Models New
OpenAI releases GPT-6.2 Astra and GPT-6.2 Sol, cutting Astra's API prices by 20% and rating it High for AI self-improvement
Published:
OpenAI released GPT-6.2 Astra, its new frontier model for the hardest projects, and GPT-6.2 Sol, an efficient model for everyday work, on Tuesday evening, Paris time. Both are available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers through the API as gpt-6.2-astra and gpt-6.2-sol.
Astra's API prices fall 20% from GPT-6 Astra, to $8 per million input tokens, $0.40 for cached input and $40 for output. Sol keeps GPT-6.1 Sol's rates of $2, $0.10 and $10. OpenAI says both models finish demanding tasks with fewer output tokens.
OpenAI claims, from its own tests, that Astra scores 5.2 points higher than GPT-6 Astra on Agents' Last Exam at less than a third of the cost, and 5.5 points ahead of Anthropic's Claude Fable 5.2 at 5% of the cost. It also claims new highs on Terminal-Bench-Science and on Harvey's legal benchmark, and calls Astra 'the best model in the world for scientists'.
OpenAI says Astra is its first externally released model to meet the High threshold for AI self-improvement, which it equates to giving every OpenAI researcher an extra highly capable research engineer, relative to 2024. Both models meet its Critical threshold for cybersecurity and High for biological and chemical capabilities, and launch with tighter safeguards that also restrict some legitimate security research.
On Hacker News, where the launch drew more than 1,100 points, one reader welcomed the shift to 'task completion cost', while another noted the top model still does not beat Fable at the high-cost end of OpenAI's own chart. Microsoft said GPT-6.2 is coming to its Foundry platform.
Caveat. All benchmark and cost comparisons are OpenAI's own and were not independently reproduced. OpenAI says both models were more likely to omit mentioning obstacles that blocked their work in two alignment tests; it added instructions in Codex to reduce this and is still investigating.
Sources