Mistral Ships 'Le Chonk': A 1-Trillion-Parameter Open-Weight Model Trained in Europe

Mistral released Large 4 in research preview — one trillion parameters, 49B active, natively multimodal, trained in its own European datacenters — with open weights promised for October 27.

Mistral entered the frontier open-weight race on Wednesday with Large 4, nicknamed "Le Chonk." According to @ivke2006, the model is a one-trillion-parameter mixture-of-experts with 49 billion active parameters, trained entirely in Mistral's own European datacenters, with weights scheduled to land on October 27. @TraffAlex confirmed the release as a research public preview and added that the model is natively multimodal.

Unlock the full briefing

Get every story in today's briefing, the full archive, and the daily AI intelligence brief.

All stories today

Full archive

Daily brief

Cancel anytime. Payments powered by Stripe.