--- title: Mistral description: "Mistral AI's model family and broader model portfolio, with general and specialist releases whose capabilities and licenses differ." canonical_url: 'https://darkfactory.dev/glossary/mistral' markdown_url: 'https://darkfactory.dev/glossary/mistral.md' collection: glossary date_published: '2026-10-07T00:00:00-04:00' date_modified: '2026-10-07T00:00:00-04:00' --- # Mistral ## Definition Mistral names Mistral AI's model family and broader model portfolio. The portfolio includes general language and multimodal models alongside specialist coding, reasoning, speech, and document models. Mistral AI is the developer; a specific model name identifies the artifact or service being used. The family name should be followed by the release and variant when discussing behavior. Mistral Small, Medium, and Large identify product lines, while names such as Codestral, Devstral, Magistral, and Voxtral identify specialist lines in the broader portfolio. They are not interchangeable names for one checkpoint. ## Origin and attribution Mistral AI released its first model, Mistral 7B, on September 27, 2023. Its announcement credits the Mistral AI team and distributes the model under Apache 2.0. Later models have their own architecture and license documentation. ## Scope and licensing Mistral's catalog includes multiple model architectures, sizes, input/output capabilities, and access arrangements. It can also list third-party models, whose developer remains separate from the host. A model may have downloadable weights, a hosted endpoint, or both. The catalog also distinguishes active models from retired releases. A historical model page remains useful provenance even when its endpoint has been withdrawn. The August 2026 license overview distinguishes Apache 2.0 releases from models governed by modified MIT terms. The latter include an additional commercial-license requirement above the overview's stated revenue threshold. Individual model cards carry the applicable terms. Read those terms for the release being deployed, separately from the API service agreement. Mistral's license overview describes its releases as open source. Under OSI's Open Source AI Definition, classification also depends on training code and data information. A permissive weight license answers the rights question for those weights; a complete-system claim needs the broader artifact review. ## Operational significance Record the model identifier, version, reasoning configuration, and serving method. A coding model and a document-recognition model solve different parts of a workflow, so evaluate each against the output it must supply. A portfolio benchmark cannot establish performance for every specialist or every task. For downloadable releases, check artifact provenance, compatibility, and resource requirements before deployment. With hosted models, track identifier changes and retirement dates. In both cases, validate generated code or extracted evidence before allowing it to drive an external action. ## Distinguish it from nearby terms - Mistral AI is an organization; Mistral model names identify its models and services. - Mixture of experts describes an architecture used in some releases, including the Mixtral line. It is not the definition of the entire portfolio. - Open-weight describes released parameters; proprietary describes retained control or restrictive access. A vendor can offer both arrangements across different products. ## Check your understanding An application switches between two Mistral releases because both are described as open models. Which capability, license, artifact, and serving details must be checked before treating the switch as equivalent? ## Also called Mistral model family, Mistral AI models ## Related terms - [Large language model (LLM)](https://darkfactory.dev/glossary/large-language-model) - [Multimodal model](https://darkfactory.dev/glossary/multimodal-model) - [Reasoning model](https://darkfactory.dev/glossary/reasoning-model) - [Open-weight model](https://darkfactory.dev/glossary/open-weight-model) - [Open-source AI](https://darkfactory.dev/glossary/open-source-ai) - [Proprietary model](https://darkfactory.dev/glossary/proprietary-model) - [Mixture of experts (MoE)](https://darkfactory.dev/glossary/mixture-of-experts) - [Model routing](https://darkfactory.dev/glossary/model-routing) - [Evaluation (eval)](https://darkfactory.dev/glossary/evaluation) ## Related factory areas - [Model selection, routing & budgets](https://darkfactory.dev/factory/model-routing-budgets) - [Runtime operations, observability, incidents & healing](https://darkfactory.dev/factory/runtime-operations) - [Security, privacy & software supply chain](https://darkfactory.dev/factory/security) ## Evidence and further reading - [Mistral AI: Mistral 7B](https://mistral.ai/news/announcing-mistral-7b/) - [Mistral AI: Model catalog](https://docs.mistral.ai/models) - [Mistral AI: Open model licenses](https://help.mistral.ai/en/articles/347393-under-which-license-are-mistral-s-open-models-available) - [Open Source Initiative: Open Source AI Definition 1.0](https://opensource.org/ai/open-source-ai-definition)