Mistral AI

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 👇:

Transcribing speech is one thing. Knowing when someone is done speaking is another.

🎙 Voxtral 3’s end-of-turn detection helps developers identify when a user has finished speaking, making it easier to build responsive voice interactions without relying on science alone.

See it in action ⬇️

🆕 Le Chonk brings the brains. Mistral Studio gets it working.

Introducing managed agents and apps, plus new capabilities for workflows, enterprise connections, observability and evaluations in Studio.

Build faster. Deploy where you choose. Keep control.
See what’s possible. 🧵

🔴 LIVE TOMORROW

Office Hours are back. We’re talking Studio and Vibe, with a guest from the Studio team.

Drop your questions below 👇 or just come hang out.
📅 Nov 5
⏰ 6PM CET / 12PM EST
📍 YouTube: youtube.com/live/bTMRF7DcK…
📍 Discord: discord.com/invite/mistral…

Can Mistral Large 4 find a critical vulnerability, prove it and fix it? We tested it on a recent Django CVE.

133 agents map the code, hunt for flaws, test attack scenarios and confirm the bug. Then Le Chonk patches it.

Triage, repro, patch. Shown at 90x speed.