Skip to content

Codex 5-hr time runs out extremely quickly #16623

Closed

Description

@bitsnbytesio

What version of Codex CLI is running?

gpt-5.4

What subscription do you have?

Business

Which model were you using?

gpt-5.4

What platform is your computer?

Linux

What terminal emulator and version are you using (if applicable)?

N/A

What issue are you seeing?

Suddenly I've noticed my 5-hour time limit is being exceeded in about 2-3 hours. I guess I ran an update and am running on the newest version which appears to be causing the issue. I was previously using gpt-5.3-codex without any issue's. At first I thought it was my imagination until other individuals started complaining about the same thing.

What steps can reproduce the bug?

Just start codex in gpt-5.4 and use it for a couple of hours.

What is the expected behavior?

It should have 5-horus of usage before I'm blocked.

Additional information

This has greatly impacted my work for the last day or two.

Activity

  1. added
    CLIIssues related to the Codex CLI
    rate-limitsIssues related to rate limits, quotas, and token usage reporting
    on Apr 2, 2026
  2. etraut-openai commented on Apr 2, 2026

    @etraut-openai
    Contributor

    GPT 5.4 is billed at a higher rate than GPT 5.3-codex, so this is expected. For details, refer to the pricing documentation.

  3. bitsnbytesio commented on Apr 3, 2026

    @bitsnbytesio
    Author

    It's the 5-hour daily limit, not the weekly limit. My weekly limit hasn't changed. It only lasts about 2-hours. My weekly limit is still plenty.

  4. liudesheng9 commented on Apr 3, 2026

    @liudesheng9

    Even in 5.4 xhigh, the usage is dramatically faster than before. Just one 30-minute task will now drain the entire 5-hour limit. Whereas before 2–3 hours of two agents running in parallel would empty the 5-hour limit. I first observed this about 20 hours ago from my end. There might have been some internal update I guess.

  5. jh889h commented on Apr 3, 2026

    @jh889h

    GPT 5.4 is billed at a higher rate than GPT 5.3-codex, so this is expected. For details, refer to the pricing documentation.

    The issue is not that GPT-5.4 costs more. The real problem is that the Business plan has effectively become worse than the Plus plan.

    During the promotional phase, both Business and Plus users had 4× usage compared to the current limits. Now, Business only has 1× usage, while Plus has 2×. This means Business users are getting significantly less value than before—and even less than Plus users—which is a downgrade from the previous situation.

  6. fssrepository commented on Apr 21, 2026

    @fssrepository

    I don't mind to jump to pro, but as a normal developer expects, not unlimited but usable limits.

    I think there are too many subscriptions at the moment:

    1. Free tier weekly limit (no 5h limit) = 5h limit of plus (5h limit window does not make any sense anymore)
    2. Go tier does not make any sense anymore.

    Naturally it is hard to tell the customer, that you need to increase the price 5 times. (based on my experience is probably 2,5-3 times, as with pro you are getting more)

    Based on add credit (where i can buy 1000 credits, i would introduce multiplier instead, temporarily double the usage limit by 2/3/4/5...)

    It's definitely, not how open ai should work. If you need to increase price, whatever reason, you shouldn't shrink quotas what users are used to. The plus at the moment is like 3x less, than before.

    I would have increased the "Go"/"Plus" price to double, but providing the similar somewhat reduced (not that noticeable) quota as before (at least on 5h window). OpenAI is also mixing the codex usage with the chatgpt usage, and i think they shouldn't as they are totally different things and usage. If someone is using codex, hardly use chatgpt, and many times might use other prompts like gemini/claude web etc.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    CLIIssues related to the Codex CLIbugSomething isn't workingrate-limitsIssues related to rate limits, quotas, and token usage reporting

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions