokens. Preview pricing as shown on Mistral's docs page. | https://docs.mistral.ai/models/mistral-large-4 |

## Vendor claims and neutral signals

- Vendor claim (Mistral): state-of-the-art among open models on cybersecurity, finance and law. Not independently verified by us.

## What we tested

Nothing yet. We have not run Mistral Large 4 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Output modalities and maximum output tokens were not captured. Recheck once the weights ship.

## Sources

- [Mistral model page](https://docs.mistral.ai/models/mistral-large-4)
- [Mistral announcement](https://mistral.ai/news/mistral-large-4/)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Nano Banana 2.1

Nano Banana 2.1 is Google's latest image-generation model in our dataset, released Oct 6, 2026. Proprietary, API access only. Google's high-efficiency Gemini image generation model (API id gemini-nano-banana-2.1), generally available since Oct 6, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/nano-banana-2-1 · JSON: https://theagenticweekly.com/api/models/nano-banana-2-1

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Oct 6, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-nano-banana-2.1` | https://ai.google.dev/gemini-api/docs/models/gemini-nano-banana-2.1 |
| Output | image | https://ai.google.dev/gemini-api/docs/models/gemini-nano-banana-2.1 |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/pricing |
| Pricing | 1K image $0.0336; 2K image $0.0504; 4K image $0.113 (per image) | https://ai.google.dev/gemini-api/docs/pricing |

## What we tested

Nothing yet. We have not run Nano Banana 2.1 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Input modalities were not captured. It ranks #6 (preliminary) on the Oct 7, 2026 LMArena text-to-image snapshot.

## Sources

- [Gemini API model page](https://ai.google.dev/gemini-api/docs/models/gemini-nano-banana-2.1)
- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-6.1 Sol

GPT-6.1 Sol is an OpenAI language model released Sep 29, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's near-flagship GPT-6.1 model: OpenAI says it performs close to GPT-6 Astra at a fifth of Astra's price. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-6-1-sol · JSON: https://theagenticweekly.com/api/models/gpt-6-1-sol

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Sep 29, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-6.1-sol` | https://developers.openai.com/api/docs/models/gpt-6.1-sol |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-6.1-sol |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-6.1-sol |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-6.1-sol |
| Output | text | https://developers.openai.com/api/docs/models/gpt-6.1-sol |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-6.1-sol |
| Pricing | $2 in / $10 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-6.1-sol |

## Vendor claims and neutral signals

- Vendor claim (OpenAI): near-Astra performance at one fifth of Astra's price. We have not independently measured this.

## Notes

- At launch it was available in ChatGPT Work and Codex, not yet in ChatGPT Chat. Knowledge cutoff listed: Apr 30, 2026.

## What we tested

Nothing yet. We have not run GPT-6.1 Sol ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI model page](https://developers.openai.com/api/docs/models/gpt-6.1-sol)
- [OpenAI announcement](https://openai.com/index/introducing-gpt-6-1-sol/)
- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Sonnet 5.5

Claude Sonnet 5.5 is an Anthropic language model released Sep 28, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. The mid-priced Claude model: Anthropic says it runs 30%+ faster than Sonnet 5 and costs up to 30% less for most work. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-sonnet-5-5 · JSON: https://theagenticweekly.com/api/models/claude-sonnet-5-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Sep 28, 2026 | https://www.anthropic.com/news |
| API id | `claude-sonnet-5-5` | https://platform.claude.com/docs/en/about-claude/models/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Input | text, image | https://platform.claude.com/docs/en/about-claude/models/overview |
| Output | text | https://platform.claude.com/docs/en/about-claude/models/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $2 in / $10 out per 1M tokens | https://platform.claude.com/docs/en/about-claude/models/overview |

## What we tested

Nothing yet. We have not run Claude Sonnet 5.5 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Maximum output tokens for Sonnet 5.5 were not captured.

## Sources

- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)
- [Anthropic newsroom](https://www.anthropic.com/news)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Opus 5.5

Claude Opus 5.5 is an Anthropic language model released Sep 22, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's recommended starting model for most workloads, which Anthropic says leads in agentic coding and knowledge work. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-opus-5-5 · JSON: https://theagenticweekly.com/api/models/claude-opus-5-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Sep 22, 2026 | https://www.anthropic.com/news |
| API id | `claude-opus-5-5` | https://platform.claude.com/docs/en/about-claude/models/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/about-claude/models/overview |
| Input | text, image | https://platform.claude.com/docs/en/about-claude/models/overview |
| Output | text | https://platform.claude.com/docs/en/about-claude/models/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $4 in / $20 out per 1M tokens | https://platform.claude.com/docs/en/about-claude/models/overview |

## Vendor claims and neutral signals

- Vendor claim (Anthropic newsroom): "leads in agentic coding and knowledge work, and costs 40% less to run than Opus 5 on typical workloads."
- Anthropic's models overview says to start with Opus 5.5 for most workloads.

## What we tested

Nothing yet. We have not run Claude Opus 5.5 ourselves, so every fact here is Reported from the sources below.

## Sources

- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)
- [Anthropic newsroom](https://www.anthropic.com/news)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.8 Flash TTS

Gemini 3.8 Flash TTS is Google's latest voice model in our dataset, released Sep 22, 2026. Proprietary, API access only. Google's flagship creative text-to-speech model, generally available since Sep 22, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-8-flash-tts · JSON: https://theagenticweekly.com/api/models/gemini-3-8-flash-tts

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Sep 22, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.8-flash-tts` | https://ai.google.dev/gemini-api/docs/pricing |
| Output | audio | https://ai.google.dev/gemini-api/docs/pricing |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/pricing |
| Pricing | Text input $0.5; Audio output $9 (per 1M tokens). Listed price runs through Dec 31, 2026. | https://ai.google.dev/gemini-api/docs/pricing |

## What we tested

Nothing yet. We have not run Gemini 3.8 Flash TTS ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Input modalities were not captured from the model page.

## Sources

- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.8 Flash-Lite TTS

Gemini 3.8 Flash-Lite TTS is a Google voice model released Sep 22, 2026; Google's newest voice model in our dataset is Gemini 3.8 Flash TTS. Proprietary, API access only. Google's fast, cost-efficient text-to-speech model, generally available since Sep 22, 2026 alongside Gemini 3.8 Flash TTS. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-8-flash-lite-tts · JSON: https://theagenticweekly.com/api/models/gemini-3-8-flash-lite-tts

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Sep 22, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.8-flash-lite-tts` | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-lite-tts |
| Context window | 8,192 tokens (8K) | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-lite-tts |
| Max output | 16,384 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-lite-tts |
| Input | text | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-lite-tts |
| Output | audio | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-lite-tts |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini 3.8 Flash-Lite TTS ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- The 16,384 output limit is the Gemini API serving limit.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Gemini 3.8 Flash-Lite TTS](https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-lite-tts)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-6 Luna

GPT-6 Luna is an OpenAI language model released Sep 22, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's low-cost GPT-6 model, the model behind ChatGPT Free and Go according to OpenAI. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-6-luna · JSON: https://theagenticweekly.com/api/models/gpt-6-luna

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Sep 22, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-6-luna` | https://developers.openai.com/api/docs/models/gpt-6-luna |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-6-luna |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-6-luna |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-6-luna |
| Output | text | https://developers.openai.com/api/docs/models/gpt-6-luna |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-6-luna |
| Pricing | $0.1 in / $0.5 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-6-luna |

## Notes

- Knowledge cutoff listed: May 18, 2026.

## What we tested

Nothing yet. We have not run GPT-6 Luna ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI model page](https://developers.openai.com/api/docs/models/gpt-6-luna)
- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-6 Sol

GPT-6 Sol is an OpenAI language model released Sep 22, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's mid-priced GPT-6 reasoning model, released Sep 22, 2026 with GPT-6 Luna and the model behind paid ChatGPT Chat plans, per OpenAI. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-6-sol · JSON: https://theagenticweekly.com/api/models/gpt-6-sol

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Sep 22, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-6-sol` | https://developers.openai.com/api/docs/models/gpt-6-sol |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-6-sol |
| Output | text | https://developers.openai.com/api/docs/models/gpt-6-sol |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-6-sol |
| Pricing | $2 in / $10 out per 1M tokens. Standard pricing for prompts up to 272K input tokens; longer prompts cost more. | https://developers.openai.com/api/docs/changelog |

## Notes

- OpenAI says GPT-6 Sol powers ChatGPT Plus, Pro, Business and Enterprise Chat after the Oct 7, 2026 Intelligent UI rollout.

## What we tested

Nothing yet. We have not run GPT-6 Sol ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window and maximum output for GPT-6 Sol were not captured separately.

## Sources

- [OpenAI model page](https://developers.openai.com/api/docs/models/gpt-6-sol)
- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI: GPT-6 for everyone](https://openai.com/index/gpt-6-for-everyone/)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Grok 4.7

Grok 4.7 is xAI's latest language model in our dataset, released Sep 21, 2026. Proprietary, API access only. xAI's newest Grok model; its docs call it "the most capable model we've built". Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/grok-4-7 · JSON: https://theagenticweekly.com/api/models/grok-4-7

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | xAI |  |
| Released | Sep 21, 2026 | https://docs.x.ai/developers/release-notes |
| API id | `grok-4.7` | https://docs.x.ai/developers/models/grok-4.7 |
| Context window | 500,000 tokens (500K) | https://docs.x.ai/developers/models/grok-4.7 |
| Input | text, image | https://docs.x.ai/developers/models/grok-4.7 |
| Output | text | https://docs.x.ai/developers/models/grok-4.7 |
| Licence | Proprietary (API) | https://docs.x.ai/developers/models/grok-4.7 |
| Pricing | $2 in / $6 out per 1M tokens | https://docs.x.ai/developers/models/grok-4.7 |

## Vendor claims and neutral signals

- Vendor claim (xAI docs): "the most capable model we've built".

## Notes

- xAI's developer docs are branded SpaceXSI at the time of writing. Reasoning effort levels: low, medium, high, xhigh.

## What we tested

Nothing yet. We have not run Grok 4.7 ourselves, so every fact here is Reported from the sources below.

## Sources

- [xAI model page](https://docs.x.ai/developers/models/grok-4.7)
- [xAI release notes](https://docs.x.ai/developers/release-notes)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Grok Voice Transcribe 2.0

Grok Voice Transcribe 2.0 is a xAI voice model released Sep 17, 2026; xAI's newest voice model in our dataset is Grok Voice 3.0. Proprietary, API access only. xAI's speech-to-text model, available since Sep 17, 2026 and the default for the Speech to Text API. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/grok-voice-transcribe-2-0 · JSON: https://theagenticweekly.com/api/models/grok-voice-transcribe-2-0

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | xAI |  |
| Released | Sep 17, 2026 | https://docs.x.ai/docs/release-notes |
| API id | `grok-voice-transcribe-2.0` | https://docs.x.ai/docs/models |
| Input | audio | https://docs.x.ai/docs/models |
| Output | text | https://docs.x.ai/docs/models |
| Licence | Proprietary (API) | https://docs.x.ai/docs/models |

## What we tested

Nothing yet. We have not run Grok Voice Transcribe 2.0 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Pricing was not captured.

## Sources

- [xAI release notes](https://docs.x.ai/docs/release-notes)
- [xAI models and pricing](https://docs.x.ai/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Fable 5.1

Claude Fable 5.1 is an Anthropic language model released Sep 15, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's top-priced generally available Claude model, announced alongside the access-limited Claude Mythos on Sep 15, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-fable-5-1 · JSON: https://theagenticweekly.com/api/models/claude-fable-5-1

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Sep 15, 2026 | https://www.anthropic.com/news |
| API id | `claude-fable-5-1` | https://platform.claude.com/docs/en/about-claude/models/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/about-claude/models/overview |
| Input | text, image | https://platform.claude.com/docs/en/about-claude/models/overview |
| Output | text | https://platform.claude.com/docs/en/about-claude/models/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $10 in / $50 out per 1M tokens | https://platform.claude.com/docs/en/about-claude/models/overview |

## What we tested

Nothing yet. We have not run Claude Fable 5.1 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Claude Mythos 5.1 is listed by Anthropic with access limited to verification programs; we did not verify its specifications, so it has no entry.

## Sources

- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)
- [Anthropic newsroom](https://www.anthropic.com/news)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.8 Live

Gemini 3.8 Live is a Google voice model released Sep 15, 2026; Google's newest voice model in our dataset is Gemini 3.8 Flash TTS. Proprietary, API access only. Google's real-time voice model for the Gemini Live API, generally available since Sep 15, 2026, with a separate Extended Thinking variant. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-8-live · JSON: https://theagenticweekly.com/api/models/gemini-3-8-live

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Sep 15, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.8-live` | https://ai.google.dev/gemini-api/docs/pricing |
| Input | text, audio | https://ai.google.dev/gemini-api/docs/pricing |
| Output | text, audio | https://ai.google.dev/gemini-api/docs/pricing |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/pricing |
| Pricing | Text input $0.75; Audio input $3; Text output $4.5; Audio output $12 (per 1M tokens) | https://ai.google.dev/gemini-api/docs/pricing |

## Notes

- The Extended Thinking variant is gemini-3.8-live-extended-thinking (same release date).

## What we tested

Nothing yet. We have not run Gemini 3.8 Live ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.8 Live Extended Thinking

Gemini 3.8 Live Extended Thinking is a Google voice model released Sep 15, 2026; Google's newest voice model in our dataset is Gemini 3.8 Flash TTS. Proprietary, API access only. Google's high-reasoning audio-to-audio Live API model with background reasoning, generally available since Sep 15, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-8-live-extended-thinking · JSON: https://theagenticweekly.com/api/models/gemini-3-8-live-extended-thinking

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Sep 15, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.8-live-extended-thinking` | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-live-extended-thinking |
| Context window | 131,072 tokens (131K) | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-live-extended-thinking |
| Max output | 65,536 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-live-extended-thinking |
| Input | text, image, audio | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-live-extended-thinking |
| Output | text, audio | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-live-extended-thinking |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini 3.8 Live Extended Thinking ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Gemini 3.8 Live Extended Thinking](https://ai.google.dev/gemini-api/docs/models/gemini-3.8-live-extended-thinking)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# DeepSeek-V4.1-Flash

DeepSeek-V4.1-Flash is a DeepSeek language model released Sep 10, 2026; DeepSeek's newest language model in our dataset is DeepSeek-V4.1-Flash-1017. Open weights (MIT). DeepSeek's MIT-licensed V4.1 Flash model with a 1M-token context, released Sep 10, 2026 and updated to the 1017 version on Oct 17, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/deepseek-v4-1-flash · JSON: https://theagenticweekly.com/api/models/deepseek-v4-1-flash

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | DeepSeek |  |
| Released | Sep 10, 2026 | https://api-docs.deepseek.com/updates/ |
| Latest update | Oct 17, 2026 | https://api-docs.deepseek.com/updates/ |
| API id | `deepseek-flash` | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Context window | 1,000,000 tokens (1M) | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Max output | 384,000 tokens | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Input | text, image | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Output | text | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |
| Licence | MIT (open weights). License as listed on the Hugging Face model repository. | https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017 |
| Pricing | ¥1 in / ¥4 out per 1M tokens; peak: ¥2 / ¥8. Official price list is in CNY. Off-peak shown; peak (Beijing time Mon-Fri 09-12 and 14-18) is double. We do not convert currencies because the USD table was not readable without JavaScript. | https://api-docs.deepseek.com/zh-cn/quick_start/pricing |

## Notes

- Superseded behind the deepseek-flash API name by DeepSeek-V4.1-Flash-1017 (Oct 17, 2026); see /models/deepseek-v4-1-flash-1017.

## What we tested

Nothing yet. We have not run DeepSeek-V4.1-Flash ourselves, so every fact here is Reported from the sources below.

## Sources

- [DeepSeek API updates](https://api-docs.deepseek.com/updates/)
- [DeepSeek pricing (zh-CN)](https://api-docs.deepseek.com/zh-cn/quick_start/pricing)
- [Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Live 1

GPT-Live 1 is OpenAI's latest voice model in our dataset, released Sep 10, 2026. Proprietary, API access only. OpenAI's full-duplex voice model that listens and speaks at once; API general availability on Sep 10, 2026 at $0.05 per minute. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-live-1 · JSON: https://theagenticweekly.com/api/models/gpt-live-1

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Sep 10, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-live-1` | https://developers.openai.com/api/docs/models/gpt-live-1 |
| Input | text, audio | https://developers.openai.com/api/docs/models/gpt-live-1 |
| Output | text, audio | https://developers.openai.com/api/docs/models/gpt-live-1 |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-live-1 |
| Pricing | Voice session, billed per second $0.05 (per minute). Backend model and tool usage are billed separately. | https://developers.openai.com/api/docs/models/gpt-live-1 |

## What we tested

Nothing yet. We have not run GPT-Live 1 ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI model page](https://developers.openai.com/api/docs/models/gpt-live-1)
- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI API announcement](https://openai.com/index/introducing-gpt-live-1-in-the-api/)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# North Small Translate

North Small Translate is Cohere's latest language model in our dataset, released Sep 9, 2026. Open weights (CC BY-NC 4.0). Cohere's open-weights mixture-of-experts machine-translation model for more than 50 languages, released Sep 9, 2026 for non-commercial use under CC BY-NC 4.0. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/north-small-translate · JSON: https://theagenticweekly.com/api/models/north-small-translate

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Cohere |  |
| Released | Sep 9, 2026 | https://docs.cohere.com/changelog |
| API id | `north-small-translate-09-2026` | https://docs.cohere.com/changelog |
| Context window | 16,000 tokens (16K). Cohere lists a 16K context length. | https://docs.cohere.com/changelog |
| Parameters | 218B total, 25B active, mixture-of-experts | https://docs.cohere.com/changelog |
| Licence | CC BY-NC 4.0 (open weights). Open weights are for non-commercial use only. | https://docs.cohere.com/changelog |

## What we tested

Nothing yet. We have not run North Small Translate ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price was not captured; the model is on the free tier of the Chat V2 API.

## Sources

- [Cohere release notes](https://docs.cohere.com/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Image-2.5 Flare

GPT-Image-2.5 Flare is OpenAI's latest image-generation model in our dataset, released Sep 8, 2026. Proprietary, API access only. OpenAI's fast image generation model for everyday use, released Sep 8, 2026 alongside GPT-Image-2.5 Sunburst. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-image-2-5-flare · JSON: https://theagenticweekly.com/api/models/gpt-image-2-5-flare

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Sep 8, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-image-2.5-flare` | https://developers.openai.com/api/docs/models/gpt-image-2.5-flare |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-image-2.5-flare |
| Output | image | https://developers.openai.com/api/docs/models/gpt-image-2.5-flare |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-image-2.5-flare |
| Pricing | $5 in / $30 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-image-2.5-flare |

## What we tested

Nothing yet. We have not run GPT-Image-2.5 Flare ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-Image-2.5 Flare](https://developers.openai.com/api/docs/models/gpt-image-2.5-flare)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Image-2.5 Sunburst

GPT-Image-2.5 Sunburst is an OpenAI image-generation model released Sep 8, 2026; OpenAI's newest image-generation model in our dataset is GPT-Image-2.5 Flare. Proprietary, API access only. OpenAI's most capable image generation and editing model, released Sep 8, 2026, and #1 on the LMArena text-to-image snapshot we cite. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-image-2-5-sunburst · JSON: https://theagenticweekly.com/api/models/gpt-image-2-5-sunburst

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Sep 8, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-image-2.5-sunburst` | https://developers.openai.com/api/docs/models/gpt-image-2.5-sunburst |
| Output | image | https://developers.openai.com/api/docs/models/gpt-image-2.5-sunburst |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-image-2.5-sunburst |
| Pricing | $5 in / $30 out per 1M tokens. Billed in tokens, not per image: text input $5, image input $8, image output $30 per 1M tokens. Cost per image depends on size, quality and references. | https://developers.openai.com/api/docs/models/gpt-image-2.5-sunburst |

## Vendor claims and neutral signals

- Neutral signal: #1 on LMArena Text-to-Image in the Oct 7, 2026 snapshot (score marked preliminary).

## What we tested

Nothing yet. We have not run GPT-Image-2.5 Sunburst ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- The sibling GPT-Image-2.5 Flare (fast) is ranked #2 on the same leaderboard; its specifications were not captured.

## Sources

- [OpenAI model page](https://developers.openai.com/api/docs/models/gpt-image-2.5-sunburst)
- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI announcement](https://openai.com/index/introducing-chatgpt-images-2-5/)
- [LMArena Text-to-Image](https://arena.ai/leaderboard/text-to-image)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-6 Astra

GPT-6 Astra is an OpenAI language model released Sep 3, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's flagship GPT-6 model for complex reasoning and coding, and its most expensive text model per token. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-6-astra · JSON: https://theagenticweekly.com/api/models/gpt-6-astra

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Sep 3, 2026 | https://openai.com/index/gpt-6-astra/ |
| API id | `gpt-6-astra` | https://developers.openai.com/api/docs/models/gpt-6-astra |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-6-astra |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-6-astra |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-6-astra |
| Output | text | https://developers.openai.com/api/docs/models/gpt-6-astra |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-6-astra |
| Pricing | $10 in / $50 out per 1M tokens. Prompts above 272K input tokens cost 2x input and 1.5x output. | https://developers.openai.com/api/docs/models/gpt-6-astra |

## Vendor claims and neutral signals

- OpenAI's models page calls it the "flagship for complex reasoning and coding".

## Notes

- Knowledge cutoff listed by OpenAI: Apr 30, 2026.

## What we tested

Nothing yet. We have not run GPT-6 Astra ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI model page](https://developers.openai.com/api/docs/models/gpt-6-astra)
- [OpenAI announcement](https://openai.com/index/gpt-6-astra/)
- [OpenAI models list](https://developers.openai.com/api/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.8 Flash

Gemini 3.8 Flash is a Google language model released Sep 2, 2026; Google's newest language model in our dataset is Gemini 4 Argon. Proprietary, API access only. Google's current fast Gemini model ("most intelligent Flash model" per Google's changelog), replacing Gemini 3.7 Flash. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-8-flash · JSON: https://theagenticweekly.com/api/models/gemini-3-8-flash

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Sep 2, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.8-flash` | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash |
| Context window | 1,048,576 tokens (1.05M) | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash |
| Max output | 65,536 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash |
| Input | text, image, video, audio, pdf | https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/pricing |
| Pricing | $0.75 in / $3.75 out per 1M tokens; from Jan 1, 2027: $1.5 / $7.5. Current pricing runs through Dec 31, 2026, then $1.50 / $7.50. | https://ai.google.dev/gemini-api/docs/pricing |

## What we tested

Nothing yet. We have not run Gemini 3.8 Flash ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Output modalities were not captured from the model page.

## Sources

- [Gemini API model page](https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash)
- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Mythos 5.1

Claude Mythos 5.1 is an Anthropic language model (limited access) released Sep 1, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's access-limited Claude Mythos successor for Project Glasswing participants, launched with Claude Fable 5.1 (release-notes date Sep 1, 2026). Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-mythos-5-1 · JSON: https://theagenticweekly.com/api/models/claude-mythos-5-1

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Sep 1, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-mythos-5-1` | https://platform.claude.com/docs/en/release-notes/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/release-notes/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $10 in / $50 out per 1M tokens | https://platform.claude.com/docs/en/release-notes/overview |

## What we tested

Nothing yet. We have not run Claude Mythos 5.1 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Anthropic's newsroom announcement for Fable 5.1 and Mythos is dated Sep 15, 2026 on our Fable 5.1 page; this page uses the API release-notes date.

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini Omni 1.1 Flash

Gemini Omni 1.1 Flash is Google's latest video-generation model in our dataset, released Aug 27, 2026. Proprietary, API access only. Google's fast video generation and editing model (API id gemini-omni-1.1-flash), generally available since Aug 27, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-omni-1-1-flash · JSON: https://theagenticweekly.com/api/models/gemini-omni-1-1-flash

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Aug 27, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-omni-1.1-flash` | https://ai.google.dev/gemini-api/docs/pricing |
| Output | video | https://ai.google.dev/gemini-api/docs/pricing |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/pricing |
| Pricing | Input per 1M tokens $1.5; Output per second of 720p video (approx.) $0.1 (mixed) | https://ai.google.dev/gemini-api/docs/pricing |

## What we tested

Nothing yet. We have not run Gemini Omni 1.1 Flash ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Input modalities were not captured. We did not verify other video models (Veo, Kling, Seedance, Grok Imagine Video, Runway) against vendor sources, so they are left out.

## Sources

- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Google developer blog](https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-omni-1-1-flash/)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.5 Transcribe

Gemini 3.5 Transcribe is a Google voice model released Aug 26, 2026; Google's newest voice model in our dataset is Gemini 3.8 Flash TTS. Proprietary, API access only. Google's speech-to-text model with speaker diarization, word-level timestamps and custom vocabulary, generally available since Aug 26, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-5-transcribe · JSON: https://theagenticweekly.com/api/models/gemini-3-5-transcribe

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Aug 26, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.5-transcribe` | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-transcribe |
| Input | audio | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-transcribe |
| Output | text | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-transcribe |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini 3.5 Transcribe ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Token limits do not apply; the page lists audio limits instead (up to 1 hour per request, 30 minutes with diarization or word timestamps).

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Gemini 3.5 Transcribe](https://ai.google.dev/gemini-api/docs/models/gemini-3.5-transcribe)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GLM-5.3-Flash

GLM-5.3-Flash is Z.ai's latest language model in our dataset, released Aug 26, 2026. Z.ai's GLM-5.3-Flash, released Aug 26, 2026 with native visual capabilities and a hybrid linear and sparse attention design. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/glm-5-3-flash · JSON: https://theagenticweekly.com/api/models/glm-5-3-flash

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Z.ai |  |
| Released | Aug 26, 2026 | https://docs.z.ai/release-notes/new-released |
| Parameters | 320B total, 18B active, hybrid linear and sparse attention | https://docs.z.ai/release-notes/new-released |

## What we tested

Nothing yet. We have not run GLM-5.3-Flash ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window, license and price were not captured from Z.ai's own pages; Mistral's changelog lists a 1M-token context and $0.15 / $0.50 per 1M tokens for its hosted copy, which we did not treat as Z.ai's own figures.

## Sources

- [Z.ai release notes](https://docs.z.ai/release-notes/new-released)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GLM-5.3

GLM-5.3 is a Z.ai language model released Aug 14, 2026; Z.ai's newest language model in our dataset is GLM-5.3-Flash. Open weights (GLM-5.3 License (MIT-style plus a MaaS condition)). Z.ai's open-weights coding and agent model, released Aug 14, 2026, with weights on Hugging Face under a permissive license that adds a security-review condition for very large API businesses. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/glm-5-3 · JSON: https://theagenticweekly.com/api/models/glm-5-3

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Z.ai |  |
| Released | Aug 14, 2026 | https://z.ai/blog/glm-5.3 |
| API id | `glm-5.3` | https://docs.z.ai/guides/llm/glm-5.3 |
| Context window | 1,048,576 tokens (1.05M) | https://docs.z.ai/guides/llm/glm-5.3 |
| Output | text | https://docs.z.ai/guides/llm/glm-5.3 |
| Licence | GLM-5.3 License (MIT-style plus a MaaS condition) (open weights). Model-as-a-service operators above US$10B revenue in 12 months must pass a Z.ai security review. | https://huggingface.co/zai-org/GLM-5.3/blob/main/LICENSE |
| Pricing | $1.4 in / $4.4 out per 1M tokens | https://docs.z.ai/guides/overview/pricing |

## Vendor claims and neutral signals

- Vendor claim (Z.ai): "the most capable open-weights model for coding", based on Z.ai's own in-house benchmark. We did not verify this.

## What we tested

Nothing yet. We have not run GLM-5.3 ourselves, so every fact here is Reported from the sources below.

## Sources

- [Z.ai release blog](https://z.ai/blog/glm-5.3)
- [Z.ai docs](https://docs.z.ai/guides/llm/glm-5.3)
- [Z.ai pricing](https://docs.z.ai/guides/overview/pricing)
- [Hugging Face](https://huggingface.co/zai-org/GLM-5.3)
- [GLM-5.3 LICENSE (Hugging Face)](https://huggingface.co/zai-org/GLM-5.3/blob/main/LICENSE)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Qwen3.8-27B

Qwen3.8-27B is Alibaba's latest language model in our dataset, released Aug 14, 2026. Open weights (Apache-2.0). Alibaba's 27B dense open-weights Qwen3.8 model under Apache 2.0, the small Qwen3.8 you can self-host. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/qwen3-8-27b · JSON: https://theagenticweekly.com/api/models/qwen3-8-27b

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Alibaba |  |
| Released | Aug 14, 2026 | https://github.com/QwenLM/Qwen3.8 |
| API id | `Qwen/Qwen3.8-27B` | https://huggingface.co/Qwen/Qwen3.8-27B |
| Context window | 262,144 tokens (262K). Native; extensible to 1,000,000 tokens per the model card. | https://huggingface.co/Qwen/Qwen3.8-27B |
| Parameters | 27B total, dense | https://huggingface.co/Qwen/Qwen3.8-27B |
| Licence | Apache-2.0 (open weights) | https://huggingface.co/Qwen/Qwen3.8-27B |

## What we tested

Nothing yet. We have not run Qwen3.8-27B ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Modalities were not captured. Alibaba says a hosted version with 1M context by default is coming; its price was not verified.

## Sources

- [Hugging Face model card](https://huggingface.co/Qwen/Qwen3.8-27B)
- [Qwen3.8 on GitHub](https://github.com/QwenLM/Qwen3.8)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Grok 4.6

Grok 4.6 is a xAI language model released Aug 12, 2026; xAI's newest language model in our dataset is Grok 4.7. Proprietary, API access only. xAI's Grok 4.6 frontier model for coding, agentic tasks and knowledge work, available on the API since Aug 12, 2026 with a 500k context window. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/grok-4-6 · JSON: https://theagenticweekly.com/api/models/grok-4-6

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | xAI |  |
| Released | Aug 12, 2026 | https://docs.x.ai/docs/release-notes |
| API id | `grok-4.6` | https://docs.x.ai/docs/release-notes |
| Context window | 500,000 tokens (500K) | https://docs.x.ai/docs/release-notes |
| Input | text, image | https://docs.x.ai/docs/release-notes |
| Output | text | https://docs.x.ai/docs/release-notes |
| Licence | Proprietary (API) | https://docs.x.ai/docs/models |
| Pricing | $2 in / $6 out per 1M tokens. Below 200k prompt tokens; $4 / $1 / $12 per 1M tokens (input / cached input / output) above 200k. | https://docs.x.ai/docs/release-notes |

## What we tested

Nothing yet. We have not run Grok 4.6 ourselves, so every fact here is Reported from the sources below.

## Sources

- [xAI release notes](https://docs.x.ai/docs/release-notes)
- [xAI models and pricing](https://docs.x.ai/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Qwen3.8-2.4T-A95B

Qwen3.8-2.4T-A95B is an Alibaba language model released Aug 12, 2026; Alibaba's newest language model in our dataset is Qwen3.8-27B. Open weights (Qwen3.8-Max License (custom)). Alibaba's 2.4T-parameter open-weights flagship (95B active), released Aug 12, 2026 under the custom Qwen3.8-Max license. The hosted Qwen3.8-Max adds vision and a 1M default context. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/qwen3-8-2-4t-a95b · JSON: https://theagenticweekly.com/api/models/qwen3-8-2-4t-a95b

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Alibaba |  |
| Released | Aug 12, 2026 | https://github.com/QwenLM/Qwen3.8 |
| API id | `Qwen/Qwen3.8-2.4T-A95B` | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
| Context window | 262,144 tokens (262K). Native; extensible to 1,010,000 tokens per the model card. | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
| Parameters | 2.4T total, 95B active, mixture-of-experts | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
| Licence | Qwen3.8-Max License (custom) (open weights). Custom license with commercial thresholds for very large products and for model-as-a-service businesses; read the license text. | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/blob/main/LICENSE |

## What we tested

Nothing yet. We have not run Qwen3.8-2.4T-A95B ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Hosted Qwen3.8-Max API pricing was not verified at the vendor.

## Sources

- [Hugging Face model card](https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B)
- [Qwen3.8 on GitHub](https://github.com/QwenLM/Qwen3.8)
- [Qwen3.8-2.4T-A95B LICENSE (Hugging Face)](https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/blob/main/LICENSE)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini Robotics ER 2

Gemini Robotics ER 2 is a Google language model (public preview) released Jul 30, 2026; Google's newest language model in our dataset is Gemini 4 Argon. Proprietary, API access only. Google's embodied-reasoning model for robotics, in public preview since Jul 30, 2026, with a streaming sibling endpoint for the Live API. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-robotics-er-2 · JSON: https://theagenticweekly.com/api/models/gemini-robotics-er-2

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Jul 30, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-robotics-er-2-preview` | https://ai.google.dev/gemini-api/docs/models |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini Robotics ER 2 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window and output limits were not captured: the model page was not available.
- Input modalities per the changelog: text, image, video and audio.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Live-Transcribe

GPT-Live-Transcribe is an OpenAI voice model released Jul 28, 2026; OpenAI's newest voice model in our dataset is GPT-Live 1. Proprietary, API access only. OpenAI's low-latency streaming speech-to-text model for live audio, released Jul 28, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-live-transcribe · JSON: https://theagenticweekly.com/api/models/gpt-live-transcribe

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Jul 28, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-live-transcribe` | https://developers.openai.com/api/docs/models/gpt-live-transcribe |
| Input | audio, text | https://developers.openai.com/api/docs/models/gpt-live-transcribe |
| Output | text | https://developers.openai.com/api/docs/models/gpt-live-transcribe |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-live-transcribe |
| Pricing | Realtime transcription $0.017 (per minute) | https://developers.openai.com/api/docs/models/gpt-live-transcribe |

## What we tested

Nothing yet. We have not run GPT-Live-Transcribe ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-Live-Transcribe](https://developers.openai.com/api/docs/models/gpt-live-transcribe)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Transcribe

GPT-Transcribe is an OpenAI voice model released Jul 28, 2026; OpenAI's newest voice model in our dataset is GPT-Live 1. Proprietary, API access only. OpenAI's speech-to-text model for audio files and committed Realtime turns, released Jul 28, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-transcribe · JSON: https://theagenticweekly.com/api/models/gpt-transcribe

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Jul 28, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-transcribe` | https://developers.openai.com/api/docs/models/gpt-transcribe |
| Input | audio, text | https://developers.openai.com/api/docs/models/gpt-transcribe |
| Output | text | https://developers.openai.com/api/docs/models/gpt-transcribe |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-transcribe |
| Pricing | Transcription $0.0045 (per minute) | https://developers.openai.com/api/docs/models/gpt-transcribe |

## What we tested

Nothing yet. We have not run GPT-Transcribe ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-Transcribe](https://developers.openai.com/api/docs/models/gpt-transcribe)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Opus 5

Claude Opus 5 is an Anthropic language model released Jul 24, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's Claude Opus generation launched Jul 24, 2026, priced $5 / $25 per million tokens with a 1M-token context; its successor Claude Opus 5.5 is priced at $4 / $20. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-opus-5 · JSON: https://theagenticweekly.com/api/models/claude-opus-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Jul 24, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-opus-5` | https://platform.claude.com/docs/en/release-notes/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/release-notes/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $5 in / $25 out per 1M tokens | https://platform.claude.com/docs/en/release-notes/overview |

## What we tested

Nothing yet. We have not run Claude Opus 5 ourselves, so every fact here is Reported from the sources below.

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.5 Flash-Lite

Gemini 3.5 Flash-Lite is a Google language model released Jul 21, 2026; Google's newest language model in our dataset is Gemini 4 Argon. Proprietary, API access only. Google's low-latency, cost-effective Gemini 3.5 Flash-Lite, generally available since Jul 21, 2026 as a low-latency subagent option for high-volume automation. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-5-flash-lite · JSON: https://theagenticweekly.com/api/models/gemini-3-5-flash-lite

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Jul 21, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.5-flash-lite` | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash-lite |
| Context window | 1,048,576 tokens (1.05M) | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash-lite |
| Max output | 65,536 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash-lite |
| Input | text, image, video, audio, pdf | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash-lite |
| Output | text | https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash-lite |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini 3.5 Flash-Lite ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Gemini 3.5 Flash-Lite](https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash-lite)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.6 Flash

Gemini 3.6 Flash is a Google language model released Jul 21, 2026; Google's newest language model in our dataset is Gemini 4 Argon. Proprietary, API access only. Google's stable Gemini 3.6 Flash, released Jul 21, 2026 with improved token efficiency and code and agentic planning. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-6-flash · JSON: https://theagenticweekly.com/api/models/gemini-3-6-flash

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Jul 21, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.6-flash` | https://ai.google.dev/gemini-api/docs/models/gemini-3.6-flash |
| Context window | 1,048,576 tokens (1.05M) | https://ai.google.dev/gemini-api/docs/models/gemini-3.6-flash |
| Max output | 65,536 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3.6-flash |
| Input | text, image, video, audio, pdf | https://ai.google.dev/gemini-api/docs/models/gemini-3.6-flash |
| Output | text | https://ai.google.dev/gemini-api/docs/models/gemini-3.6-flash |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini 3.6 Flash ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Gemini 3.6 Flash](https://ai.google.dev/gemini-api/docs/models/gemini-3.6-flash)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Kimi K3

Kimi K3 is a Moonshot AI language model released Jul 16, 2026; Moonshot AI's newest language model in our dataset is Kimi K3.1 Pro. Open weights (Kimi K3 License (custom)). Moonshot's 2.8-trillion-parameter open-weights model with native vision and a 1M-token context, released under a custom Kimi K3 license. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/kimi-k3 · JSON: https://theagenticweekly.com/api/models/kimi-k3

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Moonshot AI |  |
| Released | Jul 16, 2026 | https://www.kimi.com/blog/kimi-k3 |
| API id | `kimi-k3` | https://platform.kimi.ai/docs/guide/kimi-k3-quickstart |
| Context window | 1,000,000 tokens (1M) | https://www.kimi.com/blog/kimi-k3 |
| Parameters | 2.8T total, 16 of 896 experts active, mixture-of-experts | https://huggingface.co/moonshotai/Kimi-K3 |
| Licence | Kimi K3 License (custom) (open weights). Model-as-a-service businesses above US$20M revenue in 12 months need a separate agreement; large products must display "Kimi K3". | https://huggingface.co/moonshotai/Kimi-K3 |
| Pricing | $3 in / $15 out per 1M tokens. Cache-miss input $3.00, cache-hit $0.30, output $15.00. | https://platform.kimi.ai/docs/pricing/chat |

## What we tested

Nothing yet. We have not run Kimi K3 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Input/output modalities beyond "native vision" were not captured.

## Sources

- [Kimi K3 blog](https://www.kimi.com/blog/kimi-k3)
- [Hugging Face model card](https://huggingface.co/moonshotai/Kimi-K3)
- [Kimi API pricing](https://platform.kimi.ai/docs/pricing/chat)
- [Kimi K3 quickstart (Moonshot platform docs)](https://platform.kimi.ai/docs/guide/kimi-k3-quickstart)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-5.6 Luna

GPT-5.6 Luna is an OpenAI language model released Jul 9, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's GPT-5.6 low-cost model for high-volume workloads, released Jul 9, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-5-6-luna · JSON: https://theagenticweekly.com/api/models/gpt-5-6-luna

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Jul 9, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-5.6-luna` | https://developers.openai.com/api/docs/models/gpt-5.6-luna |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-5.6-luna |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-5.6-luna |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-5.6-luna |
| Output | text | https://developers.openai.com/api/docs/models/gpt-5.6-luna |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-5.6-luna |
| Pricing | $0.2 in / $1.2 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-5.6-luna |

## What we tested

Nothing yet. We have not run GPT-5.6 Luna ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-5.6 Luna](https://developers.openai.com/api/docs/models/gpt-5.6-luna)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-5.6 Sol

GPT-5.6 Sol is an OpenAI language model released Jul 9, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's GPT-5.6 flagship, released Jul 9, 2026 with a 1.05M-token context. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-5-6-sol · JSON: https://theagenticweekly.com/api/models/gpt-5-6-sol

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Jul 9, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-5.6-sol` | https://developers.openai.com/api/docs/models/gpt-5.6-sol |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-5.6-sol |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-5.6-sol |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-5.6-sol |
| Output | text | https://developers.openai.com/api/docs/models/gpt-5.6-sol |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-5.6-sol |
| Pricing | $4 in / $20 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-5.6-sol |

## What we tested

Nothing yet. We have not run GPT-5.6 Sol ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- OpenAI lists promotional pricing for GPT-5.6 Sol through at least Nov 21, 2026.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-5.6 Sol](https://developers.openai.com/api/docs/models/gpt-5.6-sol)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-5.6 Terra

GPT-5.6 Terra is an OpenAI language model released Jul 9, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's GPT-5.6 mid-tier model, released Jul 9, 2026 to balance intelligence and cost. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-5-6-terra · JSON: https://theagenticweekly.com/api/models/gpt-5-6-terra

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Jul 9, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-5.6-terra` | https://developers.openai.com/api/docs/models/gpt-5.6-terra |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-5.6-terra |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-5.6-terra |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-5.6-terra |
| Output | text | https://developers.openai.com/api/docs/models/gpt-5.6-terra |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-5.6-terra |
| Pricing | $2 in / $12 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-5.6-terra |

## What we tested

Nothing yet. We have not run GPT-5.6 Terra ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-5.6 Terra](https://developers.openai.com/api/docs/models/gpt-5.6-terra)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Grok 4.5

Grok 4.5 is a xAI language model released Jul 8, 2026; xAI's newest language model in our dataset is Grok 4.7. Proprietary, API access only. xAI's Grok 4.5 model for coding, agentic tasks and knowledge work, available on the API since Jul 8, 2026 with configurable reasoning effort. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/grok-4-5 · JSON: https://theagenticweekly.com/api/models/grok-4-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | xAI |  |
| Released | Jul 8, 2026 | https://docs.x.ai/docs/release-notes |
| API id | `grok-4.5` | https://docs.x.ai/docs/release-notes |
| Context window | 500,000 tokens (500K) | https://docs.x.ai/docs/models |
| Input | text, image | https://docs.x.ai/docs/models |
| Output | text | https://docs.x.ai/docs/models |
| Licence | Proprietary (API) | https://docs.x.ai/docs/models |
| Pricing | $2 in / $6 out per 1M tokens | https://docs.x.ai/docs/models |

## What we tested

Nothing yet. We have not run Grok 4.5 ourselves, so every fact here is Reported from the sources below.

## Sources

- [xAI release notes](https://docs.x.ai/docs/release-notes)
- [xAI models and pricing](https://docs.x.ai/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Cohere Transcribe Arabic

Cohere Transcribe Arabic is Cohere's latest voice model in our dataset, released Jul 7, 2026. Open weights (Apache 2.0). Cohere's open-source Arabic speech-to-text model, a 2B fine-tune of Cohere Transcribe released Jul 7, 2026 under Apache 2.0. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/cohere-transcribe-arabic · JSON: https://theagenticweekly.com/api/models/cohere-transcribe-arabic

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Cohere |  |
| Released | Jul 7, 2026 | https://docs.cohere.com/changelog |
| API id | `cohere-transcribe-arabic-07-2026` | https://docs.cohere.com/changelog |
| Parameters | 2B total, conformer-based encoder-decoder | https://docs.cohere.com/changelog |
| Licence | Apache 2.0 (open weights) | https://docs.cohere.com/changelog |

## What we tested

Nothing yet. We have not run Cohere Transcribe Arabic ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price was not captured.

## Sources

- [Cohere release notes](https://docs.cohere.com/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Realtime-2.1

GPT-Realtime-2.1 is an OpenAI voice model released Jul 6, 2026; OpenAI's newest voice model in our dataset is GPT-Live 1. Proprietary, API access only. OpenAI's realtime speech-to-speech reasoning model, released Jul 6, 2026 with better alphanumeric recognition and interruption handling. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-realtime-2-1 · JSON: https://theagenticweekly.com/api/models/gpt-realtime-2-1

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Jul 6, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-realtime-2.1` | https://developers.openai.com/api/docs/models/gpt-realtime-2.1 |
| Context window | 128,000 tokens (128K) | https://developers.openai.com/api/docs/models/gpt-realtime-2.1 |
| Max output | 32,000 tokens | https://developers.openai.com/api/docs/models/gpt-realtime-2.1 |
| Input | text, audio, image | https://developers.openai.com/api/docs/models/gpt-realtime-2.1 |
| Output | text, audio | https://developers.openai.com/api/docs/models/gpt-realtime-2.1 |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-realtime-2.1 |
| Pricing | $4 in / $24 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-realtime-2.1 |

## What we tested

Nothing yet. We have not run GPT-Realtime-2.1 ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-Realtime-2.1](https://developers.openai.com/api/docs/models/gpt-realtime-2.1)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Realtime-2.1 mini

GPT-Realtime-2.1 mini is an OpenAI voice model released Jul 6, 2026; OpenAI's newest voice model in our dataset is GPT-Live 1. Proprietary, API access only. OpenAI's faster, lower-cost distilled realtime voice model, released Jul 6, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-realtime-2-1-mini · JSON: https://theagenticweekly.com/api/models/gpt-realtime-2-1-mini

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Jul 6, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-realtime-2.1-mini` | https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini |
| Context window | 128,000 tokens (128K) | https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini |
| Max output | 32,000 tokens | https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini |
| Input | text, audio, image | https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini |
| Output | text, audio | https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini |
| Pricing | $0.6 in / $2.4 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini |

## What we tested

Nothing yet. We have not run GPT-Realtime-2.1 mini ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-Realtime-2.1 mini](https://developers.openai.com/api/docs/models/gpt-realtime-2.1-mini)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Sonnet 5

Claude Sonnet 5 is an Anthropic language model released Jun 30, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's Claude Sonnet generation launched Jun 30, 2026 at $2 / $10 per million tokens, a price Anthropic made standard on Aug 10, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-sonnet-5 · JSON: https://theagenticweekly.com/api/models/claude-sonnet-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Jun 30, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-sonnet-5` | https://platform.claude.com/docs/en/release-notes/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/release-notes/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $2 in / $10 out per 1M tokens | https://platform.claude.com/docs/en/release-notes/overview |

## What we tested

Nothing yet. We have not run Claude Sonnet 5 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Priority Tier is not available on Claude Sonnet 5 per the release notes.

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Nano Banana 2 Lite

Nano Banana 2 Lite is a Google image-generation model released Jun 30, 2026; Google's newest image-generation model in our dataset is Nano Banana 2.1. Proprietary, API access only. Google's ultra-low-latency, cost-effective image generation and editing model (gemini-3.1-flash-lite-image), generally available since Jun 30, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/nano-banana-2-lite · JSON: https://theagenticweekly.com/api/models/nano-banana-2-lite

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Jun 30, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.1-flash-lite-image` | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image |
| Context window | 65,536 tokens (66K) | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image |
| Max output | 4,096 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image |
| Input | text, image, video, pdf | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image |
| Output | image, text | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Nano Banana 2 Lite ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Nano Banana 2 Lite](https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GLM-5.2

GLM-5.2 is a Z.ai language model released Jun 16, 2026; Z.ai's newest language model in our dataset is GLM-5.3-Flash. Z.ai's GLM-5.2, released Jun 16, 2026 with a 1M-token context; Mistral's changelog lists it as retiring on Oct 31, 2026 in favor of GLM-5.3. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/glm-5-2 · JSON: https://theagenticweekly.com/api/models/glm-5-2

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Z.ai |  |
| Released | Jun 16, 2026 | https://docs.z.ai/release-notes/new-released |
| Context window | 1,000,000 tokens (1M). Z.ai describes this as 1M lossless context. | https://docs.z.ai/release-notes/new-released |

## What we tested

Nothing yet. We have not run GLM-5.2 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- License and price were not captured.

## Sources

- [Z.ai release notes](https://docs.z.ai/release-notes/new-released)
- [Mistral changelog](https://docs.mistral.ai/getting-started/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Fable 5

Claude Fable 5 is an Anthropic language model released Jun 9, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's most capable model open to all customers when it launched on Jun 9, 2026, alongside the access-limited Claude Mythos 5; succeeded by Claude Fable 5.1. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-fable-5 · JSON: https://theagenticweekly.com/api/models/claude-fable-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Jun 9, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-fable-5` | https://platform.claude.com/docs/en/release-notes/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/release-notes/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $10 in / $50 out per 1M tokens | https://platform.claude.com/docs/en/release-notes/overview |

## What we tested

Nothing yet. We have not run Claude Fable 5 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Input and output modalities were not captured from the release notes for this model.
- The price is stated in the release notes of its successor (Fable 5.1, "$10 / $50 ... the same as Claude Fable 5").

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Mythos 5

Claude Mythos 5 is an Anthropic language model (limited access) released Jun 9, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's access-limited Claude model for Project Glasswing participants, launched Jun 9, 2026 alongside Claude Fable 5. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-mythos-5 · JSON: https://theagenticweekly.com/api/models/claude-mythos-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Jun 9, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-mythos-5` | https://platform.claude.com/docs/en/release-notes/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/release-notes/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |

## What we tested

Nothing yet. We have not run Claude Mythos 5 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price was not captured.
- Access is limited to Project Glasswing participants per the release notes.

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# North Mini Code

North Mini Code is a Cohere language model released Jun 9, 2026; Cohere's newest language model in our dataset is North Small Translate. Open weights (Apache 2.0). Cohere's first agentic coding model, released Jun 9, 2026 as a 30B-total, 3B-active mixture-of-experts model under Apache 2.0. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/north-mini-code · JSON: https://theagenticweekly.com/api/models/north-mini-code

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Cohere |  |
| Released | Jun 9, 2026 | https://docs.cohere.com/changelog |
| API id | `north-mini-code-1-0` | https://docs.cohere.com/changelog |
| Context window | 256,000 tokens (256K). 256K input; 64K output. | https://docs.cohere.com/changelog |
| Max output | 64,000 tokens | https://docs.cohere.com/changelog |
| Parameters | 30B total, 3B active, mixture-of-experts | https://docs.cohere.com/changelog |
| Licence | Apache 2.0 (open weights) | https://docs.cohere.com/changelog |

## What we tested

Nothing yet. We have not run North Mini Code ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price was not captured.

## Sources

- [Cohere release notes](https://docs.cohere.com/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Opus 4.8

Claude Opus 4.8 is an Anthropic language model released May 28, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's Claude Opus 4.8, launched May 28, 2026 with a 1M-token context by default and 128k max output. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-opus-4-8 · JSON: https://theagenticweekly.com/api/models/claude-opus-4-8

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | May 28, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-opus-4-8` | https://platform.claude.com/docs/en/release-notes/overview |
| Context window | 1,000,000 tokens (1M) | https://platform.claude.com/docs/en/release-notes/overview |
| Max output | 128,000 tokens | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $5 in / $25 out per 1M tokens | https://platform.claude.com/docs/en/release-notes/overview |

## What we tested

Nothing yet. We have not run Claude Opus 4.8 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price is stated in the Claude Opus 5 release note ("the same pricing as Claude Opus 4.8").

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Nano Banana Pro

Nano Banana Pro is a Google image-generation model released May 28, 2026; Google's newest image-generation model in our dataset is Nano Banana 2.1. Proprietary, API access only. Google's gemini-3-pro-image model for image generation and editing, generally available since May 28, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/nano-banana-pro · JSON: https://theagenticweekly.com/api/models/nano-banana-pro

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | May 28, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3-pro-image` | https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-image |
| Context window | 65,536 tokens (66K) | https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-image |
| Max output | 32,768 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-image |
| Input | image, text | https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-image |
| Output | image, text | https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-image |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Nano Banana Pro ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Nano Banana Pro](https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-image)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Command A+

Command A+ is a Cohere language model released May 20, 2026; Cohere's newest language model in our dataset is North Small Translate. Open weights (Apache 2.0). Cohere's Command A+, released May 20, 2026 as a mixture-of-experts model with vision, reasoning, translation and agentic abilities, under Apache 2.0. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/command-a-plus · JSON: https://theagenticweekly.com/api/models/command-a-plus

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Cohere |  |
| Released | May 20, 2026 | https://docs.cohere.com/changelog |
| API id | `command-a-plus-05-2026` | https://docs.cohere.com/changelog |
| Context window | 128,000 tokens (128K). 128K input; 64K output. | https://docs.cohere.com/changelog |
| Max output | 64,000 tokens | https://docs.cohere.com/changelog |
| Parameters | 218B total, 25B active, mixture-of-experts | https://docs.cohere.com/changelog |
| Licence | Apache 2.0 (open weights) | https://docs.cohere.com/changelog |

## What we tested

Nothing yet. We have not run Command A+ ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price was not captured.

## Sources

- [Cohere release notes](https://docs.cohere.com/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Grok Build 0.1

Grok Build 0.1 is a xAI language model (limited access) released May 19, 2026; xAI's newest language model in our dataset is Grok 4.7. Proprietary, API access only. xAI's coding model trained for agentic coding workflows, in early access since May 19, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/grok-build-0-1 · JSON: https://theagenticweekly.com/api/models/grok-build-0-1

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | xAI |  |
| Released | May 19, 2026 | https://docs.x.ai/docs/release-notes |
| API id | `grok-build-0.1` | https://docs.x.ai/docs/release-notes |
| Context window | 256,000 tokens (256K) | https://docs.x.ai/docs/models |
| Input | text, image | https://docs.x.ai/docs/models |
| Output | text | https://docs.x.ai/docs/models |
| Licence | Proprietary (API) | https://docs.x.ai/docs/models |
| Pricing | $1 in / $2 out per 1M tokens | https://docs.x.ai/docs/models |

## What we tested

Nothing yet. We have not run Grok Build 0.1 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- The release notes describe the model as "currently in early access".

## Sources

- [xAI release notes](https://docs.x.ai/docs/release-notes)
- [xAI models and pricing](https://docs.x.ai/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.1 Flash-Lite

Gemini 3.1 Flash-Lite is a Google language model released May 7, 2026; Google's newest language model in our dataset is Gemini 4 Argon. Proprietary, API access only. Google's Gemini 3.1 Flash-Lite, generally available since May 7, 2026; a smaller, lower-latency tier than Flash. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-1-flash-lite · JSON: https://theagenticweekly.com/api/models/gemini-3-1-flash-lite

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | May 7, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.1-flash-lite` | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite |
| Context window | 1,048,576 tokens (1.05M) | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite |
| Max output | 65,536 tokens | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite |
| Input | text, image, video, audio, pdf | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite |
| Output | text | https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini 3.1 Flash-Lite ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)
- [Gemini API model page: Gemini 3.1 Flash-Lite](https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Mistral Medium 3.5

Mistral Medium 3.5 is a Mistral AI language model released Apr 28, 2026; Mistral AI's newest language model in our dataset is Mistral Large 4. Open weights (Modified MIT). Mistral's frontier-class multimodal model for agentic and coding use, released Apr 28, 2026 as open weights under a Modified MIT license. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/mistral-medium-3-5 · JSON: https://theagenticweekly.com/api/models/mistral-medium-3-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Mistral AI |  |
| Released | Apr 28, 2026 | https://docs.mistral.ai/getting-started/changelog |
| API id | `mistral-medium-3-5` | https://docs.mistral.ai/models/mistral-medium-3-5-26-04 |
| Context window | 256,000 tokens (256K) | https://docs.mistral.ai/models/mistral-medium-3-5-26-04 |
| Licence | Modified MIT (open weights) | https://docs.mistral.ai/getting-started/changelog |
| Pricing | $1.5 in / $7.5 out per 1M tokens | https://docs.mistral.ai/models/mistral-medium-3-5-26-04 |

## What we tested

Nothing yet. We have not run Mistral Medium 3.5 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- The license terms of "Modified MIT" were not captured; check the model card before commercial use.

## Sources

- [Mistral changelog](https://docs.mistral.ai/getting-started/changelog)
- [Mistral model card: Medium 3.5](https://docs.mistral.ai/models/mistral-medium-3-5-26-04)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-5.5

GPT-5.5 is an OpenAI language model released Apr 24, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's GPT-5.5 flagship, released Apr 24, 2026 with a 1.05M-token context window. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-5-5 · JSON: https://theagenticweekly.com/api/models/gpt-5-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Apr 24, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-5.5` | https://developers.openai.com/api/docs/models/gpt-5.5 |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-5.5 |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-5.5 |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-5.5 |
| Output | text | https://developers.openai.com/api/docs/models/gpt-5.5 |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-5.5 |
| Pricing | $5 in / $30 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-5.5 |

## What we tested

Nothing yet. We have not run GPT-5.5 ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-5.5](https://developers.openai.com/api/docs/models/gpt-5.5)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-5.5 Pro

GPT-5.5 Pro is an OpenAI language model released Apr 24, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's GPT-5.5 Pro, released Apr 24, 2026 for harder problems; it uses more compute and has no cached-input discount. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-5-5-pro · JSON: https://theagenticweekly.com/api/models/gpt-5-5-pro

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Apr 24, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-5.5-pro` | https://developers.openai.com/api/docs/models/gpt-5.5-pro |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-5.5-pro |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-5.5-pro |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-5.5-pro |
| Output | text | https://developers.openai.com/api/docs/models/gpt-5.5-pro |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-5.5-pro |
| Pricing | $30 in / $180 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-5.5-pro |

## What we tested

Nothing yet. We have not run GPT-5.5 Pro ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-5.5 Pro](https://developers.openai.com/api/docs/models/gpt-5.5-pro)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Qwen3.6-27B

Qwen3.6-27B is an Alibaba language model released Apr 22, 2026; Alibaba's newest language model in our dataset is Qwen3.8-27B. Open weights (Apache-2.0). Alibaba's 27B dense multimodal open-weights model under Apache 2.0, released Apr 22, 2026. It is often listed as new in late October because gateways added it then; Qwen3.8-27B (Aug 14, 2026) is the newer 27B. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/qwen3-6-27b · JSON: https://theagenticweekly.com/api/models/qwen3-6-27b

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Alibaba |  |
| Released | Apr 22, 2026 | https://github.com/QwenLM/Qwen3.8 |
| API id | `Qwen/Qwen3.6-27B` | https://huggingface.co/Qwen/Qwen3.6-27B |
| Context window | 262,144 tokens (262K). Native; extensible to 1,010,000 tokens per the model card. | https://huggingface.co/Qwen/Qwen3.6-27B |
| Parameters | 27B total, dense | https://huggingface.co/Qwen/Qwen3.6-27B |
| Licence | Apache-2.0 (open weights) | https://huggingface.co/Qwen/Qwen3.6-27B |

## What we tested

Nothing yet. We have not run Qwen3.6-27B ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Hosted API pricing was not verified: Alibaba's launch post says Model Studio API access was coming soon.
- Aggregators such as LLM Gateway list it as added on Oct 21, 2026; that is the date a gateway added it, not Alibaba's release date.

## Sources

- [Hugging Face model card](https://huggingface.co/Qwen/Qwen3.6-27B)
- [Qwen on GitHub](https://github.com/QwenLM/Qwen3.8)
- [Alibaba Cloud launch post](https://www.alibabacloud.com/blog/qwen-3-6-27b-flagship-level-coding-in-a-27b-dense-model_603063)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-Image-2

GPT-Image-2 is an OpenAI image-generation model released Apr 21, 2026; OpenAI's newest image-generation model in our dataset is GPT-Image-2.5 Flare. Proprietary, API access only. OpenAI's GPT Image 2 image generation and editing model, released Apr 21, 2026; succeeded by GPT-Image-2.5. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-image-2 · JSON: https://theagenticweekly.com/api/models/gpt-image-2

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Apr 21, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-image-2` | https://developers.openai.com/api/docs/models/gpt-image-2 |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-image-2 |
| Output | image | https://developers.openai.com/api/docs/models/gpt-image-2 |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-image-2 |

## What we tested

Nothing yet. We have not run GPT-Image-2 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Pricing was not captured; the model page lists no price table.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-Image-2](https://developers.openai.com/api/docs/models/gpt-image-2)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Opus 4.7

Claude Opus 4.7 is an Anthropic language model released Apr 16, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's Claude Opus 4.7, launched Apr 16, 2026 at $5 / $25 per million tokens; it introduced the tokenizer later reused by Claude Fable 5 and Mythos 5. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-opus-4-7 · JSON: https://theagenticweekly.com/api/models/claude-opus-4-7

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Apr 16, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-opus-4-7` | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $5 in / $25 out per 1M tokens | https://platform.claude.com/docs/en/release-notes/overview |

## What we tested

Nothing yet. We have not run Claude Opus 4.7 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window and max output were not captured for this model.

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GLM-5.1

GLM-5.1 is a Z.ai language model released Apr 7, 2026; Z.ai's newest language model in our dataset is GLM-5.3-Flash. Z.ai's GLM-5.1, released Apr 7, 2026 and designed for long-horizon tasks of up to 8 hours in a single run. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/glm-5-1 · JSON: https://theagenticweekly.com/api/models/glm-5-1

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Z.ai |  |
| Released | Apr 7, 2026 | https://docs.z.ai/release-notes/new-released |

## What we tested

Nothing yet. We have not run GLM-5.1 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window, license and price were not captured.

## Sources

- [Z.ai release notes](https://docs.z.ai/release-notes/new-released)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemma 4

Gemma 4 is a Google language model released Mar 31, 2026; Google's newest language model in our dataset is Gemini 4 Argon. Open weights (Apache 2.0). Google's open-weights Gemma 4 family in E2B, E4B, 12B, 26B A4B and 31B sizes, released from Mar 31, 2026, the easiest open family to run on your own hardware. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemma-4 · JSON: https://theagenticweekly.com/api/models/gemma-4

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Mar 31, 2026 | https://ai.google.dev/gemma/docs/releases |
| Context window | 256,000 tokens (256K). 128K for the small E2B/E4B sizes; 256K for the medium sizes. | https://ai.google.dev/gemma/docs/core |
| Input | text, image, video, audio | https://ai.google.dev/gemma/docs/core |
| Output | text | https://ai.google.dev/gemma/docs/core |
| Licence | Apache 2.0 (open weights) | https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/ |

## What we tested

Nothing yet. We have not run Gemma 4 ourselves, so every fact here is Reported from the sources below.

## Sources

- [Gemma 4 docs](https://ai.google.dev/gemma/docs/core)
- [Gemma release notes](https://ai.google.dev/gemma/docs/releases)
- [Google blog](https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Cohere Transcribe

Cohere Transcribe is a Cohere voice model released Mar 26, 2026; Cohere's newest voice model in our dataset is Cohere Transcribe Arabic. Cohere's first transcription model, an audio-in, text-out speech recognition model released Mar 26, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/cohere-transcribe · JSON: https://theagenticweekly.com/api/models/cohere-transcribe

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Cohere |  |
| Released | Mar 26, 2026 | https://docs.cohere.com/changelog |
| API id | `cohere-transcribe-03-2026` | https://docs.cohere.com/changelog |
| Input | audio | https://docs.cohere.com/changelog |
| Output | text | https://docs.cohere.com/changelog |

## What we tested

Nothing yet. We have not run Cohere Transcribe ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- License and price were not captured.

## Sources

- [Cohere release notes](https://docs.cohere.com/changelog)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Voxtral TTS

Voxtral TTS is a Mistral AI voice model released Mar 23, 2026; Mistral AI's newest voice model in our dataset is Voxtral 3. Mistral's text-to-speech model with zero-shot voice cloning and streaming, released Mar 23, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/voxtral-tts · JSON: https://theagenticweekly.com/api/models/voxtral-tts

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Mistral AI |  |
| Released | Mar 23, 2026 | https://docs.mistral.ai/getting-started/changelog |
| API id | `voxtral-mini-tts-2603` | https://docs.mistral.ai/models/voxtral-tts-26-03 |
| Output | audio | https://docs.mistral.ai/models/voxtral-tts-26-03 |
| Pricing | Speech output $16 (per 1M characters) | https://docs.mistral.ai/models/voxtral-tts-26-03 |

## What we tested

Nothing yet. We have not run Voxtral TTS ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- License and weights status were not captured.

## Sources

- [Mistral changelog](https://docs.mistral.ai/getting-started/changelog)
- [Mistral model card: Voxtral TTS](https://docs.mistral.ai/models/voxtral-tts-26-03)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-5.4 mini

GPT-5.4 mini is an OpenAI language model released Mar 17, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's GPT-5.4 mini, released Mar 17, 2026 for faster, high-volume work with a 400K context. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-5-4-mini · JSON: https://theagenticweekly.com/api/models/gpt-5-4-mini

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Mar 17, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-5.4-mini` | https://developers.openai.com/api/docs/models/gpt-5.4-mini |
| Context window | 400,000 tokens (400K) | https://developers.openai.com/api/docs/models/gpt-5.4-mini |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-5.4-mini |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-5.4-mini |
| Output | text | https://developers.openai.com/api/docs/models/gpt-5.4-mini |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-5.4-mini |
| Pricing | $0.75 in / $4.5 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-5.4-mini |

## What we tested

Nothing yet. We have not run GPT-5.4 mini ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-5.4 mini](https://developers.openai.com/api/docs/models/gpt-5.4-mini)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Mistral Small 4

Mistral Small 4 is a Mistral AI language model released Mar 16, 2026; Mistral AI's newest language model in our dataset is Mistral Large 4. Open weights (Apache 2.0). Mistral's hybrid model unifying instruct, reasoning and coding in one multimodal model, released Mar 16, 2026 under Apache 2.0. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/mistral-small-4 · JSON: https://theagenticweekly.com/api/models/mistral-small-4

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Mistral AI |  |
| Released | Mar 16, 2026 | https://docs.mistral.ai/getting-started/changelog |
| API id | `mistral-small-2603` | https://docs.mistral.ai/models/mistral-small-4-0-26-03 |
| Context window | 256,000 tokens (256K) | https://docs.mistral.ai/models/mistral-small-4-0-26-03 |
| Parameters | 119B total, 6.5B active | https://docs.mistral.ai/models/mistral-small-4-0-26-03 |
| Licence | Apache 2.0 (open weights) | https://docs.mistral.ai/models/mistral-small-4-0-26-03 |
| Pricing | $0.15 in / $0.6 out per 1M tokens | https://docs.mistral.ai/models/mistral-small-4-0-26-03 |

## What we tested

Nothing yet. We have not run Mistral Small 4 ourselves, so every fact here is Reported from the sources below.

## Sources

- [Mistral changelog](https://docs.mistral.ai/getting-started/changelog)
- [Mistral model card: Small 4](https://docs.mistral.ai/models/mistral-small-4-0-26-03)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Grok 4.20

Grok 4.20 is a xAI language model released Mar 10, 2026; xAI's newest language model in our dataset is Grok 4.7. Proprietary, API access only. xAI's Grok 4.20 (grok-4.20-0309-reasoning), made available on Mar 10, 2026 alongside a Multi-agent variant. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/grok-4-20 · JSON: https://theagenticweekly.com/api/models/grok-4-20

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | xAI |  |
| Released | Mar 10, 2026 | https://docs.x.ai/docs/release-notes |
| API id | `grok-4.20-0309-reasoning` | https://docs.x.ai/docs/models |
| Context window | 1,000,000 tokens (1M) | https://docs.x.ai/docs/models |
| Input | text, image | https://docs.x.ai/docs/models |
| Output | text | https://docs.x.ai/docs/models |
| Licence | Proprietary (API) | https://docs.x.ai/docs/models |
| Pricing | $1.25 in / $2.5 out per 1M tokens | https://docs.x.ai/docs/models |

## What we tested

Nothing yet. We have not run Grok 4.20 ourselves, so every fact here is Reported from the sources below.

## Sources

- [xAI release notes](https://docs.x.ai/docs/release-notes)
- [xAI models and pricing](https://docs.x.ai/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GPT-5.4

GPT-5.4 is an OpenAI language model released Mar 5, 2026; OpenAI's newest language model in our dataset is GPT-6.1 Terra. Proprietary, API access only. OpenAI's GPT-5.4, released Mar 5, 2026 with built-in computer use through the Responses API. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gpt-5-4 · JSON: https://theagenticweekly.com/api/models/gpt-5-4

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | OpenAI |  |
| Released | Mar 5, 2026 | https://developers.openai.com/api/docs/changelog |
| API id | `gpt-5.4` | https://developers.openai.com/api/docs/models/gpt-5.4 |
| Context window | 1,050,000 tokens (1.05M) | https://developers.openai.com/api/docs/models/gpt-5.4 |
| Max output | 128,000 tokens | https://developers.openai.com/api/docs/models/gpt-5.4 |
| Input | text, image | https://developers.openai.com/api/docs/models/gpt-5.4 |
| Output | text | https://developers.openai.com/api/docs/models/gpt-5.4 |
| Licence | Proprietary (API) | https://developers.openai.com/api/docs/models/gpt-5.4 |
| Pricing | $2.5 in / $15 out per 1M tokens | https://developers.openai.com/api/docs/models/gpt-5.4 |

## What we tested

Nothing yet. We have not run GPT-5.4 ourselves, so every fact here is Reported from the sources below.

## Sources

- [OpenAI API changelog](https://developers.openai.com/api/docs/changelog)
- [OpenAI model page: GPT-5.4](https://developers.openai.com/api/docs/models/gpt-5.4)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Gemini 3.1 Pro

Gemini 3.1 Pro is a Google language model (public preview) released Feb 19, 2026; Google's newest language model in our dataset is Gemini 4 Argon. Proprietary, API access only. Google's Gemini 3.1 Pro, released in preview on Feb 19, 2026 under the gemini-3.1-pro-preview endpoint. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/gemini-3-1-pro · JSON: https://theagenticweekly.com/api/models/gemini-3-1-pro

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Google |  |
| Released | Feb 19, 2026 | https://ai.google.dev/gemini-api/docs/changelog |
| API id | `gemini-3.1-pro-preview` | https://ai.google.dev/gemini-api/docs/models |
| Licence | Proprietary (API) | https://ai.google.dev/gemini-api/docs/models |

## What we tested

Nothing yet. We have not run Gemini 3.1 Pro ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window, output limits and modalities were not captured.

## Sources

- [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog)
- [Gemini API models list](https://ai.google.dev/gemini-api/docs/models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Sonnet 4.6

Claude Sonnet 4.6 is an Anthropic language model released Feb 17, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's Claude Sonnet 4.6, launched Feb 17, 2026, the predecessor of Claude Sonnet 5; a 1M-token context was offered in beta. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-sonnet-4-6 · JSON: https://theagenticweekly.com/api/models/claude-sonnet-4-6

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Feb 17, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-sonnet-4-6` | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |

## What we tested

Nothing yet. We have not run Claude Sonnet 4.6 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- The 1M-token context window is described as beta in the release notes, so no standard context value is recorded.
- Price and max output were not captured for this model.

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GLM-5

GLM-5 is a Z.ai language model released Feb 12, 2026; Z.ai's newest language model in our dataset is GLM-5.3-Flash. Z.ai's GLM-5, released Feb 12, 2026 for complex system engineering and long-range agent tasks. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/glm-5 · JSON: https://theagenticweekly.com/api/models/glm-5

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Z.ai |  |
| Released | Feb 12, 2026 | https://docs.z.ai/release-notes/new-released |

## What we tested

Nothing yet. We have not run GLM-5 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Context window, license and price were not captured.

## Sources

- [Z.ai release notes](https://docs.z.ai/release-notes/new-released)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Claude Opus 4.6

Claude Opus 4.6 is an Anthropic language model released Feb 5, 2026; Anthropic's newest language model in our dataset is Claude Haiku 5.5. Proprietary, API access only. Anthropic's Claude Opus 4.6, launched Feb 5, 2026 at $5 / $25 per million tokens, the predecessor of Claude Opus 4.7. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/claude-opus-4-6 · JSON: https://theagenticweekly.com/api/models/claude-opus-4-6

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Anthropic |  |
| Released | Feb 5, 2026 | https://platform.claude.com/docs/en/release-notes/overview |
| API id | `claude-opus-4-6` | https://platform.claude.com/docs/en/release-notes/overview |
| Licence | Proprietary (API) | https://platform.claude.com/docs/en/about-claude/models/overview |
| Pricing | $5 in / $25 out per 1M tokens | https://platform.claude.com/docs/en/release-notes/overview |

## What we tested

Nothing yet. We have not run Claude Opus 4.6 ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- Price is stated in the Claude Opus 4.7 release note ("the same $5 / $25 USD per MTok pricing as Opus 4.6").
- Context window and max output were not captured for this model.

## Sources

- [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview)
- [Anthropic models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# GLM-Image

GLM-Image is Z.ai's latest image-generation model in our dataset, released Jan 14, 2026. Z.ai's GLM-Image, an image generation model combining autoregressive semantic understanding with diffusion decoding, launched Jan 14, 2026. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/glm-image · JSON: https://theagenticweekly.com/api/models/glm-image

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Z.ai |  |
| Released | Jan 14, 2026 | https://docs.z.ai/release-notes/new-released |
| Output | image | https://docs.z.ai/release-notes/new-released |

## What we tested

Nothing yet. We have not run GLM-Image ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- License and price were not captured.

## Sources

- [Z.ai release notes](https://docs.z.ai/release-notes/new-released)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Ministral 3 14B

Ministral 3 14B is a Mistral AI language model released Dec 2, 2025; Mistral AI's newest language model in our dataset is Mistral Large 4. Open weights (Apache 2.0). Mistral's largest Ministral 3 model, released Dec 2, 2025 under Apache 2.0 and optimized for local deployment. Verified against vendor sources on Oct 26, 2026.

Last verified: 2026-10-26 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/ministral-3-14b · JSON: https://theagenticweekly.com/api/models/ministral-3-14b

## Key facts

| Fact | Value | Source |
|---|---|---|
| Vendor | Mistral AI |  |
| Released | Dec 2, 2025 | https://docs.mistral.ai/getting-started/changelog |
| API id | `ministral-14b-2512` | https://docs.mistral.ai/models/ministral-3-14b-25-12 |
| Context window | 256,000 tokens (256K) | https://docs.mistral.ai/models/ministral-3-14b-25-12 |
| Licence | Apache 2.0 (open weights) | https://docs.mistral.ai/models/ministral-3-14b-25-12 |
| Pricing | $0.2 in / $0.2 out per 1M tokens | https://docs.mistral.ai/models/ministral-3-14b-25-12 |

## What we tested

Nothing yet. We have not run Ministral 3 14B ourselves, so every fact here is Reported from the sources below.

## Sources

- [Mistral changelog](https://docs.mistral.ai/getting-started/changelog)
- [Mistral model card: Ministral 3 14B](https://docs.mistral.ai/models/ministral-3-14b-25-12)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.

---

# Comparisons

- GPT-6.1 Terra vs Claude Opus 5.5: price per task, context and when Opus still fits: https://theagenticweekly.com/compare/gpt-6-1-terra-vs-claude-opus-5-5 (markdown: https://theagenticweekly.com/compare/gpt-6-1-terra-vs-claude-opus-5-5.md). GPT-6.1 Terra lists at one quarter of Claude Opus 5.5's per-token price ($1/$5 vs $4/$20). Specs, access and vendor claims, read 2026-10-26.
- GPT-6.1 Terra vs GPT-6.1 Sol: price, specs and which to pick: https://theagenticweekly.com/compare/gpt-6-1-terra-vs-gpt-6-1-sol (markdown: https://theagenticweekly.com/compare/gpt-6-1-terra-vs-gpt-6-1-sol.md). GPT-6.1 Terra costs half of Sol per token ($1/$5 vs $2/$10). Specs, Codex and Copilot access, and when to pick each, from vendor pages read 2026-10-26.
- ChatGPT vs ComfyUI: which should you use?: https://theagenticweekly.com/compare/chatgpt-vs-comfyui (markdown: https://theagenticweekly.com/compare/chatgpt-vs-comfyui.md)
- Codex CLI vs Gemini CLI: which should you use?: https://theagenticweekly.com/compare/codex-cli-vs-gemini-cli (markdown: https://theagenticweekly.com/compare/codex-cli-vs-gemini-cli.md)
- ComfyUI vs Midjourney: which should you use?: https://theagenticweekly.com/compare/comfyui-vs-midjourney (markdown: https://theagenticweekly.com/compare/comfyui-vs-midjourney.md)
- Goodcall vs Jobber Receptionist: which should you use?: https://theagenticweekly.com/compare/goodcall-vs-jobber (markdown: https://theagenticweekly.com/compare/goodcall-vs-jobber.md)
- Activepieces vs Pipedream: which should you use?: https://theagenticweekly.com/compare/activepieces-vs-pipedream (markdown: https://theagenticweekly.com/compare/activepieces-vs-pipedream.md). Activepieces vs Pipedream: Activepieces open source vs Pipedream closed. Sourced facts from vendor pages, read Oct 25, 2026.
- Augment Code vs Devin: which should you use?: https://theagenticweekly.com/compare/augment-code-vs-devin (markdown: https://theagenticweekly.com/compare/augment-code-vs-devin.md). Augment Code vs Devin: ai coding agents compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- AutoGen vs Agno: which should you use?: https://theagenticweekly.com/compare/autogen-vs-agno (markdown: https://theagenticweekly.com/compare/autogen-vs-agno.md). AutoGen vs Agno: Free and open source vs Open source + platform; MIT vs Apache-2.0 license. Sourced facts from vendor pages, read Oct 25, 2026.
- AutoGen vs Semantic Kernel: which should you use?: https://theagenticweekly.com/compare/autogen-vs-semantic-kernel (markdown: https://theagenticweekly.com/compare/autogen-vs-semantic-kernel.md). AutoGen vs Semantic Kernel: MIT vs MIT license; 61,562 vs 28,695 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Beautiful.ai vs Canva AI: which should you use?: https://theagenticweekly.com/compare/beautiful-ai-vs-canva-ai (markdown: https://theagenticweekly.com/compare/beautiful-ai-vs-canva-ai.md). Beautiful.ai vs Canva AI: ai design and presentations compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Bolt vs Firebase Studio: which should you use?: https://theagenticweekly.com/compare/bolt-new-vs-firebase-studio (markdown: https://theagenticweekly.com/compare/bolt-new-vs-firebase-studio.md). Bolt vs Firebase Studio: vibe coding compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Claude Code vs Codex CLI: which should you use?: https://theagenticweekly.com/compare/claude-code-vs-codex-cli (markdown: https://theagenticweekly.com/compare/claude-code-vs-codex-cli.md). Claude Code vs Codex CLI: both smoke-tested on 2026-10-25 with a local model. Install routes, licenses, latest releases, surfaces and what we did not test.
- Claude Code vs Cursor: which should you use?: https://theagenticweekly.com/compare/claude-code-vs-cursor (markdown: https://theagenticweekly.com/compare/claude-code-vs-cursor.md). Claude Code vs Cursor compared from official sources: where each runs, pricing plans read on 2026-10-25, models, and what we did and did not test.
- Claude Code vs Gemini CLI: which should you use?: https://theagenticweekly.com/compare/claude-code-vs-gemini-cli (markdown: https://theagenticweekly.com/compare/claude-code-vs-gemini-cli.md). Claude Code vs Gemini CLI: Gemini CLI open source vs Claude Code closed; 153,364 vs 108,386 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- ComfyUI vs FLUX (Black Forest Labs): which should you use?: https://theagenticweekly.com/compare/comfyui-vs-black-forest-labs (markdown: https://theagenticweekly.com/compare/comfyui-vs-black-forest-labs.md). ComfyUI vs FLUX (Black Forest Labs): FLUX (Black Forest Labs) last push Jul 31, 2025. Sourced facts from vendor pages, read Oct 25, 2026.
- Composio vs Browser Use: which should you use?: https://theagenticweekly.com/compare/composio-vs-browser-use (markdown: https://theagenticweekly.com/compare/composio-vs-browser-use.md). Composio vs Browser Use: Freemium vs Pay as you go; MIT vs MIT license; 119,131 vs 30,747 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Copy.ai vs Writesonic: which should you use?: https://theagenticweekly.com/compare/copy-ai-vs-writesonic (markdown: https://theagenticweekly.com/compare/copy-ai-vs-writesonic.md). Copy.ai vs Writesonic: writing compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- CrewAI vs AutoGen: which should you use?: https://theagenticweekly.com/compare/crewai-vs-autogen (markdown: https://theagenticweekly.com/compare/crewai-vs-autogen.md). CrewAI vs AutoGen: MIT vs MIT license; 61,562 vs 60,100 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Cursor vs GitHub Copilot: which should you use?: https://theagenticweekly.com/compare/cursor-vs-github-copilot (markdown: https://theagenticweekly.com/compare/cursor-vs-github-copilot.md). Cursor vs GitHub Copilot compared from official pages read on 2026-10-25: where each runs, plan prices, how usage is billed, limits, and what we did not test.
- Cursor vs Windsurf (now Devin Desktop): which should you use?: https://theagenticweekly.com/compare/cursor-vs-windsurf (markdown: https://theagenticweekly.com/compare/cursor-vs-windsurf.md). Cursor vs Windsurf (now Devin Desktop): ai coding agents compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Firecrawl vs Context7: which should you use?: https://theagenticweekly.com/compare/firecrawl-vs-context7 (markdown: https://theagenticweekly.com/compare/firecrawl-vs-context7.md). Firecrawl vs Context7: AGPL-3.0 vs MIT license; Context7 has a public repo (63,676 stars). Sourced facts from vendor pages, read Oct 25, 2026.
- Flowise vs Langflow: which should you use?: https://theagenticweekly.com/compare/flowise-vs-langflow (markdown: https://theagenticweekly.com/compare/flowise-vs-langflow.md). Flowise vs Langflow: Flowise repo archived; Open source (archived) vs Open source. Sourced facts from vendor pages, read Oct 25, 2026.
- GitHub Copilot vs Qodo: which should you use?: https://theagenticweekly.com/compare/github-copilot-vs-qodo (markdown: https://theagenticweekly.com/compare/github-copilot-vs-qodo.md). GitHub Copilot vs Qodo: ai coding agents compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- goose vs opencode: which should you use?: https://theagenticweekly.com/compare/goose-vs-opencode (markdown: https://theagenticweekly.com/compare/goose-vs-opencode.md). goose vs opencode: two open-source coding agents, both smoke-tested on 2026-10-25 with a local model. Licenses, latest releases, install routes and what we did not test.
- Grammarly vs Jasper: which should you use?: https://theagenticweekly.com/compare/grammarly-vs-jasper (markdown: https://theagenticweekly.com/compare/grammarly-vs-jasper.md). Grammarly vs Jasper: writing compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Gumloop vs Lindy: which should you use?: https://theagenticweekly.com/compare/gumloop-vs-lindy (markdown: https://theagenticweekly.com/compare/gumloop-vs-lindy.md). Gumloop vs Lindy: agent building compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Ideogram vs Leonardo.Ai: which should you use?: https://theagenticweekly.com/compare/ideogram-vs-leonardo-ai (markdown: https://theagenticweekly.com/compare/ideogram-vs-leonardo-ai.md). Ideogram vs Leonardo.Ai: image generation compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Ideogram vs Recraft: which should you use?: https://theagenticweekly.com/compare/ideogram-vs-recraft (markdown: https://theagenticweekly.com/compare/ideogram-vs-recraft.md). Ideogram vs Recraft: image generation compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Jan vs GPT4All: which should you use?: https://theagenticweekly.com/compare/jan-vs-gpt4all (markdown: https://theagenticweekly.com/compare/jan-vs-gpt4all.md). Jan vs GPT4All: Free and open source vs Free; GPT4All last push May 27, 2025; Apache-2.0 vs MIT license. Sourced facts from vendor pages, read Oct 25, 2026.
- Kiro vs Amazon Q Developer: which should you use?: https://theagenticweekly.com/compare/kiro-vs-amazon-q-developer (markdown: https://theagenticweekly.com/compare/kiro-vs-amazon-q-developer.md). Kiro vs Amazon Q Developer: ai coding agents compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Kling AI vs Luma: which should you use?: https://theagenticweekly.com/compare/kling-vs-luma (markdown: https://theagenticweekly.com/compare/kling-vs-luma.md). Kling AI vs Luma: ai video compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- LangChain vs LlamaIndex: which should you use?: https://theagenticweekly.com/compare/langchain-vs-llamaindex (markdown: https://theagenticweekly.com/compare/langchain-vs-llamaindex.md). LangChain vs LlamaIndex: agent building compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- LangGraph vs CrewAI: which should you use?: https://theagenticweekly.com/compare/langgraph-vs-crewai (markdown: https://theagenticweekly.com/compare/langgraph-vs-crewai.md). LangGraph vs CrewAI, checked against official docs and GitHub on 2026-10-25: how each models agents, licenses, hosting, and which to pick for what.
- Letta vs Mem0: which should you use?: https://theagenticweekly.com/compare/letta-vs-mem0 (markdown: https://theagenticweekly.com/compare/letta-vs-mem0.md). Letta vs Mem0: agent building compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- llama.cpp vs GPT4All: which should you use?: https://theagenticweekly.com/compare/llama-cpp-vs-gpt4all (markdown: https://theagenticweekly.com/compare/llama-cpp-vs-gpt4all.md). llama.cpp vs GPT4All: Free and open source vs Free; GPT4All last push May 27, 2025; 132,360 vs 77,357 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- LlamaIndex vs Haystack: which should you use?: https://theagenticweekly.com/compare/llamaindex-vs-haystack (markdown: https://theagenticweekly.com/compare/llamaindex-vs-haystack.md). LlamaIndex vs Haystack: MIT vs Apache-2.0 license; 52,662 vs 26,842 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Lovable vs Replit: which should you use?: https://theagenticweekly.com/compare/lovable-vs-replit (markdown: https://theagenticweekly.com/compare/lovable-vs-replit.md). Lovable vs Replit compared from vendor docs read on 2026-10-25: how credits and plans work, what the free tiers allow, prices, code access and publishing.
- MCP reference servers vs GitHub MCP Server: which should you use?: https://theagenticweekly.com/compare/mcp-servers-vs-github-mcp-server (markdown: https://theagenticweekly.com/compare/mcp-servers-vs-github-mcp-server.md). MCP reference servers vs GitHub MCP Server: MIT to Apache-2.0… vs MIT license; 91,561 vs 33,952 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Mem0 vs LangChain: which should you use?: https://theagenticweekly.com/compare/mem0-vs-langchain (markdown: https://theagenticweekly.com/compare/mem0-vs-langchain.md). Mem0 vs LangChain: Freemium vs Free and open source; Apache-2.0 vs MIT license; 148,451 vs 67,804 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Microsoft Copilot vs Notion AI: which should you use?: https://theagenticweekly.com/compare/microsoft-copilot-vs-notion-ai (markdown: https://theagenticweekly.com/compare/microsoft-copilot-vs-notion-ai.md). Microsoft Copilot vs Notion AI: ai assistants compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Ollama vs llama.cpp: which should you use?: https://theagenticweekly.com/compare/ollama-vs-llama-cpp (markdown: https://theagenticweekly.com/compare/ollama-vs-llama-cpp.md). Ollama vs llama.cpp: Free + paid cloud plans vs Free and open source; 184,762 vs 132,360 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Ollama vs Open WebUI: which should you use?: https://theagenticweekly.com/compare/ollama-vs-open-webui (markdown: https://theagenticweekly.com/compare/ollama-vs-open-webui.md). Ollama vs Open WebUI: Ollama open source vs Open WebUI closed; Free + paid cloud plans vs Free self-hosted. Sourced facts from vendor pages, read Oct 25, 2026.
- OpenHands vs Devin: which should you use?: https://theagenticweekly.com/compare/openhands-vs-devin (markdown: https://theagenticweekly.com/compare/openhands-vs-devin.md). OpenHands vs Devin: OpenHands open source vs Devin closed. Sourced facts from vendor pages, read Oct 25, 2026.
- OpusClip vs Descript: which should you use?: https://theagenticweekly.com/compare/opus-clip-vs-descript (markdown: https://theagenticweekly.com/compare/opus-clip-vs-descript.md). OpusClip vs Descript: ai video compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Perplexity vs Elicit: which should you use?: https://theagenticweekly.com/compare/perplexity-vs-elicit (markdown: https://theagenticweekly.com/compare/perplexity-vs-elicit.md). Perplexity vs Elicit: ai search and research compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Perplexity vs You.com: which should you use?: https://theagenticweekly.com/compare/perplexity-vs-you-com (markdown: https://theagenticweekly.com/compare/perplexity-vs-you-com.md). Perplexity vs You.com: ai search and research compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Playwright MCP vs Chrome DevTools MCP: which should you use?: https://theagenticweekly.com/compare/playwright-mcp-vs-chrome-devtools-mcp (markdown: https://theagenticweekly.com/compare/playwright-mcp-vs-chrome-devtools-mcp.md). Playwright MCP vs Chrome DevTools MCP: mcp servers compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Pydantic AI vs Mastra: which should you use?: https://theagenticweekly.com/compare/pydantic-ai-vs-mastra (markdown: https://theagenticweekly.com/compare/pydantic-ai-vs-mastra.md). Pydantic AI vs Mastra: Pydantic AI open source vs Mastra closed; 29,073 vs 20,868 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Qodo vs Amazon Q Developer: which should you use?: https://theagenticweekly.com/compare/qodo-vs-amazon-q-developer (markdown: https://theagenticweekly.com/compare/qodo-vs-amazon-q-developer.md). Qodo vs Amazon Q Developer: ai coding agents compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Semantic Kernel vs Google ADK: which should you use?: https://theagenticweekly.com/compare/semantic-kernel-vs-google-adk (markdown: https://theagenticweekly.com/compare/semantic-kernel-vs-google-adk.md). Semantic Kernel vs Google ADK: MIT vs Apache-2.0 license; 28,695 vs 22,006 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- smolagents vs Pydantic AI: which should you use?: https://theagenticweekly.com/compare/smolagents-vs-pydantic-ai (markdown: https://theagenticweekly.com/compare/smolagents-vs-pydantic-ai.md). smolagents vs Pydantic AI: Apache-2.0 vs MIT license; 30,051 vs 20,868 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Speechify vs Resemble AI: which should you use?: https://theagenticweekly.com/compare/speechify-vs-resemble-ai (markdown: https://theagenticweekly.com/compare/speechify-vs-resemble-ai.md). Speechify vs Resemble AI: ai voice and speech compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Supabase MCP vs Notion MCP Server: which should you use?: https://theagenticweekly.com/compare/supabase-mcp-vs-notion-mcp (markdown: https://theagenticweekly.com/compare/supabase-mcp-vs-notion-mcp.md). Supabase MCP vs Notion MCP Server: Apache-2.0 vs MIT license; 4,678 vs 2,969 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Together AI vs Replicate: which should you use?: https://theagenticweekly.com/compare/together-ai-vs-replicate (markdown: https://theagenticweekly.com/compare/together-ai-vs-replicate.md). Together AI vs Replicate: local llms and model platforms compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Vapi vs Retell AI: which should you use?: https://theagenticweekly.com/compare/vapi-vs-retell (markdown: https://theagenticweekly.com/compare/vapi-vs-retell.md). Vapi vs Retell AI: ai voice and speech compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Vercel AI SDK vs OpenAI Agents SDK: which should you use?: https://theagenticweekly.com/compare/vercel-ai-sdk-vs-openai-agents-sdk (markdown: https://theagenticweekly.com/compare/vercel-ai-sdk-vs-openai-agents-sdk.md). Vercel AI SDK vs OpenAI Agents SDK: Apache-2.0 vs MIT license; 30,279 vs 27,511 stars. Sourced facts from vendor pages, read Oct 25, 2026.
- Voiceflow vs Botpress: which should you use?: https://theagenticweekly.com/compare/voiceflow-vs-botpress (markdown: https://theagenticweekly.com/compare/voiceflow-vs-botpress.md). Voiceflow vs Botpress: agent building compared on license, pricing, platforms and latest release. Sourced from vendor pages read Oct 25, 2026.
- Zed vs Kilo Code: which should you use?: https://theagenticweekly.com/compare/zed-vs-kilo-code (markdown: https://theagenticweekly.com/compare/zed-vs-kilo-code.md). Zed vs Kilo Code: Freemium vs Free and open source with usage-based inference. Sourced facts from vendor pages, read Oct 25, 2026.

# Alternatives pages

- Alternatives to Base44: https://theagenticweekly.com/alternatives/base44 (markdown: https://theagenticweekly.com/alternatives/base44.md)
- Alternatives to Bolt: https://theagenticweekly.com/alternatives/bolt-new (markdown: https://theagenticweekly.com/alternatives/bolt-new.md)
- Alternatives to Canva AI: https://theagenticweekly.com/alternatives/canva-ai (markdown: https://theagenticweekly.com/alternatives/canva-ai.md)
- Alternatives to Copy.ai: https://theagenticweekly.com/alternatives/copy-ai (markdown: https://theagenticweekly.com/alternatives/copy-ai.md)
- Alternatives to Deepgram: https://theagenticweekly.com/alternatives/deepgram (markdown: https://theagenticweekly.com/alternatives/deepgram.md)
- Alternatives to Descript: https://theagenticweekly.com/alternatives/descript (markdown: https://theagenticweekly.com/alternatives/descript.md)
- Alternatives to Devin: https://theagenticweekly.com/alternatives/devin (markdown: https://theagenticweekly.com/alternatives/devin.md)
- Alternatives to Firecrawl: https://theagenticweekly.com/alternatives/firecrawl (markdown: https://theagenticweekly.com/alternatives/firecrawl.md)
- Alternatives to Fireflies.ai: https://theagenticweekly.com/alternatives/fireflies (markdown: https://theagenticweekly.com/alternatives/fireflies.md)
- Alternatives to Goodcall: https://theagenticweekly.com/alternatives/goodcall (markdown: https://theagenticweekly.com/alternatives/goodcall.md)
- Alternatives to Grammarly: https://theagenticweekly.com/alternatives/grammarly (markdown: https://theagenticweekly.com/alternatives/grammarly.md)
- Alternatives to Granola: https://theagenticweekly.com/alternatives/granola (markdown: https://theagenticweekly.com/alternatives/granola.md)
- Alternatives to Grok: https://theagenticweekly.com/alternatives/grok (markdown: https://theagenticweekly.com/alternatives/grok.md)
- Alternatives to HeyGen: https://theagenticweekly.com/alternatives/heygen (markdown: https://theagenticweekly.com/alternatives/heygen.md)
- Alternatives to Hugging Face: https://theagenticweekly.com/alternatives/hugging-face (markdown: https://theagenticweekly.com/alternatives/hugging-face.md)
- Alternatives to Ideogram: https://theagenticweekly.com/alternatives/ideogram (markdown: https://theagenticweekly.com/alternatives/ideogram.md)
- Alternatives to Jasper: https://theagenticweekly.com/alternatives/jasper (markdown: https://theagenticweekly.com/alternatives/jasper.md)
- Alternatives to Jobber Receptionist: https://theagenticweekly.com/alternatives/jobber (markdown: https://theagenticweekly.com/alternatives/jobber.md)
- Alternatives to Kiro: https://theagenticweekly.com/alternatives/kiro (markdown: https://theagenticweekly.com/alternatives/kiro.md)
- Alternatives to Krea: https://theagenticweekly.com/alternatives/krea (markdown: https://theagenticweekly.com/alternatives/krea.md)
- Alternatives to Leonardo.Ai: https://theagenticweekly.com/alternatives/leonardo-ai (markdown: https://theagenticweekly.com/alternatives/leonardo-ai.md)
- Alternatives to LM Studio: https://theagenticweekly.com/alternatives/lm-studio (markdown: https://theagenticweekly.com/alternatives/lm-studio.md)
- Alternatives to Make: https://theagenticweekly.com/alternatives/make (markdown: https://theagenticweekly.com/alternatives/make.md)
- Alternatives to Mistral Le Chat: https://theagenticweekly.com/alternatives/le-chat (markdown: https://theagenticweekly.com/alternatives/le-chat.md)
- Alternatives to Murf AI: https://theagenticweekly.com/alternatives/murf-ai (markdown: https://theagenticweekly.com/alternatives/murf-ai.md)
- Alternatives to Notion AI: https://theagenticweekly.com/alternatives/notion-ai (markdown: https://theagenticweekly.com/alternatives/notion-ai.md)
- Alternatives to Ollama: https://theagenticweekly.com/alternatives/ollama (markdown: https://theagenticweekly.com/alternatives/ollama.md)
- Alternatives to Open WebUI: https://theagenticweekly.com/alternatives/open-webui (markdown: https://theagenticweekly.com/alternatives/open-webui.md)
- Alternatives to OpenRouter: https://theagenticweekly.com/alternatives/openrouter (markdown: https://theagenticweekly.com/alternatives/openrouter.md)
- Alternatives to Otter.ai: https://theagenticweekly.com/alternatives/otter (markdown: https://theagenticweekly.com/alternatives/otter.md)
- Alternatives to Perplexity: https://theagenticweekly.com/alternatives/perplexity (markdown: https://theagenticweekly.com/alternatives/perplexity.md)
- Alternatives to Pika: https://theagenticweekly.com/alternatives/pika (markdown: https://theagenticweekly.com/alternatives/pika.md)
- Alternatives to Pipedream: https://theagenticweekly.com/alternatives/pipedream (markdown: https://theagenticweekly.com/alternatives/pipedream.md)
- Alternatives to Synthesia: https://theagenticweekly.com/alternatives/synthesia (markdown: https://theagenticweekly.com/alternatives/synthesia.md)
- Alternatives to Vapi: https://theagenticweekly.com/alternatives/vapi (markdown: https://theagenticweekly.com/alternatives/vapi.md)
- Alternatives to Warp: https://theagenticweekly.com/alternatives/warp (markdown: https://theagenticweekly.com/alternatives/warp.md)
- Alternatives to Writesonic: https://theagenticweekly.com/alternatives/writesonic (markdown: https://theagenticweekly.com/alternatives/writesonic.md)
- Alternatives to Zapier: https://theagenticweekly.com/alternatives/zapier (markdown: https://theagenticweekly.com/alternatives/zapier.md)
- Alternatives to Zed: https://theagenticweekly.com/alternatives/zed (markdown: https://theagenticweekly.com/alternatives/zed.md)
- Alternatives to LangGraph: https://theagenticweekly.com/alternatives/langgraph (markdown: https://theagenticweekly.com/alternatives/langgraph.md)
- Alternatives to Activepieces: https://theagenticweekly.com/alternatives/activepieces (markdown: https://theagenticweekly.com/alternatives/activepieces.md)
- Alternatives to Claude Code: https://theagenticweekly.com/alternatives/claude-code (markdown: https://theagenticweekly.com/alternatives/claude-code.md)
- Alternatives to Codex CLI: https://theagenticweekly.com/alternatives/codex-cli (markdown: https://theagenticweekly.com/alternatives/codex-cli.md)
- Alternatives to CrewAI: https://theagenticweekly.com/alternatives/crewai (markdown: https://theagenticweekly.com/alternatives/crewai.md)
- Alternatives to Dify: https://theagenticweekly.com/alternatives/dify (markdown: https://theagenticweekly.com/alternatives/dify.md)
- Alternatives to Gemini Notebook: https://theagenticweekly.com/alternatives/gemini-notebook (markdown: https://theagenticweekly.com/alternatives/gemini-notebook.md)
- Alternatives to goose: https://theagenticweekly.com/alternatives/goose (markdown: https://theagenticweekly.com/alternatives/goose.md)
- Alternatives to Luma: https://theagenticweekly.com/alternatives/luma (markdown: https://theagenticweekly.com/alternatives/luma.md)
- Alternatives to OpenHands: https://theagenticweekly.com/alternatives/openhands (markdown: https://theagenticweekly.com/alternatives/openhands.md)
- Alternatives to Replit: https://theagenticweekly.com/alternatives/replit (markdown: https://theagenticweekly.com/alternatives/replit.md)
- Alternatives to Google ADK: https://theagenticweekly.com/alternatives/google-adk (markdown: https://theagenticweekly.com/alternatives/google-adk.md)
- Alternatives to LangChain: https://theagenticweekly.com/alternatives/langchain (markdown: https://theagenticweekly.com/alternatives/langchain.md)
- Alternatives to opencode: https://theagenticweekly.com/alternatives/opencode (markdown: https://theagenticweekly.com/alternatives/opencode.md)
- Alternatives to Claude: https://theagenticweekly.com/alternatives/claude (markdown: https://theagenticweekly.com/alternatives/claude.md)
- Alternatives to Haystack: https://theagenticweekly.com/alternatives/haystack (markdown: https://theagenticweekly.com/alternatives/haystack.md)
- Alternatives to Mastra: https://theagenticweekly.com/alternatives/mastra (markdown: https://theagenticweekly.com/alternatives/mastra.md)
- Alternatives to ChatGPT: https://theagenticweekly.com/alternatives/chatgpt (markdown: https://theagenticweekly.com/alternatives/chatgpt.md)
- Alternatives to FastMCP: https://theagenticweekly.com/alternatives/fastmcp (markdown: https://theagenticweekly.com/alternatives/fastmcp.md)
- Alternatives to Gemini CLI: https://theagenticweekly.com/alternatives/gemini-cli (markdown: https://theagenticweekly.com/alternatives/gemini-cli.md)
- Alternatives to Jan: https://theagenticweekly.com/alternatives/jan (markdown: https://theagenticweekly.com/alternatives/jan.md)
- Alternatives to n8n: https://theagenticweekly.com/alternatives/n8n (markdown: https://theagenticweekly.com/alternatives/n8n.md)
- Alternatives to ElevenLabs: https://theagenticweekly.com/alternatives/elevenlabs (markdown: https://theagenticweekly.com/alternatives/elevenlabs.md)
- Alternatives to v0: https://theagenticweekly.com/alternatives/v0 (markdown: https://theagenticweekly.com/alternatives/v0.md)
- Alternatives to ComfyUI: https://theagenticweekly.com/alternatives/comfyui (markdown: https://theagenticweekly.com/alternatives/comfyui.md)
- Alternatives to Cline: https://theagenticweekly.com/alternatives/cline (markdown: https://theagenticweekly.com/alternatives/cline.md)
- Alternatives to Cursor: https://theagenticweekly.com/alternatives/cursor (markdown: https://theagenticweekly.com/alternatives/cursor.md)
- Alternatives to Browser Use: https://theagenticweekly.com/alternatives/browser-use (markdown: https://theagenticweekly.com/alternatives/browser-use.md)
- Alternatives to GitHub Copilot: https://theagenticweekly.com/alternatives/github-copilot (markdown: https://theagenticweekly.com/alternatives/github-copilot.md)
- Alternatives to Kling AI: https://theagenticweekly.com/alternatives/kling (markdown: https://theagenticweekly.com/alternatives/kling.md)
- Alternatives to OpenAI Agents SDK: https://theagenticweekly.com/alternatives/openai-agents-sdk (markdown: https://theagenticweekly.com/alternatives/openai-agents-sdk.md)
- Alternatives to Runway: https://theagenticweekly.com/alternatives/runway (markdown: https://theagenticweekly.com/alternatives/runway.md)
- Alternatives to Lovable: https://theagenticweekly.com/alternatives/lovable (markdown: https://theagenticweekly.com/alternatives/lovable.md)
- Alternatives to Midjourney: https://theagenticweekly.com/alternatives/midjourney (markdown: https://theagenticweekly.com/alternatives/midjourney.md)
- Alternatives to Semantic Kernel: https://theagenticweekly.com/alternatives/semantic-kernel (markdown: https://theagenticweekly.com/alternatives/semantic-kernel.md)
- Alternatives to LlamaIndex: https://theagenticweekly.com/alternatives/llamaindex (markdown: https://theagenticweekly.com/alternatives/llamaindex.md)
- Alternatives to Suno: https://theagenticweekly.com/alternatives/suno (markdown: https://theagenticweekly.com/alternatives/suno.md)
- Alternatives to Flowise: https://theagenticweekly.com/alternatives/flowise (markdown: https://theagenticweekly.com/alternatives/flowise.md)
- Alternatives to Windsurf (now Devin Desktop): https://theagenticweekly.com/alternatives/windsurf (markdown: https://theagenticweekly.com/alternatives/windsurf.md)
- Alternatives to Aider: https://theagenticweekly.com/alternatives/aider (markdown: https://theagenticweekly.com/alternatives/aider.md)
- Alternatives to ACE-Step: https://theagenticweekly.com/alternatives/ace-step (markdown: https://theagenticweekly.com/alternatives/ace-step.md)
- Alternatives to AutoGen: https://theagenticweekly.com/alternatives/autogen (markdown: https://theagenticweekly.com/alternatives/autogen.md)

# Explainers and pricing pages

- What is agentic AI? Definition, how it works and examples (2026): https://theagenticweekly.com/what-is/agentic-ai (markdown: https://theagenticweekly.com/what-is/agentic-ai.md). Agentic AI means AI systems that set goals, plan and act with tools and little human input. Definition, components, vs generative AI and AI agents, read 2026-10-25.
- What is an AI coding agent? How it differs from autocomplete (2026): https://theagenticweekly.com/what-is/ai-coding-agent (markdown: https://theagenticweekly.com/what-is/ai-coding-agent.md). An AI coding agent reads your codebase, edits files, runs commands and tests, and iterates. How it differs from autocomplete, with examples and limits. Read 2026-10-25.
- What is Aider? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/aider (markdown: https://theagenticweekly.com/what-is/aider.md). What is Aider? Listed under AI coding agents, from Aider-AI; license: Apache-2.0; 49,752 GitHub stars. Facts from official pages read Oct 25, 2026.
- What is Browser Use? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/browser-use (markdown: https://theagenticweekly.com/what-is/browser-use.md). What is Browser Use? Listed under Agent building, from Browser Use; license: MIT; 119,131 GitHub stars. Facts from official pages read Oct 25, 2026.
- Cartesia pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/cartesia-pricing (markdown: https://theagenticweekly.com/what-is/cartesia-pricing.md). Cartesia pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- Deepgram pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/deepgram-pricing (markdown: https://theagenticweekly.com/what-is/deepgram-pricing.md). Deepgram pricing as of Oct 25, 2026 (Pay as you go): plan table, free tier and what the vendor's page does not say.
- Devin pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/devin-pricing (markdown: https://theagenticweekly.com/what-is/devin-pricing.md). Devin pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- What is Dify? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/dify (markdown: https://theagenticweekly.com/what-is/dify.md). What is Dify? Listed under Agent building, from LangGenius; license: Modified Apache 2.0; 159,534 GitHub stars. Facts from official pages read Oct 25, 2026.
- ElevenLabs pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/elevenlabs-pricing (markdown: https://theagenticweekly.com/what-is/elevenlabs-pricing.md). ElevenLabs pricing as of Oct 25, 2026 (Credit-based subscription): plan table, free tier and what the vendor's page does not say.
- What is Firecrawl? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/firecrawl (markdown: https://theagenticweekly.com/what-is/firecrawl.md). What is Firecrawl? Listed under MCP servers, from Firecrawl; license: AGPL-3.0; pricing: Freemium. Facts from official pages read Oct 25, 2026.
- GitHub Copilot pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/github-copilot-pricing (markdown: https://theagenticweekly.com/what-is/github-copilot-pricing.md). GitHub Copilot pricing as of Oct 25, 2026 (Free + paid plans with usage credits): plan table, free tier and what the vendor's page does not say.
- Grammarly pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/grammarly-pricing (markdown: https://theagenticweekly.com/what-is/grammarly-pricing.md). Grammarly pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- HeyGen pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/heygen-pricing (markdown: https://theagenticweekly.com/what-is/heygen-pricing.md). HeyGen pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- What is Hugging Face? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/hugging-face (markdown: https://theagenticweekly.com/what-is/hugging-face.md). What is Hugging Face? Listed under Local LLMs and model platforms, from Hugging Face; license: Varies by repository. Facts from official pages read Oct 25, 2026.
- What is Ideogram? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/ideogram (markdown: https://theagenticweekly.com/what-is/ideogram.md). What is Ideogram? Listed under Image generation, from Ideogram; license: Proprietary. Facts from official pages read Oct 25, 2026.
- Ideogram pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/ideogram-pricing (markdown: https://theagenticweekly.com/what-is/ideogram-pricing.md). Ideogram pricing as of Oct 25, 2026 (Freemium, subscription and pay-as-you-go): plan table, free tier and what the vendor's page does not say.
- Kiro pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/kiro-pricing (markdown: https://theagenticweekly.com/what-is/kiro-pricing.md). Kiro pricing as of Oct 25, 2026 (Freemium (credit-based)): plan table, free tier and what the vendor's page does not say.
- What is LangChain? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/langchain (markdown: https://theagenticweekly.com/what-is/langchain.md). What is LangChain? Listed under Agent building, from LangChain; license: MIT; 148,451 GitHub stars. Facts from official pages read Oct 25, 2026.
- Mistral Le Chat pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/le-chat-pricing (markdown: https://theagenticweekly.com/what-is/le-chat-pricing.md). Mistral Le Chat pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- Leonardo.Ai pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/leonardo-ai-pricing (markdown: https://theagenticweekly.com/what-is/leonardo-ai-pricing.md). Leonardo.Ai pricing as of Oct 25, 2026 (Freemium with subscriptions and pay-as-you-go API): plan table, free tier and what the vendor's page does not say.
- What is llama.cpp? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/llama-cpp (markdown: https://theagenticweekly.com/what-is/llama-cpp.md). What is llama.cpp? Listed under Local LLMs and model platforms, from ggml-org; license: MIT; 132,360 GitHub stars. Facts from official pages read Oct 25, 2026.
- What is LM Studio? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/lm-studio (markdown: https://theagenticweekly.com/what-is/lm-studio.md). What is LM Studio? Listed under Local LLMs and model platforms, from LM Studio; license: Proprietary application. Facts from official pages read Oct 25, 2026.
- What is Make? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/make (markdown: https://theagenticweekly.com/what-is/make.md). What is Make? Listed under No-code automation, from Make; license: Proprietary; pricing: Freemium, credits-based. Facts from official pages read Oct 25, 2026.
- What is an MCP server? A plain-English guide (2026): https://theagenticweekly.com/what-is/mcp-server (markdown: https://theagenticweekly.com/what-is/mcp-server.md). An MCP server meaning, explained from the official MCP docs: how hosts, clients and servers work, local vs remote, popular examples, and what we tested.
- Mem0 pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/mem0-pricing (markdown: https://theagenticweekly.com/what-is/mem0-pricing.md). Mem0 pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- Murf AI pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/murf-ai-pricing (markdown: https://theagenticweekly.com/what-is/murf-ai-pricing.md). Murf AI pricing as of Oct 25, 2026 (Free API trial, pay as you go and enterprise): plan table, free tier and what the vendor's page does not say.
- What is n8n? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/n8n (markdown: https://theagenticweekly.com/what-is/n8n.md). What is n8n? Listed under Agent building, from n8n; pricing: Free self-hosted Community Edition + paid plans. Facts from official pages read Oct 25, 2026.
- Notion AI pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/notion-ai-pricing (markdown: https://theagenticweekly.com/what-is/notion-ai-pricing.md). Notion AI pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- What is Ollama? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/ollama (markdown: https://theagenticweekly.com/what-is/ollama.md). What is Ollama? Listed under Local LLMs and model platforms, from Ollama; license: MIT; 184,762 GitHub stars. Facts from official pages read Oct 25, 2026.
- Ollama pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/ollama-pricing (markdown: https://theagenticweekly.com/what-is/ollama-pricing.md). Ollama pricing as of Oct 25, 2026 (Free + paid cloud plans): plan table, free tier and what the vendor's page does not say.
- What is Open WebUI? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/open-webui (markdown: https://theagenticweekly.com/what-is/open-webui.md). What is Open WebUI? Listed under Local LLMs and model platforms, from Open WebUI; license: Open WebUI License. Facts from official pages read Oct 25, 2026.
- What is OpenHands? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/openhands (markdown: https://theagenticweekly.com/what-is/openhands.md). What is OpenHands? Listed under AI coding agents, from OpenHands; license: MIT; 91,839 GitHub stars. Facts from official pages read Oct 25, 2026.
- What is Perplexity? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/perplexity (markdown: https://theagenticweekly.com/what-is/perplexity.md). What is Perplexity? Listed under AI search and research, from Perplexity AI; license: Proprietary; pricing: Freemium. Facts from official pages read Oct 25, 2026.
- Perplexity pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/perplexity-pricing (markdown: https://theagenticweekly.com/what-is/perplexity-pricing.md). Perplexity pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- Runway pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/runway-pricing (markdown: https://theagenticweekly.com/what-is/runway-pricing.md). Runway pricing as of Oct 25, 2026 (Credit-based subscription): plan table, free tier and what the vendor's page does not say.
- v0 pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/v0-pricing (markdown: https://theagenticweekly.com/what-is/v0-pricing.md). v0 pricing as of Oct 25, 2026 (Free + paid plans with token-based usage): plan table, free tier and what the vendor's page does not say.
- What is Vapi? Facts, license and pricing (Oct 2026): https://theagenticweekly.com/what-is/vapi (markdown: https://theagenticweekly.com/what-is/vapi.md). What is Vapi? Listed under AI voice and speech, from Vapi; license: Proprietary; pricing: Pay as you go. Facts from official pages read Oct 25, 2026.
- What is vibe coding? Meaning, how it works and the risks (2026): https://theagenticweekly.com/what-is/vibe-coding (markdown: https://theagenticweekly.com/what-is/vibe-coding.md). Vibe coding means building software by describing it to an AI and accepting what it writes. Definition, how it works, risks and tools, from sources read 2026-10-25.
- Warp pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/warp-pricing (markdown: https://theagenticweekly.com/what-is/warp-pricing.md). Warp pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- Zapier pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/zapier-pricing (markdown: https://theagenticweekly.com/what-is/zapier-pricing.md). Zapier pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.
- Zed pricing (Oct 2026): plans and free tier: https://theagenticweekly.com/what-is/zed-pricing (markdown: https://theagenticweekly.com/what-is/zed-pricing.md). Zed pricing as of Oct 25, 2026 (Freemium): plan table, free tier and what the vendor's page does not say.

# Guides (setup steps; each says which steps we ran and which come from vendor docs)

- Guides index: https://theagenticweekly.com/guides (markdown: https://theagenticweekly.com/guides.md)
- How to add an MCP server to Claude Code: https://theagenticweekly.com/guides/add-mcp-server-to-claude-code (markdown: https://theagenticweekly.com/guides/add-mcp-server-to-claude-code.md). Reported (not run by us): from the Claude Code docs. The claude mcp add commands for remote and local MCP servers, scopes and status checks, from Claude Code's docs. Reported; we have not run them. Last verified 2026-10-26.
- Best model for OpenCode, Cline and Claude Code: options by budget: https://theagenticweekly.com/guides/best-model-for-opencode-cline-claude-code (markdown: https://theagenticweekly.com/guides/best-model-for-opencode-cline-claude-code.md). Reported (not run by us): prices and setup from vendor docs; our tool tests do not measure model quality. Model options for OpenCode, Cline and Claude Code by budget, with list prices and setup steps from vendor docs. No quality ranking; not benchmarked by us. Last verified 2026-10-26.
- How to build an MCP server with FastMCP: https://theagenticweekly.com/guides/build-an-mcp-server-with-fastmcp (markdown: https://theagenticweekly.com/guides/build-an-mcp-server-with-fastmcp.md). Tested (smoke test, 2026-10-25, no LLM involved), stdio only. Install FastMCP and write a three-tool Python MCP server. Tested for the stdio run in our log; HTTP transport and client setup are from the FastMCP docs. Last verified 2026-10-26.
- ChatGPT Work vs ChatGPT Chat vs Codex: what each is for: https://theagenticweekly.com/guides/chatgpt-work-vs-chatgpt-chat (markdown: https://theagenticweekly.com/guides/chatgpt-work-vs-chatgpt-chat.md). Reported (not run by us): from OpenAI's documentation; we have not used ChatGPT Work. What ChatGPT Work is and how it differs from Chat and Codex, reported from OpenAI's docs. We have not used it. Last verified 2026-10-26.
- Is Claude Opus 5.5 free? Which plans include it: https://theagenticweekly.com/guides/claude-opus-5-5-free-or-paid (markdown: https://theagenticweekly.com/guides/claude-opus-5-5-free-or-paid.md). Reported (not run by us): from Anthropic's pricing page; we did not test access with an account. Is Claude Opus 5.5 free? No: not on the Free plan; included on Pro, Max, Team and Enterprise, or by API. Reported from Anthropic's pricing page. Last verified 2026-10-26.
- Figma MCP server: how to enable it: https://theagenticweekly.com/guides/figma-mcp-server (markdown: https://theagenticweekly.com/guides/figma-mcp-server.md). Reported (not run by us): from Figma's Help Center and developer docs. Who can use Figma's MCP server, remote versus desktop setup and the documented Claude Code commands. Reported from Figma's docs; we have not run it. Last verified 2026-10-26.
- Gemini 4 Argon API pricing: $2 and $10 per million tokens until Nov 30, then $4 and $20: https://theagenticweekly.com/guides/gemini-4-argon-api-pricing (markdown: https://theagenticweekly.com/guides/gemini-4-argon-api-pricing.md). Reported (not run by us): from Google's pricing page; we have not called the API. Gemini 4 Argon API price: $2/$10 per million tokens through Nov 30, 2026, then $4/$20. Reported from Google's pricing page; not run by us. Last verified 2026-10-26.
- Git MCP server: how to set it up and what it can do: https://theagenticweekly.com/guides/git-mcp-server (markdown: https://theagenticweekly.com/guides/git-mcp-server.md). Tested (smoke test, 2026-10-25, no LLM involved). Install the reference Git MCP server with uv or pip. Tested for the logged no-model smoke test (start, 12 tools, git_status); client setup is from the README. Last verified 2026-10-26.
- GitHub MCP server: how to set it up: https://theagenticweekly.com/guides/github-mcp-server (markdown: https://theagenticweekly.com/guides/github-mcp-server.md). Reported (not run by us): from the README and Claude Code docs. Set up GitHub's official MCP server, remote or local, with a token or OAuth. Reported from the README and Claude Code docs; we have not run it. Last verified 2026-10-26.
- Which coding agents can use GPT-6.1 Terra?: https://theagenticweekly.com/guides/gpt-6-1-terra-in-coding-agents (markdown: https://theagenticweekly.com/guides/gpt-6-1-terra-in-coding-agents.md). Reported (not run by us): from vendor docs and changelogs. Where GPT-6.1 Terra works today: Copilot, Codex and ChatGPT Work confirmed; Cursor, Claude Code, Cline and OpenCode not confirmed. Reported from vendor docs. Last verified 2026-10-26.
- Is Kimi K3.1 Pro open-weight? Status of the Nov 4 promise: https://theagenticweekly.com/guides/kimi-k3-1-pro-open-weights (markdown: https://theagenticweekly.com/guides/kimi-k3-1-pro-open-weights.md). Reported (not run by us): from Moonshot's blog and a Hugging Face listing check. Status of Kimi K3.1 Pro open weights: promised by Nov 4, 2026, not on Hugging Face as of Oct 26. Reported; not run by us. Last verified 2026-10-26.
- Playwright MCP server: setup, tools and what we measured: https://theagenticweekly.com/guides/playwright-mcp-server (markdown: https://theagenticweekly.com/guides/playwright-mcp-server.md). Tested (smoke test, 2026-10-25, no LLM involved). Install and run Microsoft's Playwright MCP server. Tested for the logged no-model smoke test (start, 25 tools, navigate); client setup is from the README. Last verified 2026-10-26.
- How to run Claude Code with a local model: https://theagenticweekly.com/guides/run-claude-code-with-a-local-model (markdown: https://theagenticweekly.com/guides/run-claude-code-with-a-local-model.md). Tested (partial, 2026-10-25): one headless edit, small local model, not Claude models. Point Claude Code at a local Ollama model. Tested in part: a 3B model made one edit on an exact prompt and failed on a vague one; the settings come from Ollama's docs. Last verified 2026-10-26.
- Which Claude model should you use? Haiku 5.5, Sonnet 5.5, Opus 5.5 or Fable 5.1: https://theagenticweekly.com/guides/which-claude-model-should-i-use (markdown: https://theagenticweekly.com/guides/which-claude-model-should-i-use.md). Reported (not run by us): Anthropic's docs plus our editorial judgement. Pick between Claude Haiku 5.5, Sonnet 5.5, Opus 5.5 and Fable 5.1 by task and budget, from Anthropic's docs and prices. Reported; not run by us. Last verified 2026-10-26.

# Original tests and research (run by us on a CPU-only box)

- Install size and a first tool-calling task across 16 agent frameworks: what we measured: https://theagenticweekly.com/research/agent-framework-install-and-first-task (markdown: https://theagenticweekly.com/research/agent-framework-install-and-first-task.md). We installed 16 Python and 4 JavaScript agent frameworks and ran one tool task on each with a local 3B model. Installs ranged from 26 to 819 MB.
- MCP servers over stdio: startup time, call latency and what broke when we ran 15 of them: https://theagenticweekly.com/research/mcp-servers-latency-and-failures (markdown: https://theagenticweekly.com/research/mcp-servers-latency-and-failures.md). We timed 15 MCP server setups over stdio. Most calls took about a millisecond; the surprises were a version change, a loopback block and a hosted lookup.
- npm on an older Node installs old versions without telling you: four measured cases and other install traps: https://theagenticweekly.com/research/older-node-silent-downgrades (markdown: https://theagenticweekly.com/research/older-node-silent-downgrades.md). On Node 20, npm quietly installed older major versions of four AI packages; we list the versions, the traps in Python installs, and the one test setting that fooled us.
- Ollama and llama.cpp on a CPU-only box: three Qwen2.5 sizes, same prompts: https://theagenticweekly.com/research/ollama-vs-llama-cpp-qwen-sizes (markdown: https://theagenticweekly.com/research/ollama-vs-llama-cpp-qwen-sizes.md). We ran Qwen2.5-Instruct at 0.5B, 1.5B and 3B (Q4_K_M) through Ollama and llama.cpp on 8 CPU cores and compared pass rates, speed and tool calling.
- A voice round trip on CPU: Piper speaks, faster-whisper listens, and int8 changes the answer: https://theagenticweekly.com/research/voice-round-trip-piper-faster-whisper (markdown: https://theagenticweekly.com/research/voice-round-trip-piper-faster-whisper.md). We synthesised 10 sentences with Piper, transcribed them back with faster-whisper in 9 configurations and measured word error rate. float32 was stable; int8 was not.
- Agno 3.1.10: our test log: https://theagenticweekly.com/tests/agno (markdown: https://theagenticweekly.com/tests/agno.md). Agno 3.1.10 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.8 s. Run 2026-10-25 on a CPU-only box.
- Aider 0.86.2: our test log: https://theagenticweekly.com/tests/aider (markdown: https://theagenticweekly.com/tests/aider.md). Aider 0.86.2 implemented a missing function with a local 3B model in 3 of 3 trials (27 to 32 s each).
- AutoGen 0.7.5: our test log: https://theagenticweekly.com/tests/autogen (markdown: https://theagenticweekly.com/tests/autogen.md). AutoGen 0.7.5 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.9 s. Run 2026-10-25 on a CPU-only box.
- Chrome DevTools MCP chrome_devtools 1.12.0: our test log: https://theagenticweekly.com/tests/chrome-devtools-mcp (markdown: https://theagenticweekly.com/tests/chrome-devtools-mcp.md). Chrome DevTools MCP 1.12.0 exposed 32 tools and opened a page in headless Chrome in 136.6 ms median.
- Claude Code 2.1.310: our test log: https://theagenticweekly.com/tests/claude-code (markdown: https://theagenticweekly.com/tests/claude-code.md). Claude Code 2.1.310 edited calc.py through its Edit tool with a local 3B model on the second prompt (85.9 s); the first, vaguer prompt made no edit.
- Cline CLI 3.0.73: our test log: https://theagenticweekly.com/tests/cline-cli (markdown: https://theagenticweekly.com/tests/cline-cli.md). Cline CLI 3.0.73 edited calc.py correctly with a local 3B model, then failed on every follow-up command with 'Invalid input'.
- Codex CLI 0.164.0: our test log: https://theagenticweekly.com/tests/codex-cli (markdown: https://theagenticweekly.com/tests/codex-cli.md). Codex CLI 0.164.0 changed calc.py by running a sed command in its sandbox with a local 3B model; an earlier plain-language edit request ran no edit.
- ComfyUI ComfyUI 0.42.0 (commit cdf272f), PyTorch 2.14.1 CPU: our test log: https://theagenticweekly.com/tests/comfyui (markdown: https://theagenticweekly.com/tests/comfyui.md). ComfyUI 0.42.0 installed in 17.42 s and its CPU-only server was ready in 30.1 s with 1014 node types, but we did not generate an image.
- Context7 MCP Context7 4.4.4: our test log: https://theagenticweekly.com/tests/context7-mcp (markdown: https://theagenticweekly.com/tests/context7-mcp.md). Context7 MCP 4.4.4 started locally in 462 ms but its tool calls go to a hosted API.
- Continue CLI 1.5.47: our test log: https://theagenticweekly.com/tests/continue-cli (markdown: https://theagenticweekly.com/tests/continue-cli.md). Continue CLI 1.5.47 answered a question and fixed a one-line bug with a local 3B model through a YAML config; a 1.6 s, 81 MB install.
- CrewAI 1.15.30: our test log: https://theagenticweekly.com/tests/crewai (markdown: https://theagenticweekly.com/tests/crewai.md). CrewAI 1.15.30 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 8.4 s. Run 2026-10-25 on a CPU-only box.
- DSPy 3.4.1: our test log: https://theagenticweekly.com/tests/dspy (markdown: https://theagenticweekly.com/tests/dspy.md). DSPy 3.4.1 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 3.7 s. Run 2026-10-25 on a CPU-only box.
- faster-whisper (speech to text) 1.2.1 (CTranslate2 4.8.4): our test log: https://theagenticweekly.com/tests/faster-whisper (markdown: https://theagenticweekly.com/tests/faster-whisper.md). faster-whisper 1.2.1 transcribed 41 s of speech at 2.4 percent word error with the base.en model in float32 (deterministic).
- Flowise npm latest on Node 24.22.1: our test log: https://theagenticweekly.com/tests/flowise (markdown: https://theagenticweekly.com/tests/flowise.md). Flowise failed to install from npm on Node 24.22.1 after 106.2 s: the better-sqlite3 native build needed a C++ toolchain this machine lacks.
- GitHub MCP Server 2.0.3: our test log: https://theagenticweekly.com/tests/github-mcp-server (markdown: https://theagenticweekly.com/tests/github-mcp-server.md). GitHub MCP Server 2.0.3 started over stdio in a median 266 ms and listed 46 tools in a median 15 ms; we did not make any authenticated call.
- Google ADK (Python) 2.13.0: our test log: https://theagenticweekly.com/tests/google-adk (markdown: https://theagenticweekly.com/tests/google-adk.md). Google ADK (Python) 2.13.0 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.7 s. Run 2026-10-25 on a CPU-only box.
- goose 1.56.0: our test log: https://theagenticweekly.com/tests/goose (markdown: https://theagenticweekly.com/tests/goose.md). goose 1.56.0 edited calc.py on the second headless run with a local 3B model (19.4 s); the first run called the edit tool but its text did not match.
- GPT4All gpt4all 2.8.2 (Python SDK): our test log: https://theagenticweekly.com/tests/gpt4all (markdown: https://theagenticweekly.com/tests/gpt4all.md). GPT4All's Python SDK 2.8.2 loaded a local 3B GGUF in 4.01 s and answered 17 + 25 with 42 in 3 of 3 trials, but kept writing past the answer.
- Haystack 3.4.1: our test log: https://theagenticweekly.com/tests/haystack (markdown: https://theagenticweekly.com/tests/haystack.md). Haystack 3.4.1 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 2.3 s. Run 2026-10-25 on a CPU-only box.
- Hugging Face huggingface_hub 2.3.0, transformers 5.20.2, torch 2.14.1 (CPU): our test log: https://theagenticweekly.com/tests/hugging-face (markdown: https://theagenticweekly.com/tests/hugging-face.md). Hugging Face's Hub searched without a login in 0.1 s, a 272 MB model downloaded in 3.61 s, and transformers ran it on CPU.
- Instruction-following test: 15 AI models, 32 checkable tasks: https://theagenticweekly.com/tests/instruction-following (markdown: https://theagenticweekly.com/tests/instruction-following.md). 32 prompts with 52 constraints checked by code (word counts, bullet counts, JSON, forbidden words), 3 runs each, 15 models. Pass rates with intervals, speed, 
- Instructor 1.17.0: our test log: https://theagenticweekly.com/tests/instructor (markdown: https://theagenticweekly.com/tests/instructor.md). Instructor 1.17.0 returned a validated Pydantic object from a local 3B model in 3 of 3 trials (structured extraction, not a tool call).
- LangChain 1.4.7: our test log: https://theagenticweekly.com/tests/langchain (markdown: https://theagenticweekly.com/tests/langchain.md). LangChain 1.4.7 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 0.8 s. Run 2026-10-25 on a CPU-only box.
- LangChain.js 1.5.19: our test log: https://theagenticweekly.com/tests/langchainjs (markdown: https://theagenticweekly.com/tests/langchainjs.md). LangChain.js 1.5.19 passed our one-tool task 3 of 3 on Node 20 and 3 of 3 on Node 24; about 101 MB in 36 packages.
- Langflow langflow 1.13.0: our test log: https://theagenticweekly.com/tests/langflow (markdown: https://theagenticweekly.com/tests/langflow.md). Langflow 1.13.0 installed in 19.76 s (1.7 GB) and its backend was healthy 91.5 s after launch, serving 26 starter flows and 416 components.
- LangGraph 1.2.17: our test log: https://theagenticweekly.com/tests/langgraph (markdown: https://theagenticweekly.com/tests/langgraph.md). LangGraph 1.2.17 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 2.5 s. Run 2026-10-25 on a CPU-only box.
- LiteLLM 1.106.4: our test log: https://theagenticweekly.com/tests/litellm (markdown: https://theagenticweekly.com/tests/litellm.md). LiteLLM 1.106.4 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 2.3 s. Run 2026-10-25 on a CPU-only box.
- llama.cpp b11889 (commit d854d98ba): our test log: https://theagenticweekly.com/tests/llama-cpp (markdown: https://theagenticweekly.com/tests/llama-cpp.md). llama.cpp b11889 generated 118 tokens/s (0.5B), 40 (1.5B) and 19.5 (3B) on 8 CPU threads with llama-bench.
- LlamaIndex 0.14.25: our test log: https://theagenticweekly.com/tests/llamaindex (markdown: https://theagenticweekly.com/tests/llamaindex.md). LlamaIndex 0.14.25 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 3.4 s. Run 2026-10-25 on a CPU-only box.
- Mastra 0.24.9 (Node 20) / 1.79.0 (Node 24): our test log: https://theagenticweekly.com/tests/mastra (markdown: https://theagenticweekly.com/tests/mastra.md). Mastra passed 5 of 5 on Node 20 (0.24.9) and 5 of 5 on Node 24 (1.79.0) with temperature pinned to 0.
- MCP Inspector (CLI mode) 0.15.0 (Node 20) / 2.12.0 (Node 24): our test log: https://theagenticweekly.com/tests/mcp-inspector (markdown: https://theagenticweekly.com/tests/mcp-inspector.md). MCP Inspector listed the filesystem server's 14 tools from the command line on both 0.15.0 (Node 20) and 2.12.0 (Node 24) in under a second.
- FastMCP 4.1.3 and the official MCP Python SDK 2.3.0 (server side) fastmcp 4.1.3; mcp 2.3.0: our test log: https://theagenticweekly.com/tests/mcp-python-servers (markdown: https://theagenticweekly.com/tests/mcp-python-servers.md). The same three-tool server written with FastMCP 4.1.3 and with the official MCP Python SDK 2.3.0 answered calls equally fast (about 1 ms).
- MCP Everything (reference) server 1.0.0 (Node 20) / 2.0.1 (Node 24): our test log: https://theagenticweekly.com/tests/mcp-server-everything (markdown: https://theagenticweekly.com/tests/mcp-server-everything.md). MCP Everything (reference) server answered over stdio in 356 ms cold (v1.0.0, Node 20) and 252 ms (v2.0.1, Node 24); median call 0.5 to 0.4 ms.
- MCP Fetch server mcp-fetch 2026.10.24: our test log: https://theagenticweekly.com/tests/mcp-server-fetch (markdown: https://theagenticweekly.com/tests/mcp-server-fetch.md). The MCP Fetch server 2026.10.24 refused to fetch a local test page by default (a deliberate loopback block).
- MCP Filesystem server 1.0.0 (Node 20) / 2.0.1 (Node 24): our test log: https://theagenticweekly.com/tests/mcp-server-filesystem (markdown: https://theagenticweekly.com/tests/mcp-server-filesystem.md). MCP Filesystem server answered over stdio in 279 ms cold (v1.0.0, Node 20) and 187 ms (v2.0.1, Node 24); median call 0.8 to 1.0 ms.
- MCP Git server mcp-git 2026.10.19: our test log: https://theagenticweekly.com/tests/mcp-server-git (markdown: https://theagenticweekly.com/tests/mcp-server-git.md). The MCP Git server 2026.10.19 started in 512 ms, listed 12 tools and answered git_status in 3.2 ms median. Run 2026-10-25 on a CPU-only box.
- MCP Memory server 1.0.0 (Node 20) / 2.0.0 (Node 24): our test log: https://theagenticweekly.com/tests/mcp-server-memory (markdown: https://theagenticweekly.com/tests/mcp-server-memory.md). MCP Memory server answered over stdio in 284 ms cold (v1.0.0, Node 20) and 147 ms (v2.0.0, Node 24); median call 0.9 to 0.6 ms.
- MCP Sequential Thinking server 1.0.0 (Node 20) / 2.0.0 (Node 24): our test log: https://theagenticweekly.com/tests/mcp-server-sequential-thinking (markdown: https://theagenticweekly.com/tests/mcp-server-sequential-thinking.md). MCP Sequential Thinking server answered over stdio in 271 ms cold (v1.0.0, Node 20) and 159 ms (v2.0.0, Node 24); median call 0.8 to 0.4 ms.
- MCP Time server mcp-time 2026.10.19: our test log: https://theagenticweekly.com/tests/mcp-server-time (markdown: https://theagenticweekly.com/tests/mcp-server-time.md). The MCP Time server 2026.10.19 started in 483 ms and answered get_current_time in 0.8 ms median. Run 2026-10-25 on a CPU-only box.
- Mem0 mem0ai 2.2.1 (open-source Python SDK): our test log: https://theagenticweekly.com/tests/mem0 (markdown: https://theagenticweekly.com/tests/mem0.md). Mem0's open-source SDK 2.2.1 stored a fact from one chat message with a local 3B model and retrieved it with a semantic search in 2.04 s.
- n8n 2.44.6: our test log: https://theagenticweekly.com/tests/n8n (markdown: https://theagenticweekly.com/tests/n8n.md). n8n 2.44.6 installed from npm on Node 24 in 1 min 54 s (2,343 packages), imported a three-node workflow from the CLI and executed it.
- Notion MCP server 2.5.2 (@notionhq/notion-mcp-server): our test log: https://theagenticweekly.com/tests/notion-mcp (markdown: https://theagenticweekly.com/tests/notion-mcp.md). Notion MCP server 2.5.2 (@notionhq/notion-mcp-server) started over stdio in a median 316 ms and listed 24 tools in a median 16 ms.
- Ollama 0.40.11: our test log: https://theagenticweekly.com/tests/ollama (markdown: https://theagenticweekly.com/tests/ollama.md). Ollama 0.40.11 served three Qwen2.5 sizes over its OpenAI-compatible API on 8 CPU cores: all three answered simple prompts.
- Open WebUI 0.12.4: our test log: https://theagenticweekly.com/tests/open-webui (markdown: https://theagenticweekly.com/tests/open-webui.md). Open WebUI 0.12.4 installed with uv in 11.6 s (6,977 MB in the virtualenv), was healthy about 11 to 12 s after start.
- OpenAI Agents SDK for JavaScript 0.21.0: our test log: https://theagenticweekly.com/tests/openai-agents-js (markdown: https://theagenticweekly.com/tests/openai-agents-js.md). OpenAI Agents SDK for JS (0.21.0) passed 2 of 5 (Node 20) and 1 of 5 (Node 24) at default sampling but 5 of 5 on both once temperature was pinned to 0.
- OpenAI Agents SDK (Python) 0.24.0: our test log: https://theagenticweekly.com/tests/openai-agents-sdk (markdown: https://theagenticweekly.com/tests/openai-agents-sdk.md). OpenAI Agents SDK (Python) 0.24.0 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 1.7 s.
- OpenAI Python SDK 3.37.0: our test log: https://theagenticweekly.com/tests/openai-python (markdown: https://theagenticweekly.com/tests/openai-python.md). The plain OpenAI Python SDK (v3.37.0) with a hand-written tool loop passed 3 of 3 trials against a local model.
- opencode 1.19.4: our test log: https://theagenticweekly.com/tests/opencode (markdown: https://theagenticweekly.com/tests/opencode.md). opencode 1.19.4 edited calc.py on the second headless run with a local 3B model; the first run tried the wrong path (src/calc.py) and failed.
- OpenHands CLI 1.16.0 (SDK 1.21.0): our test log: https://theagenticweekly.com/tests/openhands-cli (markdown: https://theagenticweekly.com/tests/openhands-cli.md). OpenHands 1.16.0 installed (14 s, 537 MB) and started its agent, but with a local 3B model it made 0 edits in 2 trials: calc.py was unchanged.
- Piper (text to speech) 1.8.0 (piper-tts): our test log: https://theagenticweekly.com/tests/piper (markdown: https://theagenticweekly.com/tests/piper.md). Piper 1.8.0 synthesised 10 sentences (about 41 s of 22.05 kHz audio) on CPU in 1.4 to 1.5 s each including process start; 3.5 s uv install, 167 MB.
- Playwright MCP Playwright 1.64.0-alpha-1790635538000: our test log: https://theagenticweekly.com/tests/playwright-mcp (markdown: https://theagenticweekly.com/tests/playwright-mcp.md). Playwright MCP 0.0.83 exposed 25 tools and navigated a headless Chrome to a local page in 40.5 ms median (cold start 383 ms).
- Pydantic AI 2.61.0: our test log: https://theagenticweekly.com/tests/pydantic-ai (markdown: https://theagenticweekly.com/tests/pydantic-ai.md). Pydantic AI 2.61.0 passed our one-tool task on a local 3B model: 3 of 3 trials; cold uv install 3.7 s. Run 2026-10-25 on a CPU-only box.
- Semantic Kernel 1.45.0: our test log: https://theagenticweekly.com/tests/semantic-kernel (markdown: https://theagenticweekly.com/tests/semantic-kernel.md). Semantic Kernel 1.45.0 returned an empty reply with our standard prompt (0 of 3) but worked once the prompt used its plugin-prefixed tool name (3 of 3).
- Sentry MCP server 0.42.0 (@sentry/mcp-server): our test log: https://theagenticweekly.com/tests/sentry-mcp (markdown: https://theagenticweekly.com/tests/sentry-mcp.md). Sentry MCP server 0.42.0 (@sentry/mcp-server) started over stdio in a median 1134 ms and listed 9 tools in a median 13 ms.
- smolagents 1.26.0: our test log: https://theagenticweekly.com/tests/smolagents (markdown: https://theagenticweekly.com/tests/smolagents.md). smolagents 1.26.0 passed our one-tool task on a local 3B model: 6 of 6 trials; cold uv install 1.5 s. Run 2026-10-25 on a CPU-only box.
- Supabase MCP server 0.14.0 (@supabase/mcp-server-supabase): our test log: https://theagenticweekly.com/tests/supabase-mcp (markdown: https://theagenticweekly.com/tests/supabase-mcp.md). Supabase MCP server 0.14.0 (@supabase/mcp-server-supabase) started over stdio in a median 468 ms and listed 19 tools in a median 363 ms.
- Vercel AI SDK 6.0.315 (Node 20) / 7.0.153 (Node 24): our test log: https://theagenticweekly.com/tests/vercel-ai-sdk (markdown: https://theagenticweekly.com/tests/vercel-ai-sdk.md). Vercel AI SDK passed our one-tool task 3 of 3 on Node 20 (ai 6.0.315) and 3 of 3 on Node 24 (ai 7.0.153); the install is 26 to 27 MB.
- whisper.cpp not run: our test log: https://theagenticweekly.com/tests/whisper-cpp (markdown: https://theagenticweekly.com/tests/whisper-cpp.md). We could not run whisper.cpp: no prebuilt Linux release was available for download and this machine has no make, g++ or cmake to build it.
- Writing test: 15 AI models, same prompts, rated blind: https://theagenticweekly.com/tests/writing (markdown: https://theagenticweekly.com/tests/writing.md). We ran 14 writing prompts through 15 open models 3 times each and rated them blind. Prompts, outputs, rubric, rater prompts and scores are published.
- Writing leaderboard: fifteen small AI models, scores by criterion: https://theagenticweekly.com/tests/writing-leaderboard (markdown: https://theagenticweekly.com/tests/writing-leaderboard.md). Blind win rates, rubric scores for clarity, accuracy, concision, tone and originality, and a head-to-head grid for 15 open models. Raw outputs published.
- Writing prompt w01: one sentence on what an API is, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w01 (markdown: https://theagenticweekly.com/tests/writing-prompt-w01.md). one sentence on what an API is (one-sentence): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writing t
- Writing prompt w02: one sentence: coffee shop open Sundays, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w02 (markdown: https://theagenticweekly.com/tests/writing-prompt-w02.md). one sentence: coffee shop open Sundays (one-sentence): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our w
- Writing prompt w03: one sentence apologising for a delay, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w03 (markdown: https://theagenticweekly.com/tests/writing-prompt-w03.md). one sentence apologising for a delay (one-sentence): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our wri
- Writing prompt w04: one-sentence tagline for an offline notes app, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w04 (markdown: https://theagenticweekly.com/tests/writing-prompt-w04.md). one-sentence tagline for an offline notes app (one-sentence): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part o
- Writing prompt w05: two sentences on why the sky is blue, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w05 (markdown: https://theagenticweekly.com/tests/writing-prompt-w05.md). two sentences on why the sky is blue (two-sentence): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our wri
- Writing prompt w06: two sentences persuading a manager, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w06 (markdown: https://theagenticweekly.com/tests/writing-prompt-w06.md). two sentences persuading a manager (two-sentence): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writi
- Writing prompt w07: two sentences on a rainy commute, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w07 (markdown: https://theagenticweekly.com/tests/writing-prompt-w07.md). two sentences on a rainy commute (two-sentence): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writing
- Writing prompt w08: water bottle product blurb, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w08 (markdown: https://theagenticweekly.com/tests/writing-prompt-w08.md). water bottle product blurb (product blurb): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writing test
- Writing prompt w09: self-watering pot product blurb, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w09 (markdown: https://theagenticweekly.com/tests/writing-prompt-w09.md). self-watering pot product blurb (product blurb): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writing
- Writing prompt w10: reply to a late-order complaint, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w10 (markdown: https://theagenticweekly.com/tests/writing-prompt-w10.md). reply to a late-order complaint (email reply): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writing t
- Writing prompt w11: reply to a meeting-change email, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w11 (markdown: https://theagenticweekly.com/tests/writing-prompt-w11.md). reply to a meeting-change email (email reply): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writing t
- Writing prompt w12: two-sentence summary of a library notice, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w12 (markdown: https://theagenticweekly.com/tests/writing-prompt-w12.md). two-sentence summary of a library notice (summary): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our writ
- Writing prompt w13: password manager for beginners, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w13 (markdown: https://theagenticweekly.com/tests/writing-prompt-w13.md). password manager for beginners (beginner explanation): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of our w
- Writing prompt w14: compound interest for a 12-year-old, answered by fifteen AI models: https://theagenticweekly.com/tests/writing-prompt-w14 (markdown: https://theagenticweekly.com/tests/writing-prompt-w14.md). compound interest for a 12-year-old (beginner explanation): all 45 outputs from 15 open AI models, with rubric scores and a code-checked format rule. Part of 

# Other pages (utility, hubs, newsletter, news and release pages)

- tools: https://theagenticweekly.com/tools
- alternatives: https://theagenticweekly.com/alternatives
- news / models: https://theagenticweekly.com/news/models
- news / tools: https://theagenticweekly.com/news/tools
- what-is: https://theagenticweekly.com/what-is
- archive / week / 2026-10-19: https://theagenticweekly.com/archive/week/2026-10-19
- archive / week / 2026-10-12: https://theagenticweekly.com/archive/week/2026-10-12
- archive / month / 2026-10: https://theagenticweekly.com/archive/month/2026-10
- best / llm-for-coding: https://theagenticweekly.com/best/llm-for-coding
- models / this-week: https://theagenticweekly.com/models/this-week
- models / releases: https://theagenticweekly.com/models/releases
- models / releases / 2026-10: https://theagenticweekly.com/models/releases/2026-10
- models / releases / 2026-09: https://theagenticweekly.com/models/releases/2026-09
- models / releases / 2026-06: https://theagenticweekly.com/models/releases/2026-06
- models / releases / 2026-05: https://theagenticweekly.com/models/releases/2026-05
- newsletter: https://theagenticweekly.com/newsletter
- newsletter / issue-1: https://theagenticweekly.com/newsletter/issue-1
- newsletter / archive: https://theagenticweekly.com/newsletter/archive
- cheat-sheet: https://theagenticweekly.com/cheat-sheet
- editorial-policy: https://theagenticweekly.com/editorial-policy
- sponsor: https://theagenticweekly.com/sponsor
- badge: https://theagenticweekly.com/badge
- submit: https://theagenticweekly.com/submit
- privacy: https://theagenticweekly.com/privacy
- terms: https://theagenticweekly.com/terms