Vibepanions Get Vibepanions

Home / Guides / Checking your limits

How to check your Claude Code and Codex usage limits

Updated 23 Sep 2026 by DevSparks

Quick answer: in Claude Code, type /usage, or open Settings > Usage on claude.ai. In Codex, type /status in the CLI, or open chatgpt.com/codex/settings/usage. Both products meter you with a 5-hour window and a weekly limit, and both show when each one resets.

What are the Claude Code and Codex usage limits?

They are the amount of work your subscription allows over a period of time, measured by two windows at once: a rolling 5-hour window and a weekly limit. They are separate from the context window, which only limits how long one conversation can get.

Claude Code. On Pro, Max, Team and seat-based Enterprise plans, Anthropic meters a five-hour session limit and weekly limits. The allowance is shared: claude.ai, Claude Code and Claude Desktop all count towards the same usage limit. Message length, attached files, tools, the model and the effort level all affect how fast you use it.

Codex. Codex is included across ChatGPT plans, and usage limits vary by plan. OpenAI's Codex pricing page gives ranges of local messages per plan and model rather than one fixed number, says local messages and cloud chats share your allowance, and notes that weekly limits may also apply. On plans with shared allowances, Codex shares a pool with ChatGPT Work.

What is the difference between the 5-hour and weekly limit?

The 5-hour limit caps a burst of work; the weekly limit caps the total. You can be stopped by either one. A long afternoon can use up the 5-hour window while plenty of the week is left, and a few heavy days can use up the week even if you never hit a 5-hour cap.

For Claude, the Settings > Usage page shows the current session (how much of the five-hour limit you have used and the time left in it) and your weekly limits, including a separate one for Fable if your plan has it. Anthropic's usage credits article says the included limit "will reset every five hours once you reach it".

For Codex, OpenAI's help article refers to "5-hour and weekly Codex usage windows". Whether the 5-hour window is on depends on your plan: from 25 August 2026 it applies again to Plus, while OpenAI's Codex lead said it would stay off "for the upcoming months" on Pro $100 and Pro $200.

When does the weekly limit reset?

At the time your usage page shows. Neither Anthropic nor OpenAI publishes one reset day for everyone in the pages we read, so read it from your own account:

How do I check my Claude Code usage?

Type /usage inside Claude Code. It shows session cost, plan usage limits and activity stats, and on a Pro, Max, Team or Enterprise plan a breakdown of what counts against your limits. /cost and /stats open the same screen.

If you use an API key instead of a subscription, there is no plan limit to check; Anthropic points you to /cost and your Console for spend.

How do I check my Codex usage?

Type /status in an active Codex CLI session. OpenAI's pricing page says it shows your remaining limits there.

Users have asked for more detail in the CLI: a feature request on the Codex repository (issue 15281) asks for the full usage and limits data in /status or a dedicated /usage command.

Why are some Claude numbers labelled estimated?

Because the real figures only exist once Claude Code has sent them. Claude Code includes rate_limits in its status line data after the first reply of a session, and only on Pro and Max. Before that, any tool that shows a Claude percentage is guessing from token counts in local logs.

Vibepanions keeps the two apart. It installs a small pass-through in front of your own status line, which then runs unchanged, and shows the 5-hour and weekly figures Claude Code reports. Until Claude Code has reported, it estimates from your own past windows and labels those numbers "estimated". Codex figures are read from what Codex writes into its session files, so they are reported, not estimated.

The difference between these checks and a monitor is timing. /usage and /status answer only when you type them; Vibepanions keeps the 5-hour ring on screen and shows which terminal session is waiting for you. There is a side-by-side of the options in Claude Code usage monitors for Mac, compared.

Did the limits change in 2026?

Yes, several times. The changes we could trace to Anthropic or OpenAI statements:

Limits move often, so treat any exact figure you read online, including here, as a snapshot. Your usage page is the current answer.

What should I do when I hit a limit?

Wait for the reset, or pay for more. The options differ by product:

To hit the limit less often, Anthropic's advice for Claude Code is to /clear between tasks, match the model to the job, point at files instead of pasting them, keep CLAUDE.md short, and ask for a plan before large changes.

See both limits without typing a command. Vibepanions runs on macOS 14 and later, reads only local files, and never asks for your Claude or OpenAI password (privacy). SGD 7.99 once, full refund within 14 days. Get Vibepanions.

Sources

Fetched 23 Sep 2026.

Next: Claude Code usage monitors for Mac, compared (2026).