Mistral VP: Le Chonk trained on just 10 MW (~4,000 H100s), begs investors for 1 gigawatt
_AndrewZhao · x · 2026-11-03
Mistral VP of Science Pierre Stock says Le Chonk was trained from scratch on just 10 megawatts of compute — roughly 4,000 H100s — a fraction of what competitors use, all on Mistral's own compute. His pitch: "Please give us 1 gigawatt." He adds the team spent a year building its RL training machine and claims Le Chonk is the best open-weight model on cyber defense, coding, and coding agents. A quoted official demo shows Mistral Large 4 writing a quantum chemistry SCF solver from a short prompt, validating it, and building a 3D visualization around it.
More from Infra
- Marvell's all-stock Lumentum deal called the signature deal of this bubble — TiernanRayTech · 2026-11-04
- SemiAnalysis: SpaceX to Become World's Biggest AI Cloud With 13GW in 1.5 Years — dylan522p · 2026-11-04
- ORNL Paper: Useful Quantum Computing Needs 4M Qubits and 800 Square Meters — anirbanbandyo · 2026-11-04
- Snowflake open-sources Arctic Platform, a modular post-training framework for LLMs — StasBekman · 2026-11-04
- Basalt 0.2.0: 300 t/s prose at higher-than-llama.cpp quality on a single RTX 5090 rig — jesdga95 · 2026-11-04
- Periodic Labs to detail SGLang-based RL infra for autonomous physics labs at SGLang Summit — BanghuaZ · 2026-11-04