Mistral AI announced Mistral Large 4, an open-weight, general-purpose multimodal model built on a granular Mixture-of-Experts architecture. The model has 49B active parameters, 1.05T total parameters, and a 1.6B vision encoder, with a 1M-token context window. It is available in public preview. Pricing is listed at $0.68 per million input tokens, $0.07 per million cached input tokens, and $2.09 per million output tokens.
No score is assigned. Sources and their independence are shown in the citation chain below.