Vercel
Ultrafast mode is now available through Vercel’s AI Gateway, giving builders faster OpenAI output for interactive apps and rapid coding iterations.
The changelog entry lists two supported models: GPT 6 Astra and GPT 6.1 Sol. Builders can request the service tier through Vercel’s AI SDK, Chat Completions API or Responses API.
This extends Vercel’s recent run of AI Gateway updates, with this release focused on response speed rather than routing or fallback logic.
How Vercel Ultrafast Mode Works
To use it through the AI SDK, select openai/gpt-6-astra or openai/gpt-6.1-sol, then set OpenAI’s serviceTier option to ultrafast. The request can also be restricted to OpenAI through the gateway provider settings.
The same service tier is available through the Chat Completions and Responses APIs. If no service tier is specified, the request uses the standard tier.
| Request | Tier Used | Token Rate |
|---|---|---|
| Ultrafast with a supported model and region | Ultrafast | 6× standard |
| Ultrafast request that falls back | Tier that serves the request | Rate for that tier |
| Ultrafast pinned to an unsupported region | Standard | Standard tier rate |
| No service tier specified | Standard | Standard tier rate |
Ultrafast supports processing in the US and globally. Requests pinned to an unsupported region, including the EU, run at the standard tier instead.
Billing follows the tier that actually handles the request. A request served through Ultrafast costs six times the standard per-token rate. If it falls back to another tier, Vercel charges the rate for that tier rather than the Ultrafast rate.
The update does not require every AI Gateway request to use the faster service. Builders choose it per request, while existing requests without the setting continue on the standard tier.
Frequently asked questions
What is Vercel Ultrafast mode?
Ultrafast mode is an OpenAI service tier available through AI Gateway. It provides faster model output for interactive applications and rapid coding iterations.
Which models support Ultrafast mode?
GPT 6 Astra and GPT 6.1 Sol are currently supported. Their AI Gateway model identifiers are openai/gpt-6-astra and openai/gpt-6.1-sol.
How much does Ultrafast mode cost?
Requests served through Ultrafast are billed at six times the standard per-token rate. Requests that fall back are billed at the rate for the tier that actually serves them.
Sources
3 checkedHow we cover tool news: Create With's tool desk drafts these reports with AI from the sources listed above and checks them against those sources before publishing.
ElevenLabs
Granola
Notion