Introducing Mistral Large 4: Le chonk
Mistral just shipped a trillion-parameter open model. Here's what that means for the European AI stack.
Why it matters
Mistral Large 4 ('le chonk') is a trillion-parameter open-weight model release that expands the frontier lab's capability envelope and challenges the US-China duopoly on frontier scale. Practitioners building on open weights now have a substantive alternative at scale; enthusiasts tracking model race velocity see Europe entering the trillion-parameter tier.
The key facts
5 to knowMistral Large 4: trillion-parameter open-weight model
Positioned as European frontier capability release
Open-weight distribution (weights available, not API-only)
Competitive positioning vs. US/China frontier labs
Exact capability benchmarks and evals not specified in source URL (Simon Willison blog link — verify against official Mistral announcement for benchmark data, reasoning performance, context window, inference speed)
The story so far
Earlier coverage of this storyline
- H Company Releases Holo4: Open-Weight Computer-Use Models That Click, Code and Call Tools Across Desktop, Web, Android and APIsMarkTechPost
- Aleph Alpha Releases Kolibri: A 78.1B Open-Weight English-German MoE Model With Only 3.46B Active ParametersMarkTechPost
- Mistral Large 4 now available on AI GatewayVercel Blog
- Mistral AI Releases Mistral Large 4 (Le Chonk): A 1.05T Parameter Multimodal MoE ModelMarkTechPost
- Mistral Large 4Simon Willison
- This story
Go to the source
Simon Willisonsimonwillison.net