OpenAI opens Ultrafast mode to every GPT-6.1 Sol API user
A speed-focused service tier that was a limited preview since August is now available to all API users, running up to 8x faster at 6x the standard price.
OpenAI rolled Ultrafast mode out of limited preview and on to every API user for GPT-6.1 Sol on 8 October. Set service_tier: "ultrafast" on a request to gpt-6.1-sol or gpt-6-astra and it runs on dedicated fast inference instead of the standard queue, subject to rate limits, with global processing and both US and EU data residency.
The trade is price for speed. GPT-6.1 Sol on Ultrafast generates at up to 8x the speed of its standard tier, around 300 tokens a second, but costs 6x more per token: $12 per million input tokens and $60 per million output tokens, against $2 and $10 on standard. GPT-6 Astra gets the same 6x multiplier on its own, higher base price.
Why it matters: this gives you a per-request lever for latency rather than a model swap. Agentic tool loops and interactive coding assistants that are bottlenecked on turnaround time, not cost, can now pay for speed only on the calls that need it, while everything else stays on the standard tier.
The caveat: GPT-5.6 Sol’s Ultrafast tier stays in preview access only — this general availability change is specific to GPT-6.1 Sol and GPT-6 Astra.
{ "model": "gpt-6.1-sol", "service_tier": "ultrafast" }