BreakingNewsFor: IT and service desk · Business operations · Finance and accounting

OpenAI ships GPT-6.2 Astra and Sol; Astra's API rate falls 20%

GPT-6.2 Astra and Sol reached ChatGPT and the API on 17 November. Astra's API rate is 20% below GPT-6 Astra; the fastest mode costs three times more.

OpenAI released two models on Tuesday 17 November 2026: GPT-6.2 Astra, its top model, and GPT-6.2 Sol, the cheaper one. Its [developer changelog](https://developers.openai.com/api/docs/changelog) lists both that day. [9to5Mac reports](https://9to5mac.com/2026/11/17/openai-upgrades-chatgpt-and-codex-with-new-gpt-6-2-astra-and-sol-models/) that both are available from the same day in ChatGPT Work and Codex on Plus, Pro, Enterprise and Edu plans.

For a team that uses ChatGPT, the model behind the chat window changes and the seat price was not reported to change. For a team that pays by usage, directly or through a vendor's product built on OpenAI, the top model got cheaper.

What it does to an API bill

OpenAI bills API use per million tokens, the chunks of text a model reads (input) and writes (output). Its [pricing page](https://developers.openai.com/api/docs/pricing) lists GPT-6.2 Astra at $8 input and $40 output, against $10 and $50 for GPT-6 Astra: $2 of $10 and $10 of $50, so 20% lower on both. Text the model has already read in a session, which OpenAI bills as cached input, falls from $1.00 to $0.40, 60% lower. GPT-6.2 Sol lists at $2 and $10, the same as GPT-6.1 Sol and a quarter of Astra's rate.

  • Jobs that can wait cost less. Batch processing is half the standard rate. A background mode in public beta is 30% below standard if the answer can take up to two hours ($5.60 and $28 for Astra) and 40% below for eight hours ($4.80 and $24).
  • Speed costs more. Fast mode is twice the standard rate. Ultrafast is three times standard through 28 February 2027 ($24 and $120 for Astra) and six times from 1 March 2027 ($48 and $240).
  • Long documents cost more. Above 272,000 input tokens in one request, input rates double and output costs 1.5 times as much for the whole request, the [model page](https://developers.openai.com/api/docs/models/gpt-6.2-astra) says.
  • Location costs more. Keeping processing in the US or the EU (data residency) carries a 10% uplift, and Ultrafast offers US residency only, according to the pricing page and the [migration guide](https://developers.openai.com/api/docs/guides/latest-model?model=gpt-6.2-astra).

What changes in ChatGPT

OpenAI says both models finish demanding tasks with fewer output tokens, so a user gets more done before reaching a usage limit, 9to5Mac reports; it gives no figure. Ultrafast, which OpenAI says writes up to 4.5 times faster and uses more of the allowance, comes to the $200-a-month Pro plan on Sol, the $500-a-month Pro plan on both models, and eligible Enterprise and Edu workspaces once an admin enables it. The report does not mention the Business plan.

Answers may read differently. OpenAI's migration guide says GPT-6.2 Astra is "more concise and direct by default", and that when a request is unclear it "may propose a next step and wait for confirmation". Saved prompts and custom instructions that say "be concise" are worth revisiting, the guide says. Astra's knowledge stops at 18 May 2026 and [Sol's](https://developers.openai.com/api/docs/models/gpt-6.2-sol) at 1 June 2026.

What OpenAI has not said

The pages I read give no retirement date for GPT-6 Astra, no usage limits per ChatGPT plan, and no figure for how much allowance Ultrafast uses in ChatGPT. The lower rate reaches you only if your vendor switches models and passes it on; nothing here says a vendor will.

What to do this week

  • IT, or whoever administers ChatGPT: leave Ultrafast off in Enterprise and Edu workspaces until OpenAI tells you what it does to your allowance, then enable it for the teams where waiting costs money.
  • Finance and business operations: list the tools you pay for by usage that run on GPT-6 Astra, and ask each vendor in writing when it moves to GPT-6.2 Astra and whether its price to you falls.
  • Team leads: rerun your five most-used saved prompts on GPT-6.2 and check length and tone before staff find the difference in a customer email.
Was this useful for your work?

Take this to work

Forward this

OpenAI released GPT-6.2 Astra and Sol on 17 November 2026, and its pricing page lists Astra at $8 and $40 per million input and output tokens, 20% below GPT-6 Astra. We should ask every vendor that bills us by usage on OpenAI's top model when it switches and whether our price falls, and keep the Ultrafast setting off until we know what it uses.

Questions to ask

  1. Which of our tools run on GPT-6 Astra today, and when does each vendor move to GPT-6.2 Astra?
  2. Will the vendor pass on the 20% lower list rate, and from which invoice?
  3. How much of our ChatGPT allowance does Ultrafast use per task, and can an admin enable it for some groups only?
  4. Do any of our jobs send more than 272,000 input tokens in one request, where the input rate doubles?
  5. Which of our scheduled jobs could wait two or eight hours for a rate 30% or 40% lower?

The arithmetic

(input tokens in millions × $8) + (output tokens in millions × $40) for GPT-6.2 Astra at the standard rate, against the same token counts × $10 and × $50 on GPT-6 Astra.

One million input and one million output tokens cost $8 + $40 = $48 on GPT-6.2 Astra and $10 + $50 = $60 on GPT-6 Astra, $12 or 20% less. The same job on Ultrafast is $24 + $120 = $144 until 28 February 2027. OpenAI does not say how many tokens a ticket, a contract or a report uses; take that from your own usage export.

Questions this article answers

What does GPT-6.2 Astra cost in the API?

OpenAI's pricing page lists GPT-6.2 Astra at $8 per million input tokens and $40 per million output tokens at the standard rate, read on 17 November 2026. That is 20% below GPT-6 Astra's $10 and $50.

Which ChatGPT plans get GPT-6.2?

9to5Mac reports that GPT-6.2 Astra and Sol are available from 17 November 2026 in ChatGPT Work and Codex on Plus, Pro, Enterprise and Edu plans. The report does not mention the Business plan.

What is Ultrafast mode and what does it cost?

A faster processing tier. OpenAI says it writes up to 4.5 times faster, 9to5Mac reports. In the API it costs three times the standard rate through 28 February 2027 and six times from 1 March 2027, OpenAI's pricing page says; in ChatGPT it uses more of the plan's allowance, by an amount not stated.

Did GPT-6.2 Sol get cheaper?

No. OpenAI's pricing page lists GPT-6.2 Sol at $2 per million input tokens and $10 per million output tokens, the same as GPT-6.1 Sol.

Send this to someone who runs a teamLinkedInXBlueskyHNEmail