Mistral AI releases Mistral Small 4

Made by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made

Released Mistral Small 4 — 119B total / 6B active MoE, 256k context, Apache 2.0. Merges Magistral (reasoning), Devstral (coding) and Small (instruct) into one, with reasoning effort adjustable as a parameter

Why it matters

Brought the trend of merging separate model families into a single set of weights to open weights

Source — Primary sources first; secondary reports are labelled “Press”

Officialmistral.aihttps://mistral.ai/news/mistral-small-4/

Added in Week 35, 2026 · See that week

On the map — ⑪ Model Architecture line

  1. Previous stopGoogle Research
  2. This record · stopMistral Small 4
  3. Next stopIntern-NCP Team
See it on the map

In the model family tree — Mistral family · 13 versions

  1. Earlier versionMistral 3
  2. This recordMistral Small 4
  3. Later versionThe latest version in this branch
See the whole Mistral family

Same week

Hugging Face archives the text-generation-inference repositoryOfficialLiu et al. publish Reflection in the DarkPaperMistral AI releases LeanstralOfficial

Nearby on the same axis

DeepSeek-V4PaperMistral AI brings in physics AI by acquiring and integrating Emmi AIOfficialNested Learning / HOPEPaperGoogle Research introduces Titans and MIRAS, the framework that generalizes itOfficial