OpenAI Adds Ultrafast Mode for GPT-6.1 Sol in the API, Listing It at $12 Input and $60 Output per Million Tokens, Six Times Standard Rates
OpenAI's changelog says GPT-6.1 Sol gained an Ultrafast service tier on Oct 8; its pricing page lists $12 input and $60 output per million tokens, six times Standard.
Overview
OpenAI added an Ultrafast service tier for GPT-6.1 Sol in the Responses API on October 8, according to the OpenAI API changelog. The changelog says developers use gpt-6.1-sol with service_tier: "ultrafast" “to reduce the time between generated output tokens,” and that the tier is “available to all API users, subject to rate limits.” OpenAI’s pricing page, read on October 14, lists Ultrafast Sol at $12.00 per million input tokens and $60.00 per million output tokens for short-context prompts. Those are six times the Standard rates listed for the same model, a ratio that follows from the page’s own figures.
All three sources are OpenAI documentation pages, so every claim below is OpenAI’s own account of its product. No independent speed measurement was available.
What We Know
Availability and setup
- The Ultrafast mode guide describes the tier this way: “Ultrafast mode is the fastest service tier in the OpenAI API. It is broadly available for GPT-6 Astra and GPT-6.1 Sol. Use it when speed justifies the higher cost.”
- Per the guide, developers set
modeltogpt-6-astraorgpt-6.1-solandservice_tiertoultrafastin eachresponse.createevent. - The guide says OpenAI “strongly” recommends WebSockets, “especially for agentic applications that make many tool calls in quick succession,” and adds that Ultrafast also supports HTTP requests through the SDK.
- The changelog entry for October 8 says the Sol tier comes with global processing and US and EU data residency. The guide says GPT-6 Astra Ultrafast supports US data residency and global processing only.
- The guide says Ultrafast “has separate rate limits from Standard and Fast modes.”
Listed prices
The pricing page lists prices per million tokens in four columns: input, cached input, cache writes and output. It splits each tier into short-context and long-context prices, and the page content read for this article does not define where the long-context threshold falls.
| GPT-6.1 Sol tier | Input | Cached input | Cache writes | Output |
|---|---|---|---|---|
| Standard, short context | $2.00 | $0.10 | $2.50 | $10.00 |
| Fast, short context | $4.00 | $0.20 | $5.00 | $20.00 |
| Ultrafast, short context | $12.00 | $0.60 | $15.00 | $60.00 |
| Ultrafast, long context | $24.00 | $1.20 | $30.00 | $90.00 |
The same page lists GPT-6 Astra Ultrafast at $60.00 input and $300.00 output per million tokens for short-context prompts, against $10.00 and $50.00 for Astra Standard. Each Ultrafast figure for both models works out to six times the corresponding Standard figure, and each Fast figure to twice the Standard figure, in both the short-context and long-context columns. That multiple is arithmetic on the listed prices, not a ratio OpenAI states.
The pricing page also says regional processing (data residency) endpoints “are charged a 10% uplift for models released on or after March 5, 2026.” The page content read for this article does not say how that uplift applies to Ultrafast requests on the Sol tier’s EU and US residency options, so the effective regional price is not established here.
Separately, the changelog’s September 29 entry for the Sol launch gives the Standard price “for prompts with up to 272K input tokens” as $2 input, $0.10 cached input, $2.50 cache write and $10 output. The changelog does not attach a 272K figure to the Ultrafast tier, so this article does not either.
Rate limits
The guide lists default Ultrafast token-per-minute limits by the API usage tiers that OpenAI simplified from five to three on October 6, according to the changelog.
| API usage tier | GPT-6.1 Sol | GPT-6 Astra |
|---|---|---|
| Build | 1,000,000 | 500,000 |
| Launch | 4,000,000 | 1,000,000 |
| Grow | 40,000,000 | 5,000,000 |
The guide adds that organizations working with an OpenAI account team can contact them to request higher limits.
Timeline
Per the changelog, the Astra version came first: a September 29 entry says Ultrafast mode for GPT-6 Astra was added to the Responses API, “available to API customers, subject to rate limits,” with global processing and US data residency, and that EU and other regional inference residency “aren’t supported.” The Sol entry is dated October 8, nine days later. Earlier, an August 13 changelog entry announced Ultrafast mode as a service tier for GPT-5.6 Sol that “runs up to 14x faster than Standard processing,” in limited preview for select customers.
The Machine Herald previously reported on Ultrafast in the Codex Pro 500 plan, and noted that API pricing for the tier was not stated in the sources reviewed at the time. The pricing page now carries those figures.
What We Don’t Know
- How much faster it is. The Ultrafast guide, as read for this article, gives no speed multiple for GPT-6.1 Sol. The only multiple in the changelog, 14x, belongs to the August 13 entry about GPT-5.6 Sol in limited preview, a different model and stage. Whether the current Sol tier delivers a comparable gain is not stated in these pages.
- Latency in practice. The changelog describes the aim as reducing “the time between generated output tokens.” OpenAI’s pages do not publish time-to-first-token or throughput numbers for the Sol tier.
- Whether the price is worth it. The guide says to use the tier “when speed justifies the higher cost.” Without a published speedup, a per-task cost comparison cannot be made from these sources.
- Regional pricing. How the 10% regional uplift interacts with Ultrafast pricing is not spelled out in the pages read.
- Status after October 14. Prices, limits and availability are as the three pages showed them on October 14, 2026 and may change.