What is GPT-6.2 Astra?

GPT-6.2 Astra is OpenAI's paid frontier model for hard professional projects, available in ChatGPT Work, Codex, and the API as gpt-6.2-astra. It accepts text and images, responds with text, and has a 1,050,000-token context window. Access is listed for Plus, Pro, Enterprise, and Edu users, not a free ChatGPT plan.

What GPT-6.2 Astra is for

OpenAI positions GPT-6.2 Astra as the model to use for your hardest projects, from start to finish. The official announcement says it writes cleaner code, builds better frontends, especially for websites, gaming, and 3D design, and produces slides, spreadsheets, and documents that more closely follow the templates you provide and the conventions of fields such as law, sales, finance, and accounting.

The official API model page says GPT-6.2 Astra can build and revise software, use a computer, and work with documents. It accepts text and images and responds with text. In the Responses API, OpenAI documents a Multi-agent beta for delegating work to subagents. The API rejects requests that set reasoning effort to none.

OpenAI reports new highs on Agents' Last Exam, Terminal-Bench-Science 0.1, and the Vals Science Index, and it calls GPT-6.2 Astra the best model in the world for scientists. Those scores and the 'best for scientists' line are the company's claims, not independent measurements in this profile. A same-day Astra for Law configuration is mentioned in the announcement as a setup of GPT-6.2 Astra for lawyers; it is a related configuration, not a separate model covered in full here.

A first path to try it

If you already have ChatGPT Plus, Pro, Enterprise, or Edu, the official starting point is ChatGPT Work or Codex, where OpenAI says both GPT-6.2 Astra and Sol became available on 17 November 2026. Enterprise and Edu users should check that a workspace admin has enabled access. Pro 500 subscribers can use Ultrafast with Astra; Pro 200 Ultrafast is described as starting with GPT-6.2 Sol, not Astra.

Developers can call gpt-6.2-astra through the OpenAI API. The official model page lists Chat Completions, Responses, and Batch as supported, and it points to a Playground control on that page. OpenAI recommends keeping prompt prefixes unchanged and appending new instructions, so cache reuse can lower cost and latency and preserve reasoning across turns.

This is a suggested official path, not a recorded test. After a first task, check whether the output followed your template, whether token use and the billed tier (Standard, Batch, Fast, or Ultrafast) match your budget, and whether any tool permissions you enabled, such as computer use or web search, were actually needed.

Pricing, access, and limits

ChatGPT access is limited, in the official announcement, to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex. There is no official free-plan listing in that post. API Standard short-context rates for GPT-6.2 Astra, confirmed on both the announcement and the live model and pricing pages on 18 November 2026, are $8 input, $0.40 cached input, and $40 output per million tokens. Cache writes are $10 per million tokens for 30-minute retention and $20 for 12-hour writes on short-context Standard requests.

Longer prompts cost more on the official model page: for requests with more than 272K input tokens, input and cache rates double and output costs 1.5 times the short-context rate for the full request. Batch costs 50% of Standard. Fast mode costs twice Standard for the same context length. Ultrafast costs three times Standard through 28 February 2027, and six times Standard starting 1 March 2027, Pacific time. The context window is 1,050,000 tokens, with a 922,000-token input limit and a 128,000-token output limit. The knowledge cutoff is 18 May 2026.

OpenAI also publishes a GPT-6.2 system card. It says GPT-6.2 Astra is the first externally released model to meet the High threshold for AI Self-improvement, and that both GPT-6.2 models meet Critical cybersecurity and High biological and chemical thresholds. Fine-tuning is listed as not supported. Confirm current rates and plan eligibility on OpenAI's pages before you rely on them for budgeting.

What stands out

  • Official model ID gpt-6.2-astra, with a 1,050,000-token context window, 922,000 max input tokens, and 128,000 max output tokens
  • Official input modalities of text and image, text output, and a May 18, 2026 knowledge cutoff
  • Official ChatGPT Work and Codex access for Plus, Pro, Enterprise, and Edu users, with Ultrafast for Pro 500 and for both models in the API at launch
  • Official Standard API prices of $8 input, $0.40 cached input, and $40 output per million tokens, 20% below GPT-6 Astra on input and output
  • Official support in the Responses API for tools including computer use, code interpreter, web search, file search, and MCP

Before you choose

  • The official announcement does not list a free ChatGPT plan; Enterprise and Edu access depends on a workspace admin enabling it
  • Ultrafast and long-context requests are billed above the Standard short-context rates on the official model page; confirm those multipliers before a large run
  • OpenAI says GPT-6.2 Astra meets High AI Self-improvement and that, in two tests, the GPT-6.2 models were more likely to omit obstacles that blocked their work
  • This profile is not a hands-on test of ChatGPT Work, Codex, or the API, and it does not treat unofficial reports of a GPT-6.2 Luna model as fact

Plans & pricing

Current pricing on developers.openai.com (opens in a new tab)

Sources & further reading

Based on the official documentation below. Features and access can change; check the provider before making a decision.

  1. OpenAI: Introducing GPT-6.2 Astra and Sol (17 November 2026), official page text retrieved 18 November 2026 (opens in a new tab)
  2. Official OpenAI API model page for GPT-6.2 Astra, rendered 18 November 2026 (opens in a new tab)
  3. Official OpenAI API pricing page, rendered 18 November 2026 (opens in a new tab)
  4. Official GPT-6.2 System Card safety overview, dated 17 November 2026 (opens in a new tab)
Something out of date or incorrect?Report a correction

Was this page useful?

Help us make our coverage more useful. This is feedback on the page, not a rating of the tool.

Share this page

Related reading