GPT-5.2 Pro is proof that we need "extra extra high" reasoning
Resolved 💬 2 comments Opened Jan 25, 2026 by briandastous Closed Jan 25, 2026
💡 Likely answer: A maintainer (etraut-openai, contributor)
responded on this thread — see the highlighted reply below.
What feature would you like to see?
On complicated problems, ChatGPT 5.2 Pro routinely finds important issues that GPT-5.2 with "extra high" reasoning misses. We need to be able to spend more tokens for deeper reasoning on hard problems.
<img width="350" height="276" alt="Image" src="https://github.com/user-attachments/assets/abb52531-1406-4544-a662-a010b976736a" />
Additional information
_No response_
2 Comments
Thanks for the suggestion. We wouldn't expose such a feature as "extra extra high", but I get your point. We continue to improve the reasoning capabilities of our models with each new generation.
It would be great to be able to utilize GPT 5.2 Pro directly from within Codex CLI, even if we similarly are warned that we have like 1-3 usages in the entire month.