What Is Mistral AI? The OpenAI Competitor That Plays by Its Own Rules

Mistral AI keeps making headlines, but it’s easy to misunderstand what the French company actually does. Judge it by how close it is to becoming “the OpenAI of Europe” and you’ll miss the point.

Mistral follows the Palantir playbook. It deploys engineers directly into governments and large corporations, tailoring AI for specific use cases rather than chasing ChatGPT-level brand recognition. Its chat app Vibe (formerly Le Chat) doesn’t come close to ChatGPT’s reach, but that’s not the goal.

The numbers tell a different story. Mistral’s annual recurring revenue hit $400 million in February, up from $20 million a year earlier. The company claims it’s on track to surpass $1 billion this year. It’s reportedly raising $3.5 billion at a $23.15 billion valuation.

Founded by ex-DeepMind and Meta researchers, Mistral offers everything from small on-device models to multimodal and reasoning systems. Some are open-weight. It recently open-sourced its Leanstral code agent.

CEO Arthur Mensch positions the company as a sovereignty play. “We exist to make sure everyone gets access to the best AI systems outside of centralized control,” he wrote. Mistral is building data centers in France and Sweden, and acquired infrastructure startup Koyeb to build what it calls “a true AI cloud.”

If you’re betting on AI diversity — non-US models that governments can run on their own terms — Mistral is worth watching.