The 5-hour allowance and the total weekly quota have inexplicably disappeared!

Resolved 💬 3 comments Opened Jul 11, 2026 by yoyoweny Closed Jul 11, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of the Codex App are you using (From “About Codex” dialog)?

26.709.11516

What subscription do you have?

Pro 200% per mouth

What platform is your computer?

MacOs

What issue are you seeing?

My 5-hour quota had just reset to 100%, and I hadn't done a thing; yet, just minutes later, I checked again and found it had inexplicably dropped to 22%—and my total weekly quota had plummeted significantly, too! Oh my god! My quota wasn't even enough to begin with! What on earth happened in those few minutes? Is there a bug with my account?
Feedback ID:019f52f0-9f04-79a1-b60b-d250084adcee

What steps can reproduce the bug?

Without me doing a thing, the credit limit just vanished inexplicably.

What is the expected behavior?

_No response_

Additional information

_No response_

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 1 month ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #32279
  • #32480
  • #32410
  • #32409
  • #32439

Powered by Codex Action

piotrsynowiec · 1 month ago

I have the same problem. It's been showing every now and then for a week or so.

Mahkhmood9 · 1 month ago

my personal recomenadion

as the models seen more users , and the models improved so much...

you should expect the same preformance out of the smaller models like Luna or Terra(the more expensive of two)
its also much smarter in terms of understand on which setup and computer you are on

with/while respecting the temptation we all suffer from

I recomend using Terra-Medium , its a very cost effective model

the 5-hours limit only meant for you - not to kill all of your weekly allowance in a one go , if things go the wrong way -- and some excessive token is happening...

I think as things go for now

  1. use Terra for most task
  2. you can use Sol for codebase review (but I dont use it because I dont feel its any better really -- but openai generally will let you use it -- I just dont see the point because you are not solving a crazy math problem ,,, for code review you can ask it to review code or use the /review)

<img width="948" height="288" alt="Image" src="https://github.com/user-attachments/assets/24dad402-5d38-4bdf-9058-076880b00564" />

I run like 300 prompts a day and I dont run out of my 5hours limit, I use a hourly reset once

for example I tried evaluating antropic claude sonnet, I am barely able to get 20 prompts done using sonnet5-low
(which is just a little bit better than gpt 5.4mini)

as per computers , as it needed to be any faster, cpus and gpus are so damn fast down --- and there is still development in term -- where all human equally have limitation in the design of them -- and compilers who made for them (which is a critical layer too - in how things will advance )