Skip to main content
Original: Simon Willison · 16/03/2026

​
Summary

Mistral has launched Mistral Small 4, a versatile model unifying capabilities from its flagship models, and introduced Leanstral for the Lean 4 coding language.

​
Key Insights

“Mistral Small 4 is the first Mistral model to unify the capabilities of our flagship models.” — Describing the new model’s features and its significance.
“It supports reasoning_effort=“none” or reasoning_effort=“high”.” — Highlighting the model’s flexibility in reasoning capabilities.
“Leanstral, an open weight model that is specifically tuned to help output the Lean 4 formally verifiable coding language.” — Introducing another model aimed at a specific coding language.

​
Topics

​
Full Article

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” 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.

[AINews] Context Graphs and Agent Traces

Swyx · explanation · 50% similar

[AINews] ElevenLabs $500m Series D at $11B, Cerebras $1B Series H at $23B, Vibe Coding -> Agentic Engineering

Swyx · explanation · 49% similar

[AINews] The Unreasonable Effectiveness of Closing the Loop

Swyx · explanation · 49% similar
Originally published at https://simonwillison.net/2026/Mar/16/mistral-small-4/#atom-everything.