One endpoint for every model we use, with per-team quotas, cost attribution, audit logging and failover between providers.
Next: generally available on 19 October 2026, in 2 weeks. Each audience arrives on its own schedule, below.
Currently reaching Product teams with a reviewed use case.
The AI platform team's own services
Phase 1 — pilot
Product teams with a reviewed use case
Phase 2 — reviewed use cases
All engineering
Phase 3 — everyone
Model access grew organically: several teams hold their own provider keys, spend is invisible until the invoice arrives, and there is no shared record of what was sent to a third party.
Every team calling a model API. The change is small — a base URL and a key — but it is mandatory, because the old keys are being revoked.
Move before 2 November. If you have a use case the gateway does not cover
yet, raise it in #ai-platform now rather than after the cutover.
From 2 November the provider keys held by individual teams are revoked. Calls made directly to a provider will fail with a 401; calls through the gateway are unaffected.
Switching over is a base URL change and a different key. The quickstart has the exact diff for each SDK we support.
Teams using models in production should move now rather than in October — the gateway gives you retry handling and a cost breakdown you do not currently have.