Mistral, the French AI developer, has published the weights for Mistral Large 4 (ML4), its trillion-parameter open source model, enabling enterprises and governments to download and run the system on their own infrastructure. The model, affectionately called "Le Chonk," was unveiled earlier this month. This move intensifies the rivalry between open-weight developers like Mistral and closed-model players such as OpenAI and Anthropic, with Mistral contending that organizations shouldn't be dependent on external AI vendors for cutting-edge capabilities. That case may be especially compelling in cybersecurity, where entities seek greater control over their data and systems. As Mistral CEO Arthur Mensch put it, AI should "be a tool for empowerment, not dominance." With Chonk, Mistral reinforces that stance, claiming it can deliver "state-of-the-art performance" at a "fraction of the compute and capital committed by closed labs."

By releasing the model's weights—the numerical training that powers the system—Mistral ensures that other companies and government agencies can now deploy ML4 for their own needs. Mistral says its model performs comparably to GLM-5.3, a Chinese open-weight competitor previously regarded as one of the best open-weight models globally, particularly for cyber work. However, Mistral's model consumes 47 percent fewer tokens per task, making it considerably more economical than GLM-5.3. The company also asserts that the model leads all other open models on agentic coding tasks according to Artificial Analysis' Coding Agent Index, an AI benchmarking site. "Amid growing cyber threats, ML4 brings state-of-the-art sovereign cyber defense capabilities to enterprises and governments globally," the company said Thursday. "Enterprise and public sector institutions can now download, deploy, and own frontier AI to detect, analyze, and respond to threats across code and IT infrastructure."

The ML4 weights are available on Hugging Face and through its API. "Concretely, it means that, for instance, as a business company or as a government, you're able to have access to frontier intelligence to defend yourselves against cyberattacks. Access that can't be cut off by definition because the model is open-weighted," he said. Stock added that the model could be used to "identify, coordinate, and respond in real time to major security incidents," and that open sourcing means companies can deploy the model on-prem within their own ecosystems. While AI cybersecurity advances are helping security professionals uncover bugs at far higher rates than before, Stock continued, he believes it's crucial for companies to "provide [this same] capability to everyone…as soon as possible to ramp up and to strengthen the ecosystem," given that "some malicious actors might try to actually leverage their own technology to attack."

Separately, Stock addressed concerns about recent incidents involving rogue AI agents, acknowledging the risks but warning against overreacting. "We don't want to fearmonger people. We think it's a problem that is solvable. It's a hard problem to have good security," he said.