5 ms·

same boat. gpt 5.5/5.6 goes further on its own much more often than opus 4.8/5 does. codex capped ~300k context when claude does 1m. I don't feel codex is sav

same boat. gpt 5.5/5.6 goes further on its own much more often than opus 4.8/5 does. codex capped ~300k context when claude does 1m. I don't feel codex is saving tokens, and result is usually not as good imo.

gatio 1mo ago
> codex capped ~300k context when claude does 1m. That's configurable in codex.. but there is a higher cost/usage to using it.