Mistral just put the weights for Mistral Large 4 (ML4), nicknamed “Le Chonk,” on Hugging Face.
What’s in it
It’s a trillion-parameter open-weight model aimed squarely at cyber defense. Mistral says it’s on par with GLM-5.3, one of the strongest open models for security work, while using 47% fewer tokens per task. It also claims the top open-model spot on Artificial Analysis’ Coding Agent Index. It’s on Hugging Face and the API. Mistral’s VP of Science, Pierre Stock, said the point is access “that can’t be cut off by definition.”
My take
This is no home-lab model. A trillion parameters needs serious iron. But for companies and governments that won’t send their logs to someone else’s cloud, open weights for defense are a big deal. The same weights are open to attackers too, so the race goes both ways.
God Bless, and May the Source Be with You!
