Mistral Large 4: Mistral launches new open-weight AI model

French AI company Mistral AI introduced Mistral Large 4 (ML4, internally nicknamed "le Chonk") on October 6, 2026. It is described as the company's largest and most capable model to date, currently available as a public preview.

What Mistral Large 4 is

According to Mistral's announcement page, ML4 is a natively multimodal mixture-of-experts model with 1 trillion total parameters and 49 billion active parameters per task. It was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's own European data centers, using training data spanning more than 160 languages, including every official EU language.

Preview now, full weights by month's end

The preview can be tried today through the API on Mistral Studio (console.mistral.ai). Mistral says it will release the full model weights by the end of October 2026. In the meantime, the company is red-teaming the model with cybersecurity partners and select government authorities using a version with reduced moderation and expanded cyber capabilities.

API pricing

Per Mistral's page, API access to ML4 is listed at $1.36 per million input tokens and $4.18 per million output tokens.

Who it's built for

Mistral positions ML4 for coding, agentic workflows, complex document and image analysis, and business use cases such as finance and legal work. The company also highlights that the model runs independently on European infrastructure under EU law, tied to its recent €3 billion Series D funding round.

Because this is an API-based model aimed at developers and businesses building their own products rather than a consumer chat app, everyday users would access it indirectly through platforms or tools that integrate Mistral.