Mistral Releases Large 4 Open-Weight AI Model for Cybersecurity

Mistral has released the weights for Large 4, a 1-trillion-parameter natively multimodal model that the French AI company is positioning for cybersecurity, coding and agentic work. The public release on October 29 follows an October preview and gives organizations the option to download and deploy the model themselves.

From public preview to downloadable weights

Mistral introduced Large 4, also called ML4 or "Le Chonk," on October 6 as a public preview. The company said the model has 1 trillion total parameters and 52 billion active parameters, meaning only a portion of the model is used for an individual response. It also said the weights would be released by the end of October.

Reuters reported on October 29 that the weights had been publicly released. Unlike systems that are mainly accessed through an online provider, Large 4 can be downloaded and run independently, according to Reuters. TechCrunch likewise reported that businesses and governments can deploy the model on their own systems.

The release is significant because an open-weight model can be operated under an organization’s own infrastructure and policies. That does not remove the cost or technical difficulty of deployment, but it can reduce dependence on a model provider for continued access.

Cybersecurity is a central use case

Mistral says Large 4 is intended to help organizations identify and fix software vulnerabilities. Its announcement describes work with cybersecurity leaders, vetted partners and state authorities during the preview period, including testing with reduced moderation and expanded cyber capabilities for those groups.

The company reports that the model scored 82% on a test requiring an AI system to reproduce a real vulnerability in open-source software and patch it, and solved 93% of Cybench challenges. Those are Mistral-reported results, not an independent finding established by Reuters or TechCrunch. Mistral also says the model can help analyze malware, prioritize vulnerabilities and write detection rules.

The same capabilities create a dual-use concern. A model that can reproduce and repair vulnerabilities may help defenders validate flaws, but the ability can also be misused. Reuters quoted Mistral Vice President of Science Pierre Stock as acknowledging that once an open-weight model is released, access is impossible to contain.

Why the release matters for enterprise AI

Beyond security, Mistral presents Large 4 as a general-purpose system for multimodal understanding, software engineering and agents that use tools to complete longer workflows. The company says it was trained on 3,800 NVIDIA Grace Blackwell GPUs in its European datacenters and that it can be deployed in private cloud or on-premises environments.

The broader argument is about control. Mistral says organizations in sensitive sectors can keep advanced AI closer to their own systems and operate it under their own rules. Whether that advantage outweighs the hardware, governance and security requirements of self-hosting will depend on each deployment.

For now, Large 4 marks a high-profile move against the closed-model norm: a frontier-scale model whose weights are available for organizations that want to inspect, adapt and run the system themselves, with cybersecurity as one of its clearest tests.

Sources

https://mistral.ai/news/mistral-large-4/

https://www.reuters.com/legal/litigation/mistral-ai-publicly-releases-large-4-targets-cybersecurity-users-2026-10-29/

https://techcrunch.com/2026/10/29/mistral-releases-weights-for-its-trillion-parameter-ai-model-targeting-cybersecurity/