Mistral Large 4, aka Le Chonk, is now in public preview. The natively multimodal model has 1T parameters with 49B active and was trained using nearly 4,000 NVIDIA Grace Blackwell Superchips. It is the best open weights model from US or Europe on aggregated benchmarks.
Mistral Large 4, aka Le Chonk, Enters Public PreviewAI News
Mistral Large 4, aka Le Chonk, is now in public preview. The natively multimodal model has 1T parameters with 49B active and was trained using nearly 4,000 NVIDIA Grace Blackwell Superchips. It is the best open weights model from US or Europe on aggregated benchmarks.