Feedback on Codex usage limits and Pro value
What version of the Codex App are you using (From “About Codex” dialog)?
Version 26.616.81150
What subscription do you have?
toke usage
What platform is your computer?
Microsoft Windows NT 10.0.22631.0 x64
What issue are you seeing?
Hello OpenAI team,
I would like to share feedback about my current experience with Codex after upgrading from Plus to Pro.
I upgraded because Pro is supposed to provide significantly more usage, but in practice the available quota does not feel proportional to the price difference. I understand that this may be subjective, and that usage can depend on context size, model choice, and working style. Still, from my experience, I feel like I only get around twice as much practical work done compared to Plus, not the advertised increase I expected.
I mostly work at medium speed and only rarely use fast mode. This is similar to how I worked before on Plus, so the difference in available work capacity feels surprisingly small.
Today is June 24, and I cannot continue working with Codex until June 29. That means I now have to wait five days, which is very difficult for an active development workflow.
I also want to mention the reset system. I appreciate that resets are available, and I do see that as generous. However, once I use a reset, the seven day window seems to start again from the reset date. In my case, I could have waited around three days, but after resetting, I now have to wait five days instead. This makes the reset behavior feel confusing and not very helpful in practice.
Overall, the current value of Pro does not feel right to me in this situation. I am paying significantly more, but the actual working capacity with Codex does not feel reliable enough for the kind of development work I am doing.
I would appreciate it if your team could review whether the quota calculation and reset behavior are working as intended, and whether the Pro value for Codex users can be improved or communicated more clearly.
Best regards,
Kaan Sevinc
What steps can reproduce the bug?
Feedback ID: 019e050d-bb9e-7b81-85c2-679135cc6962
What is the expected behavior?
As a Pro user, I would expect the Codex usage quota to provide a clearly noticeable increase in practical working capacity compared to Plus, especially since Pro is advertised as offering significantly more usage.
The quota calculation should feel transparent, accurate, and predictable. If usage depends on model speed, context size, or task complexity, this should be communicated clearly in the interface.
I would also expect the reset option to help restore access without creating a worse waiting period than before. If using a reset restarts the seven day window, this behavior should be clearly explained before confirming the reset, so users can make an informed decision.
Additional information
_No response_
2 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Update: I changed to Pro 20x, and the usage situation is still unsustainable
I want to add another update regarding my Codex usage, because unfortunately the situation has become even more frustrating.
I am currently in a very intense crunch phase with a large development backlog. I did not want to go from roughly €100 to more than €200 per account, because that is simply a significant amount of money for me. However, I eventually decided to upgrade to the Pro 20x usage tier because I thought this would finally give me enough headroom to work through my backlog properly without constantly worrying about hitting the weekly limit.
For the first few hours, I used Strong. I then switched fairly quickly to Medium because I did not see enough additional value from Strong to justify the much higher usage. In fact, I was extremely disappointed by the error rate and overall quality I was getting from what is supposed to be the top tier model.
What is especially frustrating is the usage itself.
It has not even been 48 hours since my weekly usage was effectively back at 100% after changing to the 20x tier.
I have already consumed approximately 70% of that weekly allowance.
That is on a plan costing more than €200, with 20x usage, in less than two days.
And my backlog is still enormous.
This is exactly why I upgraded in the first place. I thought that, at least during this crunch phase, I would finally be able to use Medium comfortably, occasionally use Strong where it actually makes sense, and simply get work done.
Instead, I am once again watching the usage meter disappear at an incredible rate while simultaneously dealing with results that frequently require additional iterations because of mistakes.
That combination is becoming extremely difficult to justify financially.
I have been using Codex since its first week, and over that time I have repeatedly observed substantial fluctuations in quality. Sometimes these differences even appear to depend on the time of day.
I regularly encounter:
I have tried practically every workflow I can think of to mitigate this. I create intermediate summaries. I restart conversations. I reduce context. I explicitly document architecture and constraints. I split work into smaller tasks.
It does not fundamentally solve the problem.
At some point it becomes very difficult not to feel that I am paying more and more primarily for the ability to retry work that should not have failed in the first place.
That is the part I find hardest to accept.
I am willing to pay a premium for a genuinely powerful development tool. I would also understand high token consumption if the corresponding output quality and reliability were consistently excellent.
But paying more than €200 for a 20x usage tier and then seeing around 70% of the weekly allowance consumed in less than 48 hours, while still spending a significant amount of that usage correcting mistakes, does not feel like a fair value proposition.
I am tired of constantly having to think about usage while I am trying to work.
The entire reason I upgraded was to remove that pressure temporarily during this crunch phase. Instead, I am already approaching the same situation again after less than two days.
Once this crunch phase is over, I will have to seriously reconsider how I use Codex and whether continuing at this price level makes sense for me.
I genuinely want Codex to work. I have used it since the beginning, I have integrated it deeply into my development workflow, and I know how capable it can be when it performs well.
That is precisely why the current experience is so disappointing.
At more than €200 per account, I should not feel after less than 48 hours that I am already running out of the product I just paid for.