Mistral open-sources trillion-parameter AI model weights, focusing on the cybersecurity domain

Mistral released the weights of Mistral Large 4, a trillion-parameter open-source AI model; enterprises and government agencies can download and deploy it themselves, with a focus on cybersecurity. Mistral claims performance comparable to GLM-5.3, 47% fewer Tokens per task, and that it surpasses all other open-source models on the Artificial Analysis agentic coding index. The weights are available on Hugging Face and via API.