Term
Mistral Medium 3
Mistral Medium 3 is a proprietary AI model from Mistral AI released in May 2025. As an enterprise model built for high performance at low operating cost, it offers around 128,000 tokens of context and can also be run on-premise or in your own VPC.
Mistral Medium 3 — explained in detail
Mistral Medium 3 is a model from the French provider Mistral AI, released on 7 May 2025. Unlike the early Mistral models, Medium 3 is a proprietary model accessible commercially via the API and is not an open-weight model. It positions itself as an enterprise model designed to deliver frontier-level performance at significantly reduced operating cost.
Key data according to the provider: a context window of around 131,000 tokens as well as low API pricing (about 0.40 US dollars per million input and 2 US dollars per million output tokens). The model is geared towards coding, STEM reasoning and adaptation to enterprise workflows. A distinctive feature is its flexible deployment: besides API access, hybrid, on-premise and in-VPC deployments are possible.
With this, Mistral Medium 3 specifically targets companies looking for a powerful yet cost-efficient and flexibly deployable alternative to the larger frontier models.
Example / practical relevance
Mistral Medium 3 is suited to productive enterprise use, for example in coding assistants, in the analysis of technical documents or in tailored workflows. The option of on-premise or VPC operation makes it particularly attractive for organisations that must retain control over their infrastructure for data-protection or compliance reasons.
Distinction
Within the Mistral portfolio, Medium 3 sits between the small, partly open models and the large flagship models. Unlike Mistrals early open-weight models, Medium 3 is proprietary. Against competitors (such as Claude Sonnet or Llama models), it positions itself on the balance of performance, cost and flexible deployment. Successors such as Mistral Medium 3.1 and 3.5 continue this line.