Codex is OpenAI’s coding agent. Sign in with your ChatGPT account and it runs on your plan’s usage limits, which on Plus means a five-hour window and a weekly one. You don’t think about either until one runs out.
It happens mid-task. Codex finishes the turn it was on, because OpenAI lets an active turn complete. The next message is the one that stops. The banner then offers credits, a reset if you have one, an upgrade, or the wait.
At that moment you need two numbers. How much is left, and when it comes back. Below is every official place to find them, the page that looks right and isn’t, and the way we keep them on screen all day.
- In the Codex CLI, type /status. OpenAI says it shows your remaining limits during an active session.
/statuslinecan keep rate limits in the footer. - On the web, open the Codex usage page. chatgpt.com/codex/settings/usage shows your limits, when each resets, and your credit balance.
- platform.openai.com/usage is the API dashboard. It tracks spend on API keys, not your ChatGPT plan’s limits.
- Plus has a five-hour window and a week. Pro currently has no five-hour limit. Codex shares its allowance with ChatGPT Work.
- Every one of those checks waits for you to ask. Context Board keeps Codex’s limits on screen as rings that empty as you work, beside Claude Code and the sessions waiting on you.
Check Codex usage with /status
The fastest check is inside the terminal you are already in. In a running Codex CLI session, type /status.
OpenAI’s Codex pricing page says this is how you “see your remaining limits during an active Codex CLI session”. Per the CLI command reference, it also prints the active model, the approval policy, the folders Codex can write to, and how much context is left.
Two more commands sit beside it.
/statusline: choose what the footer of the CLI shows. Rate limits is one of the items, along with the model, context and git branch, and your choice is saved toconfig.toml. Turn it on and the limits stay visible in that terminal without typing anything./usage: your account’s token activity. Type/usage daily,/usage weeklyor/usage cumulativeto jump straight to a view. The same menu is where you redeem an earned reset, if you have one.
In the Codex desktop app, /status shows the chat ID, context usage and rate limits, and /usage opens your usage and billing settings, according to OpenAI’s app command list.
The catch is that every one of these lives inside a session. No session open, no number. People have asked OpenAI for a usage command that runs on its own. GitHub issue 29838 was opened in June 2026, and in October it is still open.
Check it on the Codex usage page
The web page is at chatgpt.com/codex/settings/usage. Sign in with the same ChatGPT account Codex uses.
OpenAI’s help article on using Codex with your ChatGPT plan tells you to check “which allowance is exhausted, your credit balance, and any reset time shown.” It is also the page to open first when a limit has already stopped you, because it lists what you can do next.
In the desktop app, the same information lives in Usage settings, where the usage meter sits. On an Enterprise workspace that pays in credits, Usage & billing shows a monthly allowance instead.
OpenAI’s own advice is to “check the dashboard every week or two to understand your pace and remaining capacity.” That is good advice for pace. It is no help at 4pm on a Thursday, when you are deciding whether Codex gets the next big task.
Why platform.openai.com/usage shows something else
It is an easy page to land on, because it is the one called Usage. It is also a different account.
platform.openai.com/usage is OpenAI’s API Usage Dashboard. It reports API activity and cost for an organization, by project, in UTC, and only organization owners or people with the Usage Dashboard permission can open it.
It only involves Codex when you run Codex with an API key. Then API pricing applies, and that is where the spend shows. Signed in with ChatGPT, your Codex work draws on your plan, so this page has nothing to say about it.
Some people run both. OpenAI suggests an API key for automation in shared places like CI, while you work on your plan at your desk. In that case your plan’s limits are on the Codex usage page, and the CI bill is on the API dashboard. Neither page shows the other.
How the five-hour and weekly limits work
There are two windows. A five-hour one, and a week.
OpenAI’s pricing page lists estimated local messages per five-hour window for Plus and Standard Business, model by model. They are ranges, and OpenAI says plainly that they “are not fixed message limits”. Under the table, one more line says “Weekly limits may also apply.”
Not every plan has both. The same page says “Pro plans currently have no five-hour limit”, so a Pro user is watching the week. Codex is included on Free and Go too, and OpenAI publishes no numbers for either.
The allowance is shared. On plans that have them, Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents all draw from one pool. Ordinary ChatGPT chat is not on that list, and ChatGPT’s image, voice and file upload limits do not apply to Codex.
Four things empty the pool faster.
- The model and the task: model choice, context, reasoning and tools all change what one message costs, so prompt length alone tells you little.
- Cloud over local: OpenAI says cloud tasks may use more of your allowance than local messages.
- Speed: Fast mode uses your included limits at 2.5 times the Standard rate.
- Pictures: image generation uses included limits about 3 to 5 times faster on average.
If you also run Claude Code, its limits work differently, because they are shared with ordinary Claude chat. How to check Claude Code usage covers that side.
When your Codex limits reset
Both limits come back on their own. The usage page and the limit banner show when.
There are also resets you can apply yourself, and they behave differently.
- A banked reset: some Plus and Pro referral promotions add one. Open the profile menu and pick the usage summary that says something like “1 reset available”. It refreshes both windows. In the CLI, the
/usagemenu also offers to redeem an earned reset when one is available. - A paid instant reset: Plus and Pro personal accounts can buy one on ChatGPT web and in the Codex desktop app. It restores your five-hour and weekly usage at once, and it cannot be saved for later.
- A one-time reset from OpenAI: occasionally given to eligible users. Not guaranteed.
The part people miss is that a reset moves your week. OpenAI’s article on paid weekly resets says the new weekly period starts with your first request after the reset, and the next automatic reset comes seven days after that. Buy one on Tuesday, send nothing until Wednesday morning, and your week now runs Wednesday to Wednesday. It also pulls your next week’s allowance forward rather than adding extra.
Before you buy a reset
Check how much you have left first. OpenAI’s article is blunt that a bought reset “applies immediately, even if you have usage remaining”. It does not wait for you to use up the rest.
So the worst time to buy one is with 30% of the week still in hand. You get a full week back, but the 30% is not added on top, so you paid for 70%. Buy it when the week is close to empty and the reset date is still days away. That is exactly the moment you need both numbers.
And one thing that will not work. OpenAI Support does not reset usage limits.
Keep Codex usage in sight on a board
Every check above has the same flaw. You go and look. The terminal you would look in is the one busy working, and the usage page is a tab you closed an hour ago.
Context Board is ours, so weigh the bias. It is a live board you build for yourself, and its AI usage widget shows what is left of your Codex plan as rings that empty as you use it. Green at half or more, amber from 20%, red under 20%. Each ring says when it comes back, like “resets in 1h 5m” or “resets Sat 11:00 AM”. Claude Code sits in the same widget, read the same way.
- Claude Code · landingasks to approve a plan6m
Codex is at 16% for the week. Big refactors go to Claude Code until Saturday.
- Webhook queue on Postgres
- Approve the landing plan
- Check Codex before the migration
On that board the decision makes itself. Codex has 16% of its week left and Claude Code has 61%, so the big job goes to Claude Code and Codex gets the small ones until Saturday.
The numbers come from Codex itself. As it works, Codex keeps a note on your computer of what OpenAI last told it about your limits. The board’s runner, a small program you set up once with one line from Connect, reads the newest one and sends the board the percentages and reset times. The widget says when that was, so a number is never older than it looks. From your sessions, only names, states and percentages leave your computer, never what you and Codex wrote.
You add AI usage yourself from the gallery, or start from the Agents template, which has it already. It shows the windows Codex reports for your plan, and it fills the next time Codex runs.
The same runner feeds the Agent activity and Needs you widgets. They list every Codex and Claude Code session on your computers, and whether each one is working, done, or waiting for you to answer a question or approve a plan. Run agents on a second computer or a server, and you link each one with the same line.
The board stays live while it is open. Context Board is free for 3 days, then $6 a month, with every widget included.
What happens when you hit a Codex limit
The turn in progress gets to finish. OpenAI’s words are that the agent “will be able to continue working on that turn, subject to fair use limits.”
After that, the limit banner and the usage page show what your plan allows.
- Credits: Plus and Pro users can buy credits and keep working without changing plan.
- A reset: a banked one if you have it, or a paid instant reset on Plus and Pro.
- An upgrade: Pro comes in $100, $200 and $500 a month tiers.
- Waiting: until the reset time the banner shows.
There is one more way out. Any user can run extra local chats with an API key, charged at standard API rates. That spend is the kind that does show up on platform.openai.com/usage, so this is how a Codex user ends up with numbers on both pages.
Which check to use when
Each place answers a slightly different question.
In the middle of a session, /status is quickest, and /statusline keeps the limits in the footer of that one terminal. To buy credits, use a reset or see your exact reset times, open the usage page. For spend on an API key, the API dashboard. To see Codex and Claude Code together, all day, with the sessions that need you, a board.
I would still open the usage page once a week, the way OpenAI suggests. Just not at 4pm, halfway through a migration, when the only thing I want to know is whether Codex can take the next task.