- On October 8, 2026, OpenAI added an Ultrafast mode for GPT-6.1 Sol to its API. It costs 6x the standard rate: $12 input and $60 output per million tokens, against $2/$10 at standard speed.
- GPT-6 reached ChatGPT's Chat tab on October 7 (GPT-6 Sol on paid plans) and October 8 (GPT-6 Luna on Free and Go). GPT-6.1 Sol is still limited to Work, Codex and the API.
- In ChatGPT, Ultrafast is only on Pro 500 and Enterprise, and it uses your included allowance 8x faster than standard speed, according to OpenAI.
Contents
On October 8, 2026, OpenAI made GPT-6.1 Sol Ultrafast available in its API, nine days after it announced the speed tier at DevDay. OpenAI's pricing page lists Ultrafast at six times GPT-6.1 Sol's standard rates. In the same stretch, GPT-6 replaced GPT-5.6 in ChatGPT's everyday Chat tab, free users included, along with a new "Intelligent UI." This story covers what changed between DevDay on September 29 and October 12, 2026. It is based on OpenAI's API changelog, pricing and model pages, and ChatGPT release notes, checked on October 12.
For the full GPT-6 launch cycle, see our GPT-6 launch coverage. For the rest of DevDay, see The Week in AI #2.
The short version
- GPT-6.1 Sol (
gpt-6.1-sol) costs $2 input, $0.10 cached input and $10 output per million tokens. That is one-fifth of GPT-6 Astra's $10/$50. It is not cheaper than GPT-6 Sol, which has the same $2/$10 headline price. - Ultrafast reached GPT-6 Astra on September 29 and GPT-6.1 Sol on October 8, at 6x the standard price.
- Chat tab: GPT-6 Sol on Plus, Pro, Business and Enterprise from October 7; GPT-6 Luna on Free and Go from October 8.
- GPT-6.1 Sol still isn't in Chat. It runs in ChatGPT Work, Codex and the API.
- GPT-5.5 retires from ChatGPT, Work and Codex on October 14, 2026, per OpenAI's Codex pricing page.
How much does GPT-6.1 Sol cost?
OpenAI calls GPT-6.1 Sol "an upgrade to GPT‑6 Sol that nearly matches GPT‑6 Astra's intelligence" on agentic coding, computer use and professional work. That claim rests on OpenAI's own evaluations, so treat it as vendor-reported. OpenAI still recommends Astra for the hardest science tasks.
The "one-fifth" in OpenAI's launch post is measured against Astra, a point some DevDay summaries blurred. Compared with GPT-6 Sol, only cached input changes, from $0.20 to $0.10.
API prices per million tokens, prompts up to 272K input tokens
| Model | Tier | Input | Cached input | Cache writes | Output |
|---|---|---|---|---|---|
| GPT-6.1 Sol | Standard | $2.00 | $0.10 | $2.50 | $10.00 |
| GPT-6.1 Sol | Batch / Flex | $1.00 | $0.05 | $1.25 | $5.00 |
| GPT-6.1 Sol | Fast | $4.00 | $0.20 | $5.00 | $20.00 |
| GPT-6.1 Sol | Ultrafast | $12.00 | $0.60 | $15.00 | $60.00 |
| GPT-6 Sol | Standard | $2.00 | $0.20 | $2.50 | $10.00 |
| GPT-6 Astra | Standard | $10.00 | $1.00 | $12.50 | $50.00 |
| GPT-6 Astra | Ultrafast | $60.00 | $6.00 | $75.00 | $300.00 |
| GPT-6 Luna | Standard | $0.10 | $0.01 | $0.125 | $0.50 |
Last verified: October 12, 2026 (OpenAI API pricing and model pages)
Prompts over 272,000 input tokens cost 2x the input and cache rates and 1.5x the output rate for the whole request. Regional data-residency endpoints add 10%.
If you're on gpt-6-sol: OpenAI's main pricing table now shows GPT-6.1 Sol in the Sol slot. GPT-6 Sol's model page still lists its prices and points to GPT-6.1 Sol as "the newer Sol model." The changelog shows no deprecation for GPT-6 Sol as of October 12.
What is Ultrafast?
OpenAI's developer guide calls Ultrafast its "fastest API service tier for latency-sensitive work." It sits above Fast mode, the tier formerly called Priority processing. At DevDay, OpenAI claimed "up to 8× faster token generation (300 tokens per second) in Codex" and up to 6x in the API. The API guide itself gives no latency figures.
- How: set
service_tiertoultrafastwithgpt-6-astraorgpt-6.1-solin the Responses API. OpenAI recommends WebSockets. - Who: "all API users," with separate rate limits. GPT-6.1 Sol's defaults run from 1 million tokens per minute (Build tier) to 40 million (Grow). On October 6, OpenAI cut its usage tiers from five to three: Build, Launch and Grow.
- Data residency: US and EU for GPT-6.1 Sol. US-only or global processing for Astra.
- Luna: no Ultrafast tier.
What changed in ChatGPT?
GPT-6 in Chat. OpenAI's October 7 post, "GPT-6 and Intelligent UI for everyone," moved Chat to GPT-6 Sol on paid plans that day and to GPT-6 Luna on Free and Go on October 8. The rollout is global, and Enterprise availability depends on admin settings. OpenAI says GPT-6 Instant "starts answering 44% sooner, on average, than GPT‑5.6 Instant" on questions that need web search. That is a vendor-reported figure.
Intelligent UI. Answers can now include charts, buttons, forms and small tools, such as a bill splitter. The release notes say it works from Instant to Extra High reasoning, but not with the Pro option, which runs on GPT-6 Astra.
Work and Codex didn't change. In OpenAI's words, "the models powering Work and Codex are not changing." Chat runs GPT-6 Sol, while Work and Codex offer GPT-6.1 Sol, GPT-6 Sol, Luna and Astra.
Ultrafast in ChatGPT is limited to Pro 500 and Enterprise. OpenAI's Pro tiers article says that among Pro plans, "Ultrafast is available only on Pro 500," and that buying credits on Pro 100 or 200 doesn't unlock it. Ultrafast uses included allowance at 8x the standard rate, and purchased credits at 6x. See our ChatGPT plans comparison.
Also since DevDay, per the release notes: audio uploads on paid plans (October 6), and composer predictions in Codex for personal Pro users (October 9, beta).
One caution: as of October 12, OpenAI's "GPT-6 and other models in ChatGPT" help article still names GPT-5.6 Luna as the Free and Go default, which contradicts the October 7 announcement.
What's confirmed, reported and uncertain
- Confirmed (OpenAI): the GPT-6.1 Sol and Ultrafast prices; Ultrafast launch dates of September 29 for Astra and October 8 for GPT-6.1 Sol; the October 7–8 Chat rollout; Ultrafast limited to Pro 500 and Enterprise in ChatGPT; GPT-5.5's retirement on October 14.
- Vendor-reported: "nearly matches Astra," "44% sooner" and 300 tokens per second.
- Uncertain:
- when GPT-6.1 Sol reaches Chat
- whether GPT-6 Sol gets a formal API deprecation
- per-plan message caps in Chat, which OpenAI doesn't publish
What it means for you
- API developers: moving from
gpt-6-soltogpt-6.1-solcosts the same and halves cached input. Ultrafast is expensive: GPT-6.1 Sol Ultrafast output ($60 per million tokens) costs more than standard Astra ($50). See token pricing isn't the story and model routing explained. - ChatGPT users: Chat updated automatically. For GPT-6.1 Sol, use Work or Codex.
- Codex users on GPT-5.5: switch models before October 14.
About this storyBased on the sources linked below. Editorial standards
