Key Points

Mistral Small 4 is a new Apache 2 licensed model with 119B parameters (Mixture-of-Experts, 6B active) released by Mistral AI.
It unifies the capabilities of Magistral for reasoning, Pixtral for multimodal tasks, and Devstral for agentic coding into a single, versatile model.
It supports two reasoning modes, reasoning_effort="none" and reasoning_effort="high", with the latter delivering verbosity comparable to previous Magistral models.
The model image is listed as 242GB on Hugging Face, and the article includes hands-on experimentation via the Mistral API and a sample command sequence.
Leanstral is also announced, an open-weight model tuned for Lean 4, highlighting a targeted language focus for formally verifiable coding.

Simon Willison’s Weblog

Sponsored by: CodeRabbit — Planner helps 10x your coding agents while minimizing rework and AI slop. Try Now.

16th March 2026 - Link Blog

Introducing Mistral Small 4. Big new release from Mistral today (despite the name) - a new Apache 2 licensed 119B parameter (Mixture-of-Experts, 6B active) model which they describe like this:

Mistral Small 4 is the first Mistral model to unify the capabilities of our flagship models, Magistral for reasoning, Pixtral for multimodal, and Devstral for agentic coding, into a single, versatile model.

It supports reasoning_effort="none" or reasoning_effort="high", with the latter providing "equivalent verbosity to previous Magistral models".

The new model is 242GB on Hugging Face.

I tried it out via the Mistral API using llm-mistral:

llm install llm-mistral
llm mistral refresh
llm -m mistral/mistral-small-2603 "Generate an SVG of a pelican riding a bicycle"

The bicycle is upside down and mangled and the pelican is a series of grey curves with a triangular beak.

I couldn't find a way to set the reasoning effort in their API documentation, so hopefully that's a feature which will land soon.

Also from Mistral today and fitting their -stral naming convention is Leanstral, an open weight model that is specifically tuned to help output the Lean 4 formally verifiable coding language. I haven't explored Lean at all so I have no way to credibly evaluate this, but it's interesting to see them target one specific language in this way.

Posted 16th March 2026 at 11:41 pm