GPT-6.2 Astra and Sol: what OpenAI released, what it costs, and the cyber limits
Quick answer: On Nov 17, 2026, OpenAI released GPT-6.2 Astra (its model for the hardest professional work) and GPT-6.2 Sol (its cheaper everyday model). Both are in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, and in the API. Astra’s API price is $8 per million input tokens and $40 per million output tokens, 20% lower than GPT-6 Astra. Sol stays at $2 and $10. OpenAI says both models meet its “Critical” cybersecurity threshold, so they ship with tighter safeguards that also restrict some legitimate security work. (OpenAI)

What changed
- Astra: priced at $8 input, $0.40 cached input and $40 output per million tokens. OpenAI cites a 20% discount on input and output tokens and a 60% discount on cached input versus GPT-6 Astra. (OpenAI)
- Sol: keeps GPT-6.1 Sol’s rates of $2 input, $0.10 cached input and $10 output per million tokens. OpenAI says it is “smarter” and “more efficient” than 6.1 Sol and lets users complete more tasks on the same paid plan. (OpenAI)
- Access: all Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex; eligible Enterprise and Edu users once their admin enables it. API model names are
gpt-6.2-astraandgpt-6.2-sol. (OpenAI) - Ultrafast mode: Pro 500 subscribers can use it with either model. For the first time it comes to Pro 200 subscribers, starting with Sol. In the API it is available for both models at launch. (OpenAI)
OpenAI’s own benchmark claims
These are OpenAI’s results, not independent tests. Treat them as claims to check on your own workload.
- On Agents’ Last Exam, OpenAI says Astra scores 5.2 points higher than GPT-6 Astra at less than a third of the cost, and 5.5 points ahead of Claude Fable 5.2 at about 5% of the cost. OpenAI notes the Fable 5.2 results exclude nine tasks that used a fallback model. (OpenAI)
- On Cognition’s FrontierCode 2.0, OpenAI says Sol comes within 0.5 points of Astra at about one-fifth the cost. (OpenAI)
The cybersecurity limits
OpenAI says GPT-6.2 Astra and Sol “meet the Critical threshold for cybersecurity” under its Preparedness Framework, and launch “with tighter safeguards for higher-risk cybersecurity assistance.” OpenAI says these protections reduce misuse risk “but also restrict some legitimate vulnerability research and security testing workflows.” OpenAI says defenders can currently get access with reduced cyber refusals through its Daybreak program (on GPT-6 Sol, not the new models). Both models also meet the “High” threshold for biological and chemical capabilities, and Astra is OpenAI’s first externally released model to meet “High” for AI self-improvement. (OpenAI)
What is still unknown
Independent results. OpenAI’s numbers come from its own testing and partner benchmarks; we have not seen third-party replication. The system card has full methodology; this brief does not summarize it.
Why it matters
For teams on the API, Astra gets cheaper per token and Sol does not change, so the cost question is token efficiency on your own tasks. For security teams, expect more refusals on vulnerability work unless you have access through OpenAI’s defender programs, by OpenAI’s own account.
FAQ
What does GPT-6.2 Astra cost? $8 per million input tokens, $0.40 cached input, $40 output. (OpenAI)
Is Sol cheaper than Astra? Yes: $2 input and $10 output per million tokens. (OpenAI)
Can I get Ultrafast on the $200 plan? Yes, with Sol only, per OpenAI.
Sources
- OpenAI, “Introducing GPT-6.2 Astra and Sol” (Nov 17, 2026): https://openai.com/index/introducing-gpt-6-2-astra-sol/
AltTab Dad
