I’m experiencing the same issue, and it has been going on for several days now.

The “Selected model is at capacity. Please try a different model.” error appears very frequently and has made Codex extremely difficult to use. In many cases, I can only exchange a few messages before the session is interrupted again.

This is not limited to GPT-5.6 Sol for me. I’ve also encountered the same behavior with GPT-5.6 Terra and GPT-5.6 Luna.

What makes it especially frustrating is that the issue is intermittent. There are periods when everything seems to work normally for a while, but then the capacity errors suddenly return and start happening repeatedly again, sometimes making Codex practically unusable.

This has been happening for several days, so it no longer feels like a short-lived capacity spike.

Could the OpenAI team please investigate whether there is a broader capacity or routing issue affecting these models?

I’m hitting this on both Luna and Terra models across effort levels in the Codex Desktop App and it is utterly infuriating.

not only in desktop, in CLI same. Luna hits the wall, then Terra hits the wall. sometimes is it enought to prompt “continue”, or change thinking level or model.

honestly? I DO NOT MIND – it slows us down a bit, just a little bit. not like it was 5h limit back to the game.

at this point I just toggle between dozen of tmux windows, and check where Codex, Gemini, Opus or Z.AI has usage issue to switch to other model.

Btw Z.AI for Coding had reset of usage yesterday as well, and guess what – it did not change the date. It is actually super cool to wake up in the morning, and see 100% available and having only two days to use it

Just joining the thread as hitting the same limit message across all 3 5.6 models in Codex.
Reverted back to 5.5 but burning through those credits on the lesser model which is frustrating