Error messages

You’ve hit your usage limit.

You’ve hit your usage limit.

Codex prints this when the usage that comes with your ChatGPT plan has run out for the current window. The first sentence is always the same:

You’ve hit your usage limit.

What follows it isn't. We read the code that writes it in OpenAI's openai/codex repo and found five different endings, picked by the plan your account is on, and all of them are in the 0.156.1 release. It's a common sight. 161 issues in that repo quote "hit your usage limit" as of September 23, 2026, and 51 of them were opened after August 1.

The ending depends on your plan

• Plus: Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits
• Pro: Visit https://chatgpt.com/codex/settings/usage to purchase more credits
• Team, Business and the usage-based enterprise plans: To get more access now, send a request to your admin
• Free and Go: Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus),
• Enterprise, Edu, or a plan Codex doesn't recognize: nothing extra, just the reset time.

So the line also tells you which plan Codex thinks you're on. One person who dropped from Pro to Plus in September got the Plus wording (the Pro upgrade link) and found their usage at zero from the start of the new plan.

Reading the reset time

When OpenAI's server sends a reset time, Codex adds it in your local time. If the reset falls later today you get something short, like try again at 11:50 PM. If it's on another day you get the full date, like try again at Sep 28th, 2026 2:14 AM. When the date is several days away, it's your weekly allowance that's spent, since the five-hour window resets within hours. With no reset time from the server it says try again later.

Codex doesn't retry this error on its own. It sits in the same group of terminal errors as the "model is at capacity" one, so the turn stops and the message stays on screen until you act.

What to do

If the time is close, wait for it. If it isn't, the ending tells you what's open to you. On Plus and Pro you can buy credits, and Codex draws on them once the plan's included usage is gone. We've covered how those credits work, what they cost and when they expire. On a Business or Team workspace the extra access goes through your admin.

If the message names a model, as in You’ve hit your usage limit for followed by a model name and Switch to another model now,, only that model's allowance is spent. Run /model and pick another one.

When /status says you still have usage left

The complaint we saw most often is a meter that disagrees with the block. In April a Business user got the admin version of the message while /status in the same session reported "5h limit: 62% left". In May someone on Business with 169 purchased credits was blocked anyway, by the desktop app's version of the wall, which reads You’re out of Codex messages. Neither issue had an answer from OpenAI staff when we read them on September 23.

If the numbers disagree, check the usage page at chatgpt.com/codex/settings/usage, which is where the message sends you anyway, before assuming the meter in the terminal is right. Then run /feedback right after a failure and link the upload on a GitHub issue, because a screenshot of the two numbers side by side is what the open reports lack.

Lines that look alike

Workspaces that pay by credit have their own wording, such as Your workspace is out of credits. Add credits to continue. and a pair about a spend cap set in the workspace. And Selected model is at capacity. Please try a different model. has nothing to do with your allowance: that one means OpenAI is short of servers for the model.