Paid credits consumed too quickly and Pro quota still blocked

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

Version 26.602.40724 • Released 05.06.2026

What subscription do you have?

plus and now pro

What platform is your computer?

Microsoft Windows NT 10.0.22631.0 x64

What issue are you seeing?

I want to give direct feedback about the current Codex and ChatGPT usage experience.

I have been a ChatGPT user since the beginning and I rely heavily on these tools, especially during active development phases. Right now I am in a crunch phase and the current usage and billing experience feels extremely frustrating and unreliable.

I was on ChatGPT Plus, but my usage was fully exhausted on June 7, with the next reset only on June 11. I then bought additional token or credit packages, including 500 and 1000 packages, spending around 60 to 70 EUR. These were consumed extremely quickly and did not feel proportional to the value received.

Because of that, I upgraded from Plus to Pro. After upgrading and paying significantly more, I continued working for about an hour. The usage display still showed around 77 percent remaining for the current Codex window, but the chat suddenly told me that I had no tokens left. This is incredibly confusing and feels broken. If the interface shows remaining quota, I should not be blocked by another message saying the opposite.

At this point, I have paid for Plus, additional credits, and now Pro, and still the system tells me I am out of tokens shortly after upgrading. This does not feel like a premium product experience. It feels unreliable, opaque, and financially punishing for users who are dependent on the tool during real work.

I understand that large language models cost money. I do not expect everything to be free. But the current system feels like it pushes users into buying more during time pressure without giving a clear, trustworthy explanation of what they are actually buying, how it is consumed, and why different quota displays contradict each other.

Please investigate this usage and quota behavior. At minimum, the system needs much clearer billing transparency, consistent quota displays, and a fairer handling of upgrades and purchased credits.

<img width="1652" height="386" alt="Image" src="https://github.com/user-attachments/assets/6b2ca390-9aa2-4905-a865-4ad5f95349de" />

What steps can reproduce the bug?

Steps to reproduce:

1. Use Codex on a ChatGPT Plus account until the normal usage limit is reached.
2. Buy additional usage credits or token packages.
3. Continue working in Codex on large development tasks.
4. Upgrade from Plus to Pro.
5. Start or continue a Codex session after the upgrade.
6. Check the Codex usage display. In my case, it still showed remaining usage, around 77 percent.
7. Try to continue the chat or task.
8. The system blocks the task and says that no tokens or usage remain, even though the quota display still shows remaining capacity.

Expected result:

After upgrading from Plus to Pro, the available Codex quota should be updated clearly and consistently. If the usage display shows remaining capacity, the user should be able to continue working. If another limit applies, the interface should clearly explain which limit was reached.

Actual result:

The Codex usage display showed remaining quota, but the chat still blocked me and said that no tokens were available. This happened shortly after upgrading from Plus to Pro, after I had already spent additional money on credit packages.

Session information:

Date: June 9, 2026
Account plan before upgrade: ChatGPT Plus
Account plan after upgrade: ChatGPT Pro
Purchased additional credits before upgrade: approximately 60 to 70 EUR
Visible usage display: around 77 percent remaining
Error shown: no tokens or usage remaining

Code snippet:

No code snippet can reproduce this bug because it appears to be an account, billing, quota, or Codex session state issue. The problem is caused by inconsistent quota handling between the visible usage display and the actual blocking behavior in Codex.

What is the expected behavior?

_No response_

Additional information

_No response_