OpenAI Ultrafast mode now available on AI Gateway
- Source
- vercel.com
- Author
- Kevin Dawkins
- Date

Context
Vercel's AI Gateway, a service that lets developers call many AI models through one , now passes through OpenAI's Ultrafast service tier for GPT-6 Astra. In its changelog, Vercel says the tier gives faster output for interactive apps and quick coding loops, and a developer requests it with a service tier option on the openai/gpt-6-astra .
The trade is price and availability. Vercel says Ultrafast requests are billed at six times the standard per- rate, that the tier covers US and global processing, and that requests pinned to unsupported regions such as the EU run at the standard tier instead. When a request falls back, billing follows the tier actually served, and standard remains the default if no tier is named. The changelog does not give speed measurements; OpenAI's own launch is covered in a separate Intel item.
- AI Gateway exposes OpenAI's Ultrafast service tier for GPT-6 Astra. You opt in per request with the openai provider option serviceTier: 'ultrafast'; standard processing remains the default when no tier is specified.
- Ultrafast requests are billed at 6x the standard per-token rate, while a request that falls back to another tier is billed at the rate of the tier actually served.
- Ultrafast supports US and global processing only. Requests pinned to unsupported regions such as the EU run at the standard (default) tier instead.
- For workflows with frequent calls, OpenAI recommends using the Responses API over a persistent WebSocket connection to reduce overhead between turns.
- API — Application programming interface: the agreed set of operations, inputs and outputs through which one piece of software uses another.
- Model — A computational system that uses learned patterns to produce an output from information supplied to it.
- Token — A piece of text, such as a word, part of a word or a punctuation mark, that a language model represents as numbers and handles as a unit.
- Tool — An operation that the software around a model makes available for the model to request, with a name, expected inputs and a kind of result.
Checking sign-in…
Loading comments…






