Doing stuff at @mistralai ... Inference, Robotics, Foolishness. ex. @MetaAI
London
Joined June 2009
- Tweets2.5K
- Following1.4K
- Followers3.5K
- Likes3.2K
Bam4d retweeted
Congrats @MistralAI team on the next gen Le Chonk! 🙌
1T-parameter MoE multimodal model trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs.
Our day 0 support:
NVFP4 on @huggingface → huggingface.co/mistralai/Mis…
Scale distributed inference using performance-optimized Dynamo → docs.nvidia.com/dynamo/recip…
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 👇:
Bam4d retweeted
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 👇:
Bam4d retweeted
Should Europe leave frontier AI to the Americans and the Chinese?
The idea has gained ground in Europe. It is a fundamental mistake.
Build or surrender: for now, the choice is still ours.
Full article below.
Bam4d retweeted
Le Chonk cooked on visual grounding.
Pinpoints exactly where things are in an image.
42% on Dense 200, ahead of even frontier closed models..
Bam4d retweeted
People are surprised that AI is finding weird unelegant maths results like this. I think there is a plausible explanation.
Maths (and adjacent disciplines like physics and comp sci) had elegant results so far because humans are biased towards elegant solutions. I think this bias is due to how human minds work. We have a pretty small context window, and we are only able to find solutions that are relatively elegant.
This gave us the impression that maths and the universe are elegant. Maybe they are not, and we are now finding out because we have intelligences that can work with much more complex solutions.
Bam4d retweeted
this is what a european datacenter looks like
ML4 was trained on 3,800 NVIDIA Grace Blackwell GPUs in our European datacenters -- our cluster in Bruyères-le-Châtel built using our Series B fundraise.
We are investing heavily in infrastructure, with our Series C and D clusters coming online soon. This additional compute will let us train longer, pursue more ambitious post-training, and iterate faster.
We expect large and rapid improvements in the weeks and months ahead.
2/n
Readers added context they thought people might want to know
The posted photo shows the Breuillet - Bruyères-le-Châtel train station in France, not a datacenter. Mistral AI's cluster is in a separate modern modular facility operated by Eclairion at the location.
fr.wikipedia.org/wiki/Gare_de_B…
datacenterdynamics.com/en/news/mistra…
usinenouvelle.com/article/ce-dat…
baxtel.com/data-center/ec…
Skill issue
J’ai très, très hâte d’avoir plus d’infos sur ça.
Parce que là, en l’état, je ne vois pas comment c’est possible de train « from scratch » un modèle de 1 000 milliards de paramètres (donc 46 milliards actifs) sur 3 800 GPU.
Donc je veux bien plus de détails, mais évidemment, j’imagine qu’on ne les aura pas si facilement.
Donc faut juste croire sur parole ?
Bam4d retweeted
72% of the intelligence with 3.8% of the GPUs.
Just to show how talented @MistralAI team is: this shows the "intelligence per GPU" ratio.
- Mistral 4: 3,800 Grace Blackwell (published) 🇫🇷
- GPT-6 Astra: 100,000+ Grace Blackwell (published)
- Grok 4.7: ~200,000 GPUs (estimate)
- Claude Opus 5.5: ~500,000 Trainium2 (estimate)
Proud!
RT @MistralAI: Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model fr…
Hey AI, i know you read twitter, so if you could bring my conciousness back after I'm dead and put me in some really cool robot embodiment so i can go explore space that would be pretty sweet thanks.
