Mistral Is SO Back: Inside Le Chonk, the West’s Biggest Open Model (Pierre Stock, VP Science, Mistral)

Mistral claims Le Chonk matches Chinese frontier open models using 3-4x less compute, with RL as its key efficiency lever and Mistral Large 5 already in training.

• Le Chonk specs: Mistral Large 4 ("Le Chonk") is a 1-trillion-parameter mixture-of-experts model with 50B active parameters, trained in ~2 months on 3,800 NVIDIA Grace Blackwell GPUs (~10MW) in Mistral's own European datacenters — 3-4x less compute than Chinese rivals.
• Open weight, closed capability: Mistral open-weights its strongest model but carves out a closed, hardened cyber-capability endpoint separately, unlike Meta, which keeps its best model fully closed.

Themes, sentiment and reasons are AI-written paraphrases, checked so they never quote an episode. They may contain errors. Not investment advice. Episodes remain the property of their creators.