I’m using codex with o3-pro to analyze tricky code and do some changes. After thinking for very long time and consuming lots of tokens and budget it gives an error without producing any result.
I tried bunch of prompts variations and it ate $13 and haven’t produced anything.
Simply reading and analyzing my code.
They must handle rate limit in Codex to avoid this otherwise it is simply waste of money for expensive models like 03-pro.
Other models couldn’t handle this task.