AI release news: fast, sourced, timestamped
A news brief is a short write-up of a release, published as soon as we can check it against the vendor’s own source. Each brief shows when our watcher first saw the item and when the source published it, so you can judge the speed yourself. Briefs are labelled Reported (built from the linked sources, not run by us) unless the page says Tested and states what was run. Nothing here is a benchmark. Corrections policy.
How to follow: RSS /news/feed.xml · JSON Feed /feed.json · email The Weekly Verdict · MCP https://theagenticweekly.com/mcp (read-only, no key).
Feeds: news RSS, Atom, JSON Feed. For software and agents: /latest.json (latest 50 briefs with timestamps), news sitemap. Weekly verdicts and setup guides live on This week.
By type: AI model release news, updated Oct 26, 2026 · AI tool release news, updated Oct 26, 2026 · New AI models this week.
Featured: Anthropic IPO: key numbers, risks, timeline and FAQ from the S-1 (sourced, not investment advice).
Latest briefs (11)
- OpenAI GPT-6.1 Terra: a $1/$5 computer-use model, and what the launch post leaves out · Reported · first seen
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. - Superset is now Paradev: parallel coding-agent app ships desktop v2.0.0 and an $11.5M seed · Reported · first seen
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. - Anthropic files S-1 for an IPO on Nasdaq under "ANT", reporting $16.3 billion revenue in six months · Reported · first seen
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. - ARTEX: an open-source multi-agent LLM system for autonomous penetration testing in isolated labs · Reported · first seen
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. - PaperPilot: an open-source Rust PDF toolkit that runs as an MCP server, CLI and in-browser WASM · Reported · first seen
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. - Octopus SI OS: an open-source, self-evolving agentic OS with a Rust kernel, released as a research preview · Reported · first seen
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. - SolidCraft: open-source parametric CAD in pure Rust, with an MCP server so AI agents can model parts · Reported · first seen
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. - MCP reference servers v2.1.0 fixes eight vulnerabilities in filesystem, everything and fetch · Reported · first seen
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. - Microsoft open-sources Agensh, a multi-agent harness its authors scaled to 1,024 agents · Reported · first seen
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. - Ollama v0.40.11 adds speech generation on macOS and the d3 decision models · Reported · first seen
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. - Open WebUI v0.12.4 adds on-demand skill loading and says it contains unlisted security fixes · Reported · first seen
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.
Want these briefs in your Google Top Stories? Add Agentic Weekly as a preferred source on Google.
Frequently asked questions
How do I follow AI release news from Agentic Weekly?
Four ways: RSS at https://theagenticweekly.com/news/feed.xml, JSON Feed at https://theagenticweekly.com/feed.json, the free weekly email The Weekly Verdict at https://theagenticweekly.com/newsletter, or the read-only MCP server at https://theagenticweekly.com/mcp (no key, no sign-in). Details are on https://theagenticweekly.com/developers.
Are the briefs tested by Agentic Weekly?
Only if a brief says Tested. Every item is labelled Reported (taken from the vendor's own source, linked) or Tested (a small smoke test whose scope is stated).
How often is it updated?
Briefs are published as soon as we can check them against the vendor's source, and each shows when our watcher first saw the item. The weekly email goes out once a week.
Free · weekly · about six minutes
Get The Weekly Verdict, free
What shipped this week, what is worth switching to, and what to ignore. We run what we can and label what we only read.
- One email a week. Each release gets one call: update now, worth a look, or skip.
- Every claim shows its evidence. Reported means we read the sources; Tested means we ran it, with the exact scope and the failures left in.
- A free AI Tools Cheat Sheet (web page and printable PDF) lands in your inbox when you confirm.
Free. Unsubscribe anytime. No spam. We email a confirmation link first and add you only after you click it. We use your email only to send this newsletter and never sell it. Privacy