Mistral Medium 3
Mistral AI
At a glance
- Price per 1M tokens (input)
- $1.50
- Price per 1M tokens (output)
- $7.50
- Context
- 262,144 tokens
- Free access
- yes (see below)
Refreshed daily; data verified August 30, 2026. Published August 30, 2026.
Mistral Medium 3 is aimed at enterprises that need a balanced mix of language, vision, and procedural capabilities without the overhead of the largest models. It works well for tasks such as document analysis with embedded images, automated workflows that require tool calling (function calling), and multi‑step problem solving that benefits from explicit reasoning. The model’s architecture blends a moderate parameter count with a training regime that emphasizes cross‑modal understanding, letting it interpret visual inputs while maintaining fluent text generation. Mistral’s approach focuses on open research practices and modular tooling, giving developers clear interfaces for extending functionality.
Specifications & pricing
| Input (per 1M tokens) | $1.50 |
|---|---|
| Output (per 1M tokens) | $7.50 |
| Context window | 262,144 tokens |
| Max output | 262,144 tokens |
| Capabilities | images, tool calling, reasoning |
LiteLLM community dataset (MIT), verified August 30, 2026. Official Mistral AI pricing.
What Mistral Medium 3 would cost on your workload — run it through the cost calculator →
Where to try Mistral Medium 3 for free
- Mistral AI offers a free chat — Le Chat (free plan). A vendor's free chat may run a different model from the same family — the exact model is not guaranteed.
Frequently asked questions
What types of business applications get the most value from Mistral Medium 3?+
The model shines in scenarios where text and images need to be interpreted together, such as processing scanned contracts, extracting information from receipts, or tagging product photos. It also excels when a workflow must invoke external services, because tool calling lets the model format and dispatch function calls automatically. Finally, any process that benefits from step‑by‑step reasoning—like troubleshooting guides or compliance checks—can use its explicit reasoning chain to improve transparency.
How do I integrate Mistral Medium 3 into my current system?+
Start by obtaining an API key from the provider’s developer portal and configuring your client library to point at the model endpoint. The API accepts both text and image payloads and returns a structured response that may include a function call object when tool calling is triggered. Sample code snippets in common languages illustrate how to send a request, handle the function call payload, and feed the result back into the model for continuation. Documentation also covers best practices for rate limiting and error handling.
In what ways does this model differ from other Mistral offerings?+
Compared with the smaller models, Mistral Medium 3 adds native image understanding and a more robust tool‑calling engine, while still keeping inference costs lower than the flagship large model. The large model provides deeper reasoning depth and higher fidelity generation but requires more compute resources. The medium variant therefore occupies a middle ground: it delivers cross‑modal capability and reliable step‑wise reasoning without the latency and expense of the largest option.
What limitations should I be aware of before deploying it?+
The model does not retain memory across separate sessions, so each request must contain all necessary context. Its image resolution handling is tuned for typical document‑scale inputs, so very high‑resolution visuals may need preprocessing. Tool calling relies on a predefined schema, meaning any function you wish to invoke must be registered ahead of time. Finally, while the reasoning chain is explicit, it may still produce plausible‑but‑incorrect steps, so validation logic is recommended for critical workflows.
Compare with others
Org chart: how to move your company onto AI
A practical map: which company roles and processes AI agents can take over, where to start, and in what order to roll it out.