Mistral AI launched Mistral Large 4, nicknamed “le Chonk,” on October 6, 2026, as a 1-trillion-parameter native multimodal mixture-of-experts model with roughly 49–52 billion active parameters per token and a 1-million-token context window. The public preview API is live on Mistral Studio while the company runs additional reinforcement learning and shares a less-restricted version with cybersecurity partners and authorities for red-teaming ahead of open-weights release around October 27. The model was trained from scratch on Mistral’s own European Grace Blackwell GPUs and is positioned as the strongest open-weight system outside China on enterprise benchmarks in coding, finance, and cyber defense. With €3 billion in fresh Series D capital funding expanded compute, the lab expects continued rapid gains, making an incremental 4.1+ update the next logical milestone once the current checkpoint stabilizes.