OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on 17 November for ChatGPT Plus, Pro, Enterprise and Edu users, in ChatGPT Work and Codex, and for developers through its API. Astra is aimed at hard professional work and Sol at everyday tasks. Astra’s API price falls 20% to $8 per million input tokens and $40 per million output tokens; Sol stays at $2 and $10.
The safety paperwork matters as much as the launch. OpenAI’s system card says Astra is “the first model we are deploying externally” that meets its High threshold for AI self-improvement, a category for models that could speed up AI research itself. Both models meet its Critical threshold for cybersecurity, as GPT-6 Astra did before them, so OpenAI has tightened default limits on higher-risk hacking help; only separately vetted organisations get looser access.
Britain’s AI Security Institute found a “substantial regression in robustness” against jailbreaks compared with GPT-6 Astra. OpenAI says it has dealt with this, but the institute had not checked the fix when the card was published.
The benchmark gains OpenAI reports are its own. Independent testers Artificial Analysis and Epoch AI place Astra close to, but not ahead of, Anthropic’s Claude Fable 5.2.
Read the system card.

