Overview

GPT-6.2 Astra is OpenAI's frontier model for the hardest projects, released on November 17, 2026 together with GPT-6.2 Sol. It is the first Astra upgrade since GPT-6 Astra in September 2026; the planned GPT-6.1 Astra was cancelled. OpenAI says the model gets better results on demanding tasks while using fewer output tokens, and it is the company's recommended model for long professional work, scientific workflows, coding in existing codebases and computer use in tools such as CAD and simulation software.
OpenAI reports that GPT-6.2 Astra sets a new high on Agents' Last Exam, 5.2 points above GPT-6 Astra at less than a third of the cost and 5.5 points above Claude Fable 5.2. It scores 9 points higher than GPT-6 Astra on Terminal-Bench-Science 0.1, sets a new high on the Vals Science Index and on Harvey LAB-AA v1.1, and leads OpenAI's internal GDPval v2, up 6.2 points from GPT-6 Astra with 22% fewer output tokens. On Artificial Analysis's AA-AnalystAgent it answers 62.5% of questions correctly on all five attempts, and on AA-3D it ranks first with 1,318 Elo.
In the API, gpt-6.2-astra accepts text and image input and returns text, with a 1,050,000-token context window (922,000 input tokens and up to 128,000 output tokens) and a knowledge cutoff of May 18, 2026. It is served on the Chat Completions, Responses and Batch endpoints. OpenAI rates it Critical for cybersecurity and High for biological and chemical capability under its Preparedness Framework, and it is the first externally released OpenAI model to meet the High threshold for AI self-improvement, so it launched with tighter cyber safeguards.
| Released | 2026-11-17 |
|---|---|
| License | Proprietary |
| Weights | API only |
| Context | 1.05M |
| Max output | 128K |
| Knowledge cutoff | 2026-05-18 |
| Modalities | Text, Vision |
| Status | Available |
Pricing
| Input | $8.00 / 1M tokens |
|---|---|
| Cached input | $0.40 / 1M tokens |
| Output | $40.00 / 1M tokens |
20% below GPT-6 Astra's input and output rates, with a 60% discount on cached input. Requests over 272K input tokens cost 2x for input and 1.5x for output; Batch is half price.
Strengths
- New highs on Agents' Last Exam, Terminal-Bench-Science 0.1, the Vals Science Index and Harvey LAB-AA v1.1, per OpenAI
- Input and output tokens 20% cheaper than GPT-6 Astra, with cached input at $0.40 per 1M tokens
- Fewer output tokens per task: 22% fewer than GPT-6 Astra on GDPval v2
- 1.05M-token context window with up to 128K output tokens

- More precise computer use in CAD and simulation tools, and slides that follow a template's layout
Best for
- Reach for it for long, multistep professional work such as financial models, decks and legal deliverables.
- Reach for it for scientific workflows run in a terminal, from data analysis to simulation.
- Reach for it for coding agents that must change and test large existing codebases.
- Reach for it for computer-use agents that operate CAD, simulation or office tools.
How to access
| Provider | Model ID |
|---|---|
| OpenAI API ↗ | gpt-6.2-astra |
GPT (Flagship / Thinking) — every version
The full lineage of the GPT (Flagship / Thinking) line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| GPT-6.2 Astracurrent | 2026-11-17 | 1.05M | Proprietary |
| GPT-6.2 Sol | 2026-11-17 | 1.05M | Proprietary |
| GPT-6.1 Terra | 2026-10-26 | 1.05M | Proprietary |
| GPT-6.1 Sol | 2026-09-29 | 1.05M | Proprietary |
| GPT-6 Sol | 2026-09-22 | 1.05M | Proprietary |
| GPT-6 Luna | 2026-09-22 | 1.05M | Proprietary |
| GPT-6 Astra | 2026-09-03 | — | Proprietary |
| GPT-5.6 Sol | 2026-07-09 | 1.05M | Proprietary |
| GPT-5.6 Terra | 2026-07-09 | 1.05M | Proprietary |
| GPT-5.6 Luna | 2026-07-09 | 1.05M | Proprietary |
| GPT-5.5 | 2026-04-23 | 1.05M | Proprietary |
| GPT-5.4 | 2026-03-05 | — | Proprietary |
| GPT-5.2 | 2025-12-11 | — | Proprietary |
| GPT-5.1 | 2025-11-12 | — | Proprietary |
| GPT-5 | 2025-08-07 | — | Proprietary |
| GPT-4o | 2024-05-13 | — | Proprietary |
FAQ
How much does GPT-6.2 Astra cost?
OpenAI lists GPT-6.2 Astra at $8.00 per 1M input tokens, $0.40 per 1M cached input tokens and $40.00 per 1M output tokens. The input and output rates are 20% lower than GPT-6 Astra's, and cached input gets a 60% discount. Requests above 272K input tokens cost more, and the Batch API halves the price.
How does GPT-6.2 Astra compare to GPT-6 Astra?
OpenAI reports that GPT-6.2 Astra beats GPT-6 Astra by 5.2 points on Agents' Last Exam, 9 points on Terminal-Bench-Science 0.1 and 6.2 points on GDPval v2, while using 22% fewer output tokens on GDPval v2. It also costs less per token, so OpenAI says many tasks cost under a third of what GPT-6 Astra charged.
Who can use GPT-6.2 Astra?
GPT-6.2 Astra launched on November 17, 2026 for Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and in the API as gpt-6.2-astra. Pro 500 subscribers can run it in Ultrafast mode, which is also available for the model in the API at launch.
Is GPT-6.2 Astra safe for cybersecurity work?
OpenAI rates GPT-6.2 Astra Critical for cybersecurity under its Preparedness Framework, so it launched with tighter safeguards that also block some legitimate security research. Approved defenders use OpenAI's Daybreak program for reduced refusals, and OpenAI says it plans to bring GPT-6.2 models to Daybreak.
