Hello OpenAI team,
Thank you for clarifying in the recently closed discussion that the five-hour limit display was temporarily removed as part of an incident response and should return once the incident is resolved.
Could you please share any estimated timeline for restoring the five-hour limit?
I am currently on a Pro plan. While the five-hour limit is absent, usage appears to draw continuously from the weekly allowance, with no capacity restored after five hours. This makes me hesitant to use Codex normally because a long-running task could consume most or all of the remaining weekly allowance in only a few hours. It is quite uncomfortable to pay for Pro while feeling unable to use the service actively for fear of exhausting the entire weekly allocation.
Could you please clarify:
- When is the five-hour limit expected to return?
- Is it currently intended that nothing resets after five hours and usage continues directly against the weekly allowance?
- How should Pro users safely manage usage while this temporary incident response remains in effect?
- If an account exhausts its weekly allowance during this period, will affected users receive a reset or any other remediation?
- Is there an official status page or announcement channel where we can monitor restoration progress?
I understand that temporary measures may be necessary while an incident is being resolved, and I appreciate the team’s work. More guidance would be very helpful for users trying to plan their weekly usage responsibly.
Thank you.
I think I would rather have the 5 hour window please, 
OpenAI
No way back ? I need 5 hour limit to manage my work load. For me OpenAI monthly subscription has become unusable.
We use Pro+ x20 and we’re glad the 5-hour limit has been removed. The x20 weekly allowance suits us perfectly (we use it the full day, every day) and we haven’t managed to reach the limit yet, despite using it too at the weekend.
Seems hard to make everyone happy.
Perhaps the best would be a configurable setting to define a budget for the task/day?
Like: run this task but the budget is 20% of the weekly limits. When reached, it pauses and the user can decide between stopping, pausing it for later or overriding it?
People these days… we’ve actually reached the point where people are asking for the return of rate limits. A few years ago everyone complained that usage caps were too restrictive and begged for them to be removed. Now we’re seeing the exact opposite: people are asking OpenAI to bring the 5-hour limit back because it’s somehow the more comfortable system. What a timeline.
I genuinely don’t understand the problem. If you have a weekly allowance, manage it like any other limited resource. If you know you’ve got important work later in the week, don’t burn through your quota on day one. Isn’t that just basic planning?
The whole argument seems to be, “I’m afraid I’ll accidentally use too much.” Then… don’t? Nobody is forcing you to run Codex continuously for five hours straight. Stop when you’ve done enough for the day and continue tomorrow. It’s called self-control and time management.
What surprises me most is that many of the people saying this call themselves software developers. Developers spend their entire careers working with quotas, budgets, rate limits, CPU time, memory constraints, API limits, and cloud spending. Managing finite resources is practically part of the job description. Yet somehow a weekly usage budget suddenly becomes impossible to handle unless there’s an automatic reset every five hours.
It’s honestly bizarre watching people ask a service to artificially limit them because they don’t trust themselves to manage their own usage. I never thought I’d see developers asking for more restrictions instead of fewer. What a strange timeline.
I’ve gotten used to not having the 5-hour limit. I wish it would stay like this.
5h limit being gone, is WAY BETTER as it allows me to get more done!
I second you on this. I have GPT x20 plan and codex consumed my entire weekly limit in 24 hours while autonomoulslyworking on two goals overnight.Now waiting for it to restore.
5 days before it restores…And also is there someone else too who feels that chatgpt now a days has issues in following instructions and with memory.
I have to remind it the goal every time.
I’m not sure I use the same and everyday the reset day is moved a bit further now I have to wait 3 days to have my full 100% keep in mind I had input more than 300$ of recharge and stil I’m at zero the 5h was good cause cold deliver a lot now I’m stuck and thinking about switching to Claude
First of all, my utmost respect to all opinions expressed here! This shows how different our worlds & workflows are. It is obvious that a setting (on or off) would be appreciated.
I absolutely LOVE the fact that Open AI experimented with removing that 5 hour window, which on Claude destroyed many of my sessions because they cut the session short ruthlessly. I usually execute some really massive plans, with good rules for when context is compacting. Since the removal of the window, I could do WAY more than I ever did. I am very happy with this.
So happy, in fact, that this is among the very reasons I upgraded to x20 (from a regular Plus) subscription.
I also love that OpenAI does not ruthlessly cut the session when it reaches 100%. Claude is very brutal. But I noticed that Codex is usually tolerant, waiting for a certain task to complete. Not only that, but when the 5 hour was in effect, it would continue with plans for quite a while. It was indeed eating from the weekly limit, but at least it didn’t stop plans mid-point.
In my view, Open AI seems to have more respect for our work. Adding the reset button so we can “game” resets, being tolerant with cut-offs, all these make, for now, Open AI, my choice for the top subscription :).
Keep up the great work!
Would be great to keep the situation as it is.
As a Pro-20x user, I am very satisfied with the fact I can consume my whole weekly quota, instead of waiting 5 hours each time.
They’ll work for 30 minutes and somehow the entire 5-hour usage window is already gone. That doesn’t make any sense.
OpenAI clearly has a usage problem right now. There’s a reason they removed the 5-hour limit/window the way usage was being calculated simply wasn’t working properly.
It will be better somewhere in settings to add checkpoint. Who wants will add and who wants to disable it
Just add a config to allow u to add them back if u want