Mistral Releases Open Weights for Trillion-Parameter AI Model Targeting Cybersecurity
French AI startup Mistral has released the weights for its trillion-parameter model, Mistral Large 4 (ML4), nicknamed 'Le Chonk.' By providing open weights via Hugging Face and API, Mistral enables businesses and governments to deploy frontier intelligence on-premise, ensuring data sovereignty and protection against service cut-offs. The model is specifically positioned for cybersecurity applications, aiming to help organizations detect and respond to threats in real time. Mistral claims ML4 achieves state-of-the-art performance with significantly higher efficiency than competitors; specifically, it uses 47% fewer tokens per task compared to the Chinese open-weight model GLM-5.3. This release intensifies the competition between open-weight developers and closed-model providers like OpenAI and Anthropic.
Tickers: $MSFT
Topics: AI Model Release Cybersecurity Mistral Large 4 Open Source AI
Read the full article at TechCrunch
Part of the story Mistral AI Releases Improved Large 4 (Le Chonk) Open-Weight Model (3 articles)