llm-mistral 0.16 adds reasoning model support, including newly released Mistral Large 4
Simon Willison released llm-mistral 0.16 on October 6, 2026. The LLM plugin for the Mistral API now supports reasoning models, with Mistral Large 4 named as the example. Model specifications and pricing were not part of the release note.
llm-mistral 0.16, released October 6, 2026, adds support for reasoning models to the plugin that connects the LLM command-line tool to the Mistral API. The release note cites the newly released Mistral Large 4 as an example of a model the update is meant to handle.
What changed
The full release note is one line: "Adds support for reasoning models, such as the newly released Mistral Large 4." It was posted by Simon Willison, the maintainer of LLM, on October 6, 2026 at 9:32 pm.
llm-mistral is a plugin for LLM, Willison's CLI tool and Python library for working with large language models. The plugin gives LLM access to Mistral models through the Mistral API. Plugins of this type are normally installed or upgraded with llm install -U llm-mistral.
The release note does not describe the implementation. It does not say how reasoning output is exposed to users, whether reasoning tokens are shown separately from the final answer, or whether new options control reasoning behavior.
What is not disclosed
The source is a short release announcement, not a model announcement. It does not provide the following details about Mistral Large 4:
- Context window: not disclosed in the source
- Pricing per 1M input and output tokens: not disclosed in the source
- Parameter count: not disclosed in the source
- Benchmark scores: not disclosed in the source
- Training cutoff date: not disclosed in the source
- Exact release date of Mistral Large 4: not disclosed; the source says only that it is "newly released"
The source also does not say whether Mistral Large 4 is a dedicated reasoning model or a hybrid that offers an optional thinking mode. Readers should check Mistral's own documentation for the model's specifications and API identifiers.
Why the plugin matters
LLM plugins are how Willison's tool picks up new providers and model features. Reasoning models complicate the usual chat interface because they can return intermediate reasoning content alongside the final answer. Client libraries have to decide how to parse, display, store and log that output. Plugin support is therefore a prerequisite for using reasoning models in LLM workflows, including logging and scripting.
Willison's site currently lists 105 posts under its llm-reasoning tag and 68 under mistral. Reasoning support has been a recurring theme in LLM's development.
What this means
This is a tooling update, not a model launch. Its main value is as a secondary signal that Mistral Large 4 is available through the Mistral API, since the plugin targets that API. The release note is the only evidence in this source for the model's existence, and it carries no specifications.
For developers who use LLM with Mistral, the practical step is to upgrade the plugin before testing Mistral Large 4 or any other Mistral reasoning model. Teams comparing Mistral Large 4 against other reasoning models should wait for Mistral's official documentation. Cost, context length and benchmark performance cannot be assessed from this release.
Related Articles
AWS Quick adds chat-built flows, multi-agent orchestration and an updated web app
AWS Quick users can now describe a business process in chat and have Quick build a flow that orchestrates agents, skills, tools and data. The release also adds multi-agent orchestration, an updated web app with a daily briefing, Meta ads and Amplitude connectors, and new language support.
Meta puts Muse AI agent in a Chrome sidebar, with 200,000 users already
Meta's chief AI officer Alexandr Wang announced a Muse Chrome extension and said 200,000 people already use it. Muse also went live in Australia, and The Information reported a $300 price for the Muse Charm device.
AngelSlim publishes five GGUF builds of Hy4-flash, smallest at 125.73 GiB
AngelSlim has published five GGUF builds of tencent/Hy4-flash, a 223B-parameter model, with the smallest at 125.73 GiB. Speed tests cover a Mac M5 Max 128GB and an RTX Spark 128GB, and llama.cpp support still requires an unmerged pull request.
Nous Research brings open-source Hermes Agent to Microsoft Teams, adds Collective Wisdom skill sharing
Nous Research says companies will soon be able to run its open-source Hermes Agent directly in Microsoft Teams, with agents sandboxed in Azure Container Apps. A new feature, Collective Wisdom, lets employees share reviewed skills with colleagues, and the shared knowledge stays inside the company.
Comments
Loading...