Mistral made its regional endpoints generally available, letting customers choose whether their AI requests are processed in Europe or the US. A new priority tier with custom rate limits and an uptime guarantee is in public preview, and the platform will begin hosting outside open-weight models, starting with Z.ai's GLM-5.2. Mistral also launched a scheme in which large European firms make multi-year compute commitments, with plans for up to 1 GW of capacity by 2030.
What changed
Customers could not pick the region their Mistral inference ran in, had no committed uptime guarantee, and could only run Mistral's own models on the platform.
What it unlocks
Running AI requests in a chosen region with a contractual uptime guarantee, and running an outside open-weight model on the same regional setup.
- up to 1 GW capacity by 2030
- 2 regions: Europe or US
What you need to act on it
- a Mistral platform account
- priority tier is in public preview, not general release
Sources