The French company Mistral has released its latest artificial intelligence model, Mistral Large 4 (ML4), known as “Le Chonk”. It offers advanced coding and cyber-defense capabilities, allowing organizations to run it on their own infrastructure. This ensures greater control over sensitive data and corporate cybersecurity, as reported by Euronews.

Mistral defines it as the “world’s best open agentic coding model”, providing businesses and governments with frontier AI cyber-defense capabilities to detect and respond to cyber threats. ML4 has demonstrated performance improvements and superior efficiency, using 47% fewer tokens than a Chinese model, according to Euronews.

This initiative is crucial for Europe, which aims to develop AI alternatives and address growing cyber threats. Running ML4 locally reduces dependence on external providers, a key aspect for technological sovereignty. Arthur Mensch, CEO of Mistral, highlighted the importance of open source, Euronews reports.

The ML4 model boasts 1.05 trillion parameters, activating approximately 52 billion for processing efficiency. It offers three reasoning modes adaptable to task complexity. Its applications include computer-aided design (CAD), financial data analysis, and territorial monitoring for various sectors, as reported by Euronews.