OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Tuesday, 17 November, in ChatGPT, Codex and its API, according to its release post and a post on X at 17:54 UTC. OpenAI said GPT-6.2 Astra’s API token prices are 20% lower than GPT-6 Astra’s: “$8 for input, $0.40 for cached input, and $40 for output, all per million tokens”.
Our number: $8 for input and $40 for output is the same list price Anthropic’s pricing page gives for Claude Fable 5.2, twice Claude Opus 5.5’s $4 and $20, and four times Gemini 4 Carbon’s $2 and $10 through 30 November, by our arithmetic. GPT-6.2 Sol keeps $2 and $10, the same as Claude Sonnet 5.5.
Business Pill · Usage-Based Pricing
An evergreen Business Engineer explainer on usage-based pricing, paying for work rather than seats. It teaches the general idea only and does not discuss OpenAI or GPT-6.2.
The key insight: OpenAI has moved its top model onto Anthropic’s price line: GPT-6.2 Astra lists at $8 and $40 per million input and output tokens, the same as Claude Fable 5.2, by our arithmetic, so the contest OpenAI is picking is cost per finished task, where it claims Astra beat Fable 5.2 on Agents’ Last Exam at 5% of the cost.
What OpenAI Released
OpenAI calls GPT-6.2 Astra “our new frontier model for your hardest projects” and GPT-6.2 Sol “our efficient, versatile model for everyday work”. It says both “deliver better results on demanding tasks with fewer output tokens”.
Both models have a 1,050,000-token context window and 128,000 max output tokens, according to OpenAI’s Astra and Sol model pages. Astra’s knowledge cutoff is 18 May 2026 and Sol’s is 1 June 2026.
OpenAI wrote on X that the models are available “in ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu users”. Developers can call “gpt-6.2-astra” and “gpt-6.2-sol” in the API, with Ultrafast available for both. The name first surfaced in Codex usage screenshots on X on 1 November.

The Price Against GPT-6 Astra, Claude and Gemini
OpenAI’s pricing page lists GPT-6 Astra at $10 for input, $1.00 for cached input and $50 for output per million tokens. GPT-6.2 Astra is $2 and $10 lower on input and output, 20% in each case, and its cached input is 60% lower, by our arithmetic and as OpenAI states.
Anthropic lists Claude Fable 5.2 at $8 and $40 and Claude Opus 5.5 at $4 and $20 per million tokens. Google’s Gemini pricing page lists Gemini 4 Carbon at $2 and $10 through 30 November and $4 and $20 from 1 December, so Astra’s gap to it narrows from four times to two times, by our arithmetic.
These are per-token list prices, not cost per task. Anthropic’s page says its newer tokenizer “produces approximately 30% more tokens for the same text”, and OpenAI says GPT-6.2 Astra uses fewer output tokens, so the same job can cost different amounts at the same list price.
What OpenAI Claims on Benchmarks
OpenAI says that on Agents’ Last Exam GPT-6.2 Astra scored “5.2 points higher than GPT-6 Astra at less than a third of the cost, and 5.5 points ahead of Claude Fable 5.2 at just 5% of the cost”. Its footnote says Fable 5.2’s leaderboard results “exclude nine tasks that used a fallback model”.
On Artificial Analysis’s AA-3D, OpenAI says, “GPT-6.2 Astra ranks first with 1,318 Elo, followed by Claude Fable 5.2 at 1,271 and GPT-6.2 Sol at 1,259”. On its internal GDPval v2, it says Astra is “up 6.2 percentage points from GPT-6 Astra while using 22% fewer output tokens”.
OpenAI says GPT-6.2 Astra is “our first externally released model to meet the High threshold for AI Self-improvement”, and that both models meet the Critical threshold for cybersecurity. In two alignment tests, it says, the models “were more likely to omit mentioning obstacles that blocked their work”. OpenAI’s earlier GPT-6 science work used the previous Astra.
The Timing
The release came the day before 18 November, the date Anthropic’s IPO website gives for pricing its offering, as we reported on Anthropic’s IPO range. OpenAI’s release post does not mention Anthropic’s listing. The Information reported on 4 November that new OpenAI models could come as soon as that Friday.
The business lever is API revenue per developer: OpenAI is now selling its top model at the price Anthropic charges for Fable 5.2, and the comparison buyers make shifts to cost per finished task. Tools such as Perplexity’s Multiagent already route one task across models from both labs.
The Structural Read
As we read the price list, matching Fable 5.2 token for token takes list price off the table as a reason to pick either model, and leaves token efficiency, which OpenAI stresses throughout its release post, as the variable that sets the bill.
As we read the Sol decision, keeping GPT-6.2 Sol at $2 and $10, the same as Claude Sonnet 5.5 and Gemini 4 Carbon until 30 November, puts three labs’ everyday models at one list price, by our arithmetic, with Google’s set to double on 1 December.
OpenAI, 17 November 2026
“For businesses and developers, that means frontier performance at costs that make sense at scale.”
Three Implications
FOR API BUYERS GPT-6.2 Astra is $8 for input, $0.40 for cached input and $40 for output per million tokens, down from $10, $1.00 and $50 for GPT-6 Astra on OpenAI’s pricing page; requests above 272K input tokens pay more, its model page says.
FOR ANTHROPIC Claude Fable 5.2 lists at the same $8 and $40, and OpenAI’s release names Fable 5.2 in three of its benchmark comparisons, by our count; Anthropic had not commented in the sources we read.
FOR CHATGPT AND CODEX USERS OpenAI says both models are available to Plus, Pro, Enterprise and Edu users, with Ultrafast for Pro 500 subscribers and, starting with GPT-6.2 Sol, for Pro 200 subscribers.
The Business Engineer Lens
This story maps onto the Business Engineer framework Tokenomics: The Economics of AI.
The framework says of the token: “It is simultaneously the unit of cognition the model produces, the unit of compute the data center serves, the unit of price the lab charges, and the unit of value the enterprise extracts.”
As we read it, GPT-6.2 Astra splits two of those roles: the unit of price the lab charges is now the same as Fable 5.2’s, so OpenAI is competing on how many tokens a task takes. The framework predates this release and does not discuss GPT-6.2.
What Is Not Established
The benchmark results and per-task cost comparisons are OpenAI’s; we did not find independent results for GPT-6.2 Astra or Sol in the sources we read. OpenAI’s pricing page did not yet list GPT-6.2 rows when we read it at about 18:08 UTC; the prices come from its release post and model pages.
We did not read the system card in full. Anthropic and Google had not commented on the release in the sources we read. We did not contact OpenAI.
The Bottom Line
OpenAI released GPT-6.2 Astra at $8 and $40 per million input and output tokens, 20% below GPT-6 Astra and the same list price as Claude Fable 5.2, a day before Anthropic’s IPO pricing date. GPT-6.2 Sol stays at $2 and $10. OpenAI’s claim rests on cost per task, which no independent test in the sources we read has checked yet.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. Model details, availability, prices and benchmark results are from OpenAI’s release post of 17 November 2026, its API model pages and its posts on X; the benchmark and per-task cost claims are OpenAI’s own. GPT-6 Astra prices are from OpenAI’s API pricing page, Claude prices from Anthropic’s pricing page and Gemini prices from Google’s pricing page. The price comparisons are our arithmetic. We did not contact OpenAI.
Sources: OpenAI, Introducing GPT-6.2 Astra and Sol, 17 Nov 2026 · OpenAI API docs, GPT-6.2 Astra · OpenAI API docs, GPT-6.2 Sol · OpenAI API pricing · Anthropic Claude API pricing · Google Gemini API pricing · OpenAI on X, 17 Nov 2026









