Mistral Large 4 changelog

Mistral Large 4 becomes available on Microsoft Foundry

Microsoft is the first cloud provider to offer Mistral Large 4, with private deployment options for Azure developers.

Who is affected:

Azure customers can run Mistral Large 4 as a managed service.

What to do:

Deploy Mistral Large 4 from Microsoft Foundry if you prefer a managed service.

Mistral Large 4 releases open weights with three reasoning modes at $0.70 input

Mistral released the weights of the improved Mistral Large 4 under a modified MIT license. It now has Low, Medium and High reasoning modes, and pricing is $0.70 input, $0.07 cached input and $2.10 output per million tokens.

Who is affected:

Developers and enterprises can self-host Mistral Large 4 or use it through the Mistral API, OpenRouter and Microsoft Foundry.

What to do:

Download the weights from Hugging Face or try the model in Mistral Studio.

Mistral Large 4 in public preview, open weights promised by end of October

Mistral released Mistral Large 4 (mistral-large-4) in public preview, an open-weight mixture-of-experts model with a 1M context window. Mistral says it will release the weights by the end of October 2026 after red-teaming. Listed API price is $1.36 input and $4.18 output per 1M tokens.

Who is affected:

Teams that want a self-hostable frontier-class model.

What to do:

Watch for the weights release at the end of October.