Mistral Releases Mistral 3, Including 675B-Parameter Model Under Apache 2.0
Mistral has released its Mistral 3 family of models, featuring Mistral Large 3, a sparse mixture-of-experts model with 675 billion total parameters and 41 billion active parameters. Additionally, three Ministral 3 edge models (3B, 8B, 14B) were released. All models are available under the Apache 2.0 license, making this the largest permissively-licensed model release to date.
The open-sourcing of a highly capable, 675-billion-parameter model under a permissive license could significantly accelerate enterprise adoption of open-weight AI alternatives. Its efficient sparse MoE architecture also offers strong performance with potentially lower real-world operational costs.
Learn one new AI thing every day.
Daily Deck sends you seven plain-English cards like this every morning. Free.
Start free