On October 21, 2026, OpenAI’s API changelog said you can choose which service tiers a project may use in Platform settings, and that organization settings can restrict tiers for all projects. Organization rules win: a project cannot turn on a tier the organization has blocked.
The policy covers `default`, `flex`, and `priority`. A request with `service_tier: "fast"` is evaluated as `priority`. Scale Tier is not part of this project policy. Related reading: GPT-6.1 Sol Ultrafast: Price, Speed and Who Can Use It. See also How OpenAI Prompt Caching Pricing Works for 12-Hour and 30-Minute TTL (October 2026).
If the request picks—or resolves to—a tier the project does not allow, the API returns HTTP 400 as an `invalid_request_error`, with `error.param` set to `service_tier`. Leaving `service_tier` out or setting it to `auto` can still fail when the resolved tier is blocked. OpenAI’s error guide says to use an allowed tier, or change the allowed list in project and organization settings.
For the Platform settings UI itself, we have not tested it yet; if you have, you can send feedback through the site contact link in the footer. Details above are for reference only.