Arthur Mensch completed his PhD at Inria and spent three years at Google DeepMind co-authoring pioneering papers on Retro, Chinchilla, and Flamingo transformers. In April 2023, Mensch joined forces with Meta AI alumni Timothée Lacroix and Guillaume Lample to launch Mistral AI.
Championing open-weights foundational models (Mistral 7B, Mixtral 8x7B, and Mistral Large), Mensch proved that hyper-optimized, efficient architectures could rival the world's largest closed proprietary models, raising over $1B from Andreessen Horowitz, Lightspeed, and Microsoft.
