OpenAI added Ultrafast processing for GPT-6.1 Sol to its Responses API on 8 October. The service tier is available to all API users, subject to rate limits, and supports US and EU data residency. The change targets the time between generated output tokens; it is a processing option for the existing model.
Based on 5 primary sources · See sources
The speed tier carries a sixfold base tariff
The current pricing table lists $12 per million input tokens and $60 per million output tokens for Sol Ultrafast. Standard costs $2 and $10 respectively for the same short-context tier. Tokens are the pieces of content counted for API billing.
Sol’s base text-token rates
USD per million tokens · prompts up to 272,000 input tokens
| Charge | Standard | Ultrafast |
|---|---|---|
| Input | $2 | $12 |
| Output | $10 | $60 |
A faster stream is only part of the wait
The Ultrafast guide recommends persistent WebSocket connections for applications making repeated tool calls: opening a fresh connection each time can erode the latency benefit. Ultrafast also has its own rate limits, separate from Standard and Fast.
OpenAI’s announcement pitches the speed gain, but we have not verified an independent Sol Ultrafast timing test with inspectable settings and outputs. The distinction between token speed and a finished answer is visible in our Haiku effort tests. Measure the complete job before treating a streaming-speed claim as a productivity result.
OpenAI announces Sol Ultrafast and its vendor-reported speed claim.
View the original post on X ↗Sources
- 8 October API rollout entry
OpenAI · Announcement ·
- Ultrafast availability, networking and regional support
OpenAI · Documentation
- Current processing-tier prices
OpenAI · Documentation
- Sol model prices and long-prompt modifiers
OpenAI · Documentation
- Original Ultrafast announcement
OpenAI Developers · Announcement ·
Written with AI assistance from the sources above. Analysis and practical implications are our interpretation. Our editorial approach.
