OpenAI’s developer changelog lists GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026. OpenAI describes Astra as its most capable model for the hardest projects and Sol as its go-to model for everyday work, from coding to agents at scale.
What to know
| GPT-6.2 Astra | GPT-6.2 Sol | |
|---|---|---|
| Model ID | gpt-6.2-astra | gpt-6.2-sol |
| Input / output (per 1M tokens) | $8 / $40 | $2 / $10 |
| Cached input | $0.40 | $0.10 |
| Context window | 1,050,000 tokens | 1,050,000 tokens |
| Max output | 128,000 tokens | 128,000 tokens |
| Knowledge cutoff | May 18, 2026 | June 1, 2026 |
All figures are from OpenAI’s model pages. Other points OpenAI lists:
- Inputs and outputs: text and image in, text out. No audio or video.
- APIs: Responses, Chat Completions and Batch. Tool calling needs the Responses API.
- Long prompts: over 272K input tokens, input and cache rates double and output costs 1.5x for the whole request. Batch is 50% of Standard; Fast mode is 2x; Ultrafast is 3x through February 28, 2027 and 6x from March 1, 2027.
- Reasoning effort: the API rejects
none. OpenAI’s migration guide suggests starting atmedium, and moving fromnoneorminimaltolow. - Removed parameters:
temperature,top_pand logprobs settings should be dropped when migrating, according to the guide.
Two behaviour changes for developers
- Edited history can lose reasoning. OpenAI says the Responses API now drops earlier reasoning items after some changes, such as editing earlier messages or tool calls mid-conversation or switching to a smaller model. That can lower quality and use more tokens, so keep the conversation prefix unchanged.
- Top-level
instructionsnow come afterinput. To keep reusable instructions cacheable, OpenAI recommends putting them in adevelopermessage as the first input item.
OpenAI also released discounted background completion windows the same day, which work with both models.
Why it matters
The new pair keeps the Astra and Sol split: Astra for the hardest jobs, Sol at a quarter of the price for most work. OpenAI’s own guide says GPT-6.2 is more concise and direct by default, so old “be brief” lines in your prompts may now be unnecessary. We collected practical prompt changes in how to prompt GPT-6.2.
Related: ChatGPT, ChatGPT vs Claude, GPT-6.1 Terra and the model tracker.
Source: OpenAI API changelog. We summarise in our own words; check the source for the full details.
Contribute your creativity