Announced a public preview of ML4, a 1-trillion-parameter open-weight model, with full weights planned for release later in October.

On October 6, Mistral AI opened a public-preview API for Mistral Large 4 (ML4), nicknamed “le Chonk,” and said its weights would follow by the end of October. The company describes it as a natively multimodal, trillion-parameter model with 49 billion active parameters; developers can test it through Mistral Studio while the weights remain unreleased. Mistral says it will use the preview period to red-team the model with cybersecurity leaders, vetted partners and state authorities, then publish further architecture, benchmark and post-training details alongside the weights. The staged rollout makes the model available for evaluation before unrestricted deployment, while positioning Mistral to challenge US and Chinese open-weight rivals and appeal to organisations seeking alternatives to closed models. Mistral’s launch materials highlight coding, cybersecurity, finance, manufacturing and visual grounding, but provide no API price or full benchmark results yet.