Mistral: Mistral Medium 3

Mistral Medium 3 is an enterprise-grade language model from Mistral, released by the provider on 7 May 2025. It sits in the text category and is positioned as a high-performance model that balances strong reasoning with multimodal input handling. If you are evaluating it for a production workload, the key point is that Mistral built it for general text generation and reasoning tasks rather than for a single narrow niche.

The model accepts text, images and files as input, and it produces text as output. Its context window is 131,072 tokens, and the maximum output length is 104,857 tokens, so it can handle long documents and still return substantial completions. The knowledge cutoff is 31 March 2025. Multimodal input plus a large context makes it suitable for tasks where you need to reason over a mix of prose and visual or file-based material in one request.

Practical use cases follow directly from those capabilities. You can use it for document analysis and summarisation, including PDFs or other files that need to be read and reasoned over. Image input allows it to describe, interpret or extract information from visuals alongside text instructions. The long context supports codebase review, multi-document comparison and long-form drafting. Its tool and structured output support also makes it a candidate for agents and pipelines that need reliable machine-readable responses rather than free text.

You can call Mistral Medium 3 through api14's OpenAI-compatible API using the model id "mistralai/mistral-medium-3". Because the interface is OpenAI-compatible, existing clients and SDKs that target the OpenAI chat completions format work with minimal changes. The model supports frequency_penalty, max_tokens, presence_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools and top_p. That combination covers sampling control, deterministic runs via seed, stop sequences, JSON or schema-constrained output, and function or tool calling.

One limit worth noting is the output modality: the model returns text only, so it cannot generate images or audio. If your task depends on image or audio generation, or on capabilities outside text and multimodal input reasoning, another model would fit better. For text generation, document reasoning and tool-driven workflows, though, Mistral Medium 3 is a reasonable candidate given its context size, file and image input, and parameter support. Check the live price table on this page for current rates before committing.