Congratulations Mistral on launching their first reasoning model *and* writing an excellent report!

Mistral is significantly scaling our compute, by investing in our own infrastructure. This will not only allow us to scale our own models, but also offer others AI compute from datacenters based here in Europe.

Another year, another best paper from Visual Geometry Group (VGG)! Sad to miss out on the celebrations. Amazing work Jianyuan Minghao Chen Nikita Karaev David Novotny (and of course Chris and Andrea)

Small update for Small 👶🏻

We started Genesis AI to solve general-purpose robotics and unlock unlimited physical labor. We’re backed by $105M from Eclipse Ventures, Khosla Ventures, HSG, Bpifrance, Alven, Northzone, Kearny Jackson, Fly Ventures, Eurazeo, Drysdale Ventures, Motier Ventures, Alexandre Berriche,

📣 Anthropic Zurich is hiring again 🇨🇭 The team has been shaping up fantastically over the last months, and I have re-opened applications for pre-training. We welcome applications from anywhere along the "scientist/engineer spectrum". If building the future of AI for the

SIGKITTEN Good idea. Let me see what we can do.

Great conversation today with President Emmanuel Macron, @ArthurMensch, & @AmandaWolt at @ImperialCollege on how international collaboration - between France & the UK and beyond - can help maximise AI's potential to accelerate scientific discovery & opportunity.

Voxtral: our first set of audio understanding models. We release in two model sizes: * At the low end, our 3B outperforms Whisper Large and matches Gemini Flash * At the high end, our 24B matches Scribe for less than half the price And: Apache 2.0!

Thrilled to finally share what we’ve been building past few months! Audio used to be a black box for me, now I’m deep in the box, with more out-of-the-box ideas cooking. Enough with the box..introducing Voxtral. Grateful for the intense and rewarding learning curve at Mistral AI.

Mistral Medium 3.1 just got graded on LMArena. On English prompts (with Style Control) it ranks 4th, above models like GPT-5-Chat, Grok 4 and Sonnet 4. Great work by the team. Try it for free with ⚡️ inference on Le Chat.

Le Warmup

You can now search, summarize, and manage your Linear issues and projects through Mistral AI Le Chat.

20 months ago, we a16z led Mistral AI's $500M Series A at $2B Some laughed. Scientists? Serving giant enterprises? Open source AI? From Europe? Good luck! Today: We closed $2B at $13.7B+ led by ASML, and $1.6B+ TCV, deploying RL at scale for critical industries LE WARMUP

Today Mistral announced our new $2B funding round led by ASML. In just two years, we've gained hundreds of M in commercial traction in the US, Europe and Asia; grown the team by 10x; and released SoTA OSS models in text, vision and audio This is really just the start. Join us!

Introducing Magistral Small 1.2 & Magistral Medium 1.2, minor updates to our Magistral 1.1 models! - Multimodality: Now equipped with a vision encoder, these models handle both text and images seamlessly. - Performance Boost: 15% improvements on math and coding benchmarks such

Our latest Magistral models just made a massive leap on the Artificial Analysis Intelligence Index—competing with models much larger in sizes.

Nice! Congrats Nikita Karaev and crew!