Mistral Large 4 hits 38 points in Intelligence Index, a big leap

Mistral has been building its own infrastructure in Europe for months. In March, the company took out an $830 million loan for a data center near Paris, with 200 megawatts of compute capacity in Europe planned by the end of 2027. In May, Mistral renamed its chatbot Le Chat to Vibe and rebuilt it as a work tool, shifting focus from consumers toward enterprise customers. The current Series D round of 3 billion euros, the largest equity round ever raised by a European tech company, funds this compute expansion.

Mistral's emphasis on cybersecurity comes after CEO Arthur Mensch warned a French parliamentary commission in May that Europe risks becoming dependent on US models for cybersecurity. He said the French military's codebases shouldn't be scanned by Anthropic's Mythos, and that Mistral's own models could find the same vulnerabilities linked to Mythos. Mistral argues that closed models' safety filters block legitimate vulnerability research, while attackers jailbreak those same models anyway. How ML4 reliably distinguishes legitimate research from attack prep, Mistral doesn't explain.

The model's competitive position hinges on the ongoing reinforcement-learning run that Mistral says shows no signs of plateauing, with significant improvements expected over the coming weeks. If that bears out, ML4 could narrow the gap to Claude Opus 5.5 (Max)'s 58 points while strengthening its security pitch. But the weights are expected at the end of October, and until then ML4 remains proprietary and red-teamed with vetted partners.