Ollama: what’s new in v0.40.11 and our verdict

Our 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.

Tested. Tested 2026-10-25 (America/Chicago). 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. Test environment: Linux x86_64 virtual machine (Debian 13), 8 vCPU, 15 GiB RAM, no GPU; no vendor account, login or API key. Agent tools were pointed at a local open-weight model (Qwen2.5-3B-Instruct (Q4_K_M) through Ollama 0.40.11), so we tested the tool, not any vendor's hosted models. 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. These are smoke tests, not a benchmark. Full log: /tests/ollama.

Full test log with exact commands, versions and raw output: Instruction-following test: 15 AI models, 32 checkable tasks, Ollama 0.40.11: our test log, Writing test: 15 AI models, same prompts, rated blind, Writing leaderboard: fifteen small AI models, scores by criterion, Writing prompt w01: one sentence on what an API is, answered by fifteen AI models, Writing prompt w02: one sentence: coffee shop open Sundays, answered by fifteen AI models, Writing prompt w03: one sentence apologising for a delay, answered by fifteen AI models, Writing prompt w04: one-sentence tagline for an offline notes app, answered by fifteen AI models, Writing prompt w05: two sentences on why the sky is blue, answered by fifteen AI models, Writing prompt w06: two sentences persuading a manager, answered by fifteen AI models, Writing prompt w07: two sentences on a rainy commute, answered by fifteen AI models, Writing prompt w08: water bottle product blurb, answered by fifteen AI models, Writing prompt w09: self-watering pot product blurb, answered by fifteen AI models, Writing prompt w10: reply to a late-order complaint, answered by fifteen AI models, Writing prompt w11: reply to a meeting-change email, answered by fifteen AI models, Writing prompt w12: two-sentence summary of a library notice, answered by fifteen AI models, Writing prompt w13: password manager for beginners, answered by fifteen AI models, Writing prompt w14: compound interest for a 12-year-old, answered by fifteen AI models.

Evidence

Tested (smoke test, 2026-10-25, local 3B model (Qwen2.5-3B)). A smoke test is not a benchmark.

What we 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.

Model used

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

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.

Date

Oct 25, 2026 (America/Chicago)

Best for

• Best AI tool for local LLM tools and model platforms: Run open models locally or in the cloud with a CLI and API.

Key facts

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.

• License: MIT (per GitHub)
• Platforms: macOS, Windows and Linux
• Pricing model: Free + paid cloud plans
• Free tier: Free: $0, pay-as-you-go usage credits for cloud models
• Paid plans: 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
• GitHub: ollama/ollama: 184,762 stars on 2026-10-25, last push 2026-10-25
• Latest release we saw: v0.40.11, 2026-10-25
• How to start: Download Ollama from ollama.com and run a model from the command line.

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

How to set it up: Get started with Ollama

These steps come from the vendor's own pages as read on 2026-10-25. We ran Ollama on 2026-10-25; the test log lists what was and was not covered.

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.

Pros and cons

Alternatives and comparisons

• LM Studio: Desktop app for running open models locally; the homepage now leads with Bionic, an agent for work and code.
• Jan: Open-source alternative to ChatGPT that runs open models locally or connects to cloud models like GPT and Claude.
• llama.cpp: LLM inference in C/C++: the open-source engine many local-model tools build on; MIT-licensed with near-daily build tags.
• Cohere: Enterprise AI company: Command models, Rerank and the North platform.

See all alternatives to Ollama

• Ollama vs llama.cpp: which should you use?
• Ollama vs Open WebUI: which should you use?

FAQ

What we have not verified

• Speed of a given model on your hardware

Sources

• Ollama (fetched 2026-10-25)
• Ollama pricing (fetched 2026-10-25)
• Ollama repository (GitHub) (fetched 2026-10-25)

Last verified Oct 25, 2026 · Agentic Weekly

Guides for Ollama

• How to run Claude Code with a local model (Tested (partial, 2026-10-25): one headless edit, small local model, not Claude models; last verified Oct 26, 2026)

Latest news about Ollama

• Ollama v0.40.11 adds speech generation on macOS and the d3 decision models (Reported; first seen 2026-10-26 03:48 UTC)

Related tools

• Cohere: Enterprise AI company: Command models, Rerank and the North platform.
• 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.
• Hugging Face: The hub where the ML community shares models, datasets and apps, with PRO, Team and Enterprise tiers.