Mistral Large 4

Mistral Large 4 is a general-purpose, open-weight multimodal model documented on Mistral AI’s developer docs site. It is positioned as a state-of-the-art model for developers who want strong reasoning and vision capabilities without being locked into a closed API, and the documentation page serves as the reference for its weights, features, and usage.

The model uses a granular Mixture-of-Experts architecture with 52B active parameters and 1.05T total parameters, plus a 1.6B vision encoder for image understanding. It supports a 1M-token context window and exposes features such as structured outputs, function calling, document Q&A, prefix completion, chat completions, batching, and agents and conversations through the Mistral API endpoints.

Pricing is listed per million tokens: $1.36 for input, $0.14 for cached input, and $4.18 for output. The page marks the model as a public preview release, so it is best suited to developers, ML engineers, and product teams evaluating or prototyping multimodal applications, including document analysis, conversational agents, and batch processing pipelines.

Before adopting it, teams should review the model lifecycle policy, regional inference options, and priority tier details linked from the docs, and test the preview version against their own workloads. Overall, Mistral Large 4 is a well-documented option for teams that value open weights, long context, and multimodal input in a single API-accessible model.