NoFOMO
Search 中EN Sign in Sign up

Peaked at #1 Now #1

Mistral Releases Le Chonk Model

Mistral AI releases open weights for Le Chonk, targeting agentic coding and cybersecurity, ranking #1 among open models on FrontierCode.

Rank over time

Key points

  • Mistral AI released the weights of a new and improved Le Chonk, ranking #1 on Cognition's FrontierCode among open models and setting a new benchmark for Pareto efficiency on Artificial Analysis.
  • The model is state-of-the-art in cyberdefense, tested by cyber vendors including CrowdStrike, Cloudflare, Rubrik and Zscaler, with weights downloadable on Hugging Face and customizable via Mistral Forge.
  • Unsloth AI said the 1 trillion parameter model can run locally on a Mac Studio with its 1-bit GGUF, which retains 91% top-1 accuracy.
  • TRL added day-0 support to post-train Le Chonk on a single node, with LoRA SFT on a 1T-parameter model running at 32.7k tokens per step in about 8 seconds and 142 GiB per GPU on 8× B300.

Key points and the reaction summary are written by AI from the posts on this page. Check the original post. How we use AI

Original post

Mistral AI @MistralAI 226K followers

Today, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity. 💻 Built to build: Le Chonk ranks #1 on @cognition’s FrontierCode among open models, and sets a new benchmark for Pareto efficiency on @artificialanalysis. 🛡 Built to defend: State-of-the-art in cyberdefense, put through the paces by the world's leading cyber vendors, including @CrowdStrike, @cloudflare, @rubrik, @zscaler, and others. Built for sovereignty: Make it yours. Download the weights on @huggingface , customize it with Mistral Forge, and deploy it wherever you need it. Read more from our team and customers 👇:
Post image
2,635likes 325reposts 198replies 151.3Kviews

View on X Save

What others are saying

4 more
  1. Unsloth AI @UnslothAI 219

    Unsloth says Mistral's 1T-parameter Le Chonk runs locally on a Mac Studio with a 1-bit GGUF retaining 91% top-1 accuracy.

    View on X
  2. vLLM @vllm_project 39

    vLLM adds day-0 support for Mistral Large 4, offering FP8 and NVFP4 checkpoints with a built-in DSpark drafter.

    View on X
  3. SGLang @sgl_project 44

    SGLang adds day-0 support for Mistral Large 4 (Le Chonk) and is first to serve it on NVIDIA Vera Rubin; a 1T MoE with 1M context.

    View on X
  4. MTS @MTSlive 47

    MTS reports Mistral has released the weights of the new Mistral Large 4 (Le Chonk) on Hugging Face.

    View on X

Top replies on X

Reaction on X: Replies were mostly congratulatory and playful, with users joking about the name ("Le BatChonk?", "Le Dark Chonk Rises") and one noting it ranks #5 open-weight model on the Vals Index; Unsloth AI added that it made GGUFs for local runs.

Sign in to see 5 top replies from X

From @Ariinahurry, @UnslothAI, @jambucke and others. Spam removed, with English and Chinese translations.

Sign up free Have an account? Sign in

Discussion 0

No comments yet. Start the conversation.

Suggest a source

Is there a first-hand source we're missing, or a topic we should watch? Tell us. Once approved, everyone's board covers it.

@username or profile link