Mistral's €3B-backed "Le Chonk" is a 1-trillion-parameter open model

What's the deal? Mistral has launched a public preview of Mistral Large 4, a 1-trillion-parameter model it nicknames "le Chonk." The French startup calls it its largest and most capable model to date, available now via its Mistral Studio API, with open weights due by the end of the month.

The details: ML4 is a natively multimodal model with 49 billion active parameters. Mistral says it performs strongly across coding, agentic workflows, and multimodal understanding.

What's the endgame? Mistral positions the model as "competitive with the strongest open-source models globally," while claiming it outperforms any open-weight model built in the US or Europe. It was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's own European datacenters.

Why now? ML4 is the first milestone on a roadmap funded by Mistral's €3 billion Series D — which the company calls the largest equity round ever raised by a European technology company. That capital is going toward scaling compute in Mistral's European datacenters.

The signal: By training and serving the model on infrastructure it operates end-to-end, independently of other digital service providers and under European law, Mistral is pitching sovereignty as a selling point. For European firms wary of US cloud and AI dependence, an open-weight frontier model built on home soil is the clearest argument yet.

Image credit: GBPublic_PR