Priority Became Fast Mode — What Changed

Fast mode and Priority processing are the same thing under two names — a rename made directly in OpenAI's own pricing documentation, with both the old and new service-tier identifiers still accepted by the API.

What actually changed

The name, and nothing else this site can find evidence of. Pricing, the underlying latency benefit, and which models offer it at all appear unchanged across the rename — this is a naming update to the pricing page's own footnote, not a repricing or a capability change bundled in alongside it.

Why a rename like this is worth its own page

A service-tier identifier that still works under its old name is exactly the kind of thing that causes quiet confusion months or years later — a script, an internal wiki page, or a teammate's mental model built around "Priority" doesn't automatically update just because the documentation moved on, and both identifiers being accepted means nothing forces the update either. Recording the rename explicitly, with the date it happened, is a smaller service than it sounds like: it's the difference between a developer finding a clean answer and one finding two names for the same thing with no explanation of which is current.

What to actually use going forward

New code and new documentation should use the current fast service-tier identifier — it's the name OpenAI's own pricing page now uses, and there's no reason to keep writing new integrations against a name that's already been superseded, even though the old one still functions.

Not offered everywhere, under either name

This tier — whichever name you call it by — isn't published for every model on the current roster. Several models, including some of the largest, have no Fast/Priority row in the pricing table at all, and that absence carries over unchanged from before the rename.