This is insanely fast, you have to try it.

Mistral is back. Seriously impressive. So fast that it put “Synthetic Reasoning Tokens” in my head…

Waiting for UK Prime Minister to show some support too...

Self-qualifying oneself as heavyweight while shipping nothing of significance looks like hubris to me

le chat doesn’t disappoint

⚔️ MiniHack Updates! ⚔️ 1️⃣ MiniHack 1.0.0 is here! Following popular demand, it now supports the new Gymnasium API and is built on NLE 1.1.0. Huge thanks to @Stephen_Oman (maintainer of The NetHack Learning Environment ) for his outstanding contribution! 🙌

Join us to make revolutionary video games using AI. The potential for a dynamically generated video game using AI is incredible!

Hugging face from my childhood

Claude Plays Pokémon continues on as a researcher's personal project. Follow along on Twitch: twitch.tv/claudeplayspok…

🎉 Thrilled to share that I’ve been awarded the Sarah A. Burgess Award from the Society for the Neuroscience of Creativity! The Society for the Neuroscience of Creativity Honoured to have my research in AI, creativity, and HCI recognised. Huge thanks to my mentors & collaborators for their support! #AI #Creativity #HCI

Introducing Mistral AI Le Chat Enterprise, the most customizable and secure agent-powered AI assistant for businesses. 🧠 Integration with your company knowledge (starting with Gmail, Google Drive, Sharepoint…) 📚 Ability to add frequently used documents for better-informed

Self deployable fully packaged frontier AI. If you’re a country you may want to look here instead, we can help!

Mistral Medium 3 independent evals: Mistral is back amongst the leading non-reasoning models with Medium 3 rivalling Llama 4 Maverick, Gemini 2.0 Flash and Claude 3.7 Sonnet Key takeaways: ➤ Intelligence: We see substantial intelligence gains across all 7 of our evals compared

I don’t think we took Le Chat seriously enough…

I really want to know what this sounds like. VRR VRR VRRRRRRR

I don’t really like to post “hot takes” but it’s quite sad to see many people dunking on each other’s work. We’re all working hard, doing amazing things that would’ve been impossible a few years ago. There’s plenty of success to go round, let’s stop throwing toys out of the pram!

Meet Devstral, our SOTA open model designed specifically for coding agents and developed with All Hands AI mistral.ai/news/devstral

excited to finally share on arxiv what we've known for a while now: All Embedding Models Learn The Same Thing embeddings from different models are SO similar that we can map between them based on structure alone. without *any* paired data feels like magic, but it's real:🧵

Very happy to release Codestral Embed, our first code embedder. It can use up to 3072 dimensions, ordered by relevance. Using embeddings of 256 dimensions with int8 precision is already sufficient to outperform all existing models.

The best part of working at Mistral AI is the cheese and wine reception after the weekly all-hands