Mistral
Mistral is the model family of Mistral AI, a Paris-based lab founded in 2023 · a mix of open-weight models and commercial models served through its API.
Text reviewed October 5, 2026
Mistral is the model family of Mistral AI, a Paris-based lab founded in 2023 · a mix of open-weight models and commercial models served through its API.
Basic
Mistral AI released Mistral 7B in September 2023 under the Apache 2.0 license, then Mixtral 8x7B, an early open-weight mixture-of-experts model, in December 2023. Its lineup now spans small, medium and large models, plus coding and reasoning models, some open-weight and some available only through Mistral's API and partner clouds.
Deep
Mistral 7B used grouped-query attention and sliding-window attention to run efficiently. Mistral serves its models through its own platform (la Plateforme), its Le Chat assistant and cloud partners, and offers on-premise deployment for enterprises.
Expert
License terms differ by model (Apache 2.0 for many open models, commercial terms for others). As with other open-weight families, the same weights can be hosted by several providers at different prices.
Mistral AI keeps shipping new Mistral models; the live block on this page lists the newest ones tracked on BenchGecko with their release dates, prices and BenchGecko scores, so this definition never has to name a "latest" model.
Depending on why you're here
- ·A French AI company and its models
- ·Check the license per model
- ·See /family/mistral for every tracked model
- ·European AI lab; dated figures on /economy/company/mistral-ai
- ·Early open-weight MoE (Mixtral 8x7B, Dec 2023)
- ·Grouped-query and sliding-window attention in Mistral 7B