mistralai/Mistral-Large-3-675B-Instruct-2512
Updated
•
247
•
160
Mistral Large 3
A state-of-the-art, open-weight, general-purpose multimodal model with a granular Mixture-of-Experts architecture.
Note A no-loss FP8 version to reduce resource requirements. Can be deployed on a node of B200s or H200s.
mistralai/Mistral-Large-3-675B-Instruct-2512-NVFP4
Updated • 202 • 35Note A high quality NVFP4 version to reduce resource requirements. Can be deployed on a node of H100s or A100s.
mistralai/Mistral-Large-3-675B-Instruct-2512-Eagle
Updated • 92 • 24Note An EAGLE speculator for Mistral Large 3 in FP8, used for speculative decoding.
mistralai/Mistral-Large-3-675B-Base-2512
Updated • 63 • 30Note Base BF16 Weights, used for fine-tuning.