Claude Code runs on your Claude Pro or Max plan, the same plan as Claude chat. That plan has two usage limits, a session limit that resets every five hours and a weekly limit. Everything you do in Claude counts against both, whether it is a chat on claude.ai or a refactor in the terminal.
Most people find out where they stand the hard way. A refactor is half done, the terminal says the limit is reached, and it tells you when it resets. Then you do the math on whether the rest of the week will hold.
Checking takes a few seconds once you know where to look. Here is every way to do it, what ccusage counts and what it doesn’t, and the board we built to keep both limits in sight beside Codex.
- In Claude Code, type /usage. On Pro or Max it shows bars for your plan’s limits, activity stats, and what used them.
/costand/statsopen the same screen. - On the web, open Settings > Usage. claude.ai/settings/usage shows how much of your five-hour session is gone, the time left in it, and when your week resets.
- The limits are shared with Claude chat. claude.ai, Claude Desktop and Claude Code all count against the same session and weekly limits. Max also has a separate weekly limit for Fable.
- ccusage answers a different question. It reads Claude Code’s logs on your computer and reports tokens and an estimated cost, which is not how much of your plan is left.
- Each of these waits for you to look. Context Board keeps both limits on screen as rings with reset times, beside Codex and the sessions waiting on you.
Check Claude Code usage with /usage
Type /usage in Claude Code. /cost and /stats are aliases, so all three open the same screen, and /stats opens it on the Stats tab. It runs straight away, even while Claude is in the middle of a response.
On a Pro, Max, Team or Enterprise plan, Anthropic’s cost and usage docs say you get “plan usage bars, activity stats, and a usage breakdown on the same screen.” The breakdown is the part worth reading when a week went faster than you expected.
- Attribution: how much of your recent usage came from each skill, subagent, plugin and MCP server.
- Behavior flags: anything that accounts for 10% or more of recent usage, such as long context or cache misses, with a tip to cut it.
- Loops: the heaviest
/loopand scheduled tasks, with how often they ran and their tokens per run.
Press d or w to switch between the last 24 hours and the last 7 days. One caveat from the docs. The breakdown is worked out from session history on this machine, so usage from another computer or from claude.ai is not in it. The plan bars are read from Anthropic, so they count everything on your plan.
The Session block at the top shows a dollar figure. Anthropic says it is meant for API users, and on Pro or Max “the session cost figure isn’t relevant for billing purposes.” If the request for your limits fails, /usage shows the bars it loaded in the past hour with a note saying how old they are. Press r to retry.
Two things look like answers and are not.
/status: in current versions it shows your version, model, account and connectivity, not your limits. One of Anthropic’s own help articles still points you to it.- A cost with no plan bars: check whether
ANTHROPIC_API_KEYis set. Anthropic’s help center says Claude Code uses that key instead of your subscription when it is there, and bills at API rates.
To keep the numbers in sight without typing anything, put them in the status line. On Pro and Max, Claude Code hands your status line script the five-hour and seven-day percentages and when each resets, from the first response of a session. Run /statusline and describe what you want. The status line docs list the fields. In the VS Code extension, /usage opens an Account & usage dialog where each bar shows how long until its limit resets.
Check it at claude.ai/settings/usage
Away from the terminal, open claude.ai/settings/usage, or Settings > Usage in Claude Desktop. It works on Pro, Max, Team and seat-based Enterprise plans.
Anthropic’s usage limit best practices describe the Plan usage limits section at the top of the page.
- Current session: how much of your five-hour session limit you have used, and how long the session has left.
- Weekly limits: when your weekly limit resets, for all models, and for Fable if your plan includes it.
Below that sit your usage credits, and a Resets section when Anthropic has given you a free reset. From Claude Code, /usage-credits opens this same page on Pro and Max.
This is the page to bookmark. It reads the same plan as /usage, and it puts the reset times in plain words.
How the five-hour and weekly limits work
Every paid plan has two limits. Anthropic’s article on how usage limits work calls them “a session limit that resets every five hours, and weekly limits that reset once every week.” Max adds a third, a separate weekly limit for Fable.
They are shared. In Anthropic’s words, “your usage of all different Claude product surfaces (claude.ai, Claude Code, Claude Desktop) counts towards the same usage limit,” and Claude Code in VS Code or JetBrains counts too. So an afternoon of long chats on claude.ai leaves less for Claude Code that evening.
Anthropic does not publish the limits as message counts. What it does say is relative. Max 5x includes five times the Pro plan’s per-session allowance, and Max 20x includes twenty times. How fast you get through it depends on the length of the conversation, the features you use, the model and the effort level.
In Claude Code, a few habits cost more than they look. The docs list them.
- Long sessions: Claude Code sends the whole conversation with every request, so a one-line question late in a long session still pays for the history.
- Breaks: your first message after a break longer than the cache lifetime, an hour on a subscription, reprocesses the full context.
- Background work: subagents, scheduled tasks and compaction all send requests of their own.
One distinction matters when you hit a wall. The session and weekly limits are shared across all models, so switching with /model does not help. A model-specific message like “You’ve hit your Opus limit” is different, and moving to another model family keeps you working.
Codex works the other way round. Its allowance is shared with ChatGPT Work but not with ordinary ChatGPT chat, and Pro plans there currently have no five-hour limit at all. How to check Codex usage has its side.
When your Claude limits reset
The session limit resets every five hours. Settings > Usage shows how long the current one has left.
The weekly limit is more predictable than people assume. Anthropic’s Pro plan and Max plan pages say it resets “at a fixed time each week that is assigned to your account”, and that the day and time “stay the same regardless of when you start using Claude or when your subscription begins.” Your next reset is in Settings > Usage.
So it is a schedule you can plan around. If your week resets Monday at 9:00 AM and you have 20% left on Thursday, the big refactor waits for Monday.
Two things leave that schedule alone. Usage credits keep you working past a limit, and Anthropic says they “don’t affect the reset timing of any of these limits.” And a free reset, which Anthropic hands out now and then, puts the session or the weekly limit back to full at once, while “your weekly limits still reset on their usual day and time.” You claim one with Reset for free, on the web or in Claude Desktop. The button is not in Claude Code, but the reset applies there too.
Here is the opinion. The five-hour limit is the loud one. It interrupts you, then comes back before dinner. The weekly one is quiet, and it costs more, because when it runs out on a Thursday nothing comes back until your reset day.
Track tokens and cost with ccusage
ccusage is a free, open-source command-line tool. Run npx ccusage@latest and it reads Claude Code’s logs from your computer, in ~/.claude/projects/ or ~/.config/claude/projects/, and reports your tokens and an estimated dollar cost. Nothing is uploaded.
It has daily, weekly, monthly and per-session reports, a blocks report for Claude Code’s five-hour windows, and a compact line for the status bar. It now reads Codex and several other coding agents as well.
It is accurate for what it measures. The cost is worked out from token counts and model prices, so if you pay for Claude Code through the API, it is close to the number you care about. The daily report is also a quick way to see which day ate your week.
On a Pro or Max plan it answers a different question. The dollar figure is a list-price estimate, not your bill, and the logs on one computer leave out your claude.ai chats and your other machines. What you mostly need on a plan is the percentage Anthropic reports, and that lives in /usage and Settings > Usage. We set the two side by side in Context Board vs ccusage.
Keep Claude Code usage in sight on a board
Every check so far waits for you to remember it. The status line comes closest, and it only shows in the terminal it lives in.
Context Board is our product, a live board you build for yourself. Its AI usage widget gives Claude Code two rings, one for the current five-hour session and one for the week. Each ring empties as you use it, turns amber under half and red under a fifth, and says when it comes back, like “resets in 47m” or “resets Mon 9:00 AM”. If your plan has a model’s own weekly limit, that shows underneath. Codex sits beside it in the same widget.
- Claude Code · apihas a question3m
Session under 20%? The next task goes to Codex.
On that board, Claude Code has 18% of its session left with 47 minutes to go, and Codex is at 88%. The next task goes to Codex, and the note on the board already says so.
The numbers come from Claude Code. Each time it checks your limits with Anthropic, it keeps a copy in its settings file on your computer. The runner, a small program you set up once with one line from Connect, reads that copy and sends the board the percentages and reset times. The widget says when Claude Code last checked. From your sessions, only names, states and percentages leave your computer, never what you and Claude Code wrote.
The same runner feeds the Agent activity and Needs you widgets. Every Claude Code session on your computers shows with its title and project, and whether it is working, done, or waiting for you to answer a question, give an OK or approve a plan. Start from the Agents template and it is all in place, or add each widget yourself from the gallery. Link a second computer or a server with the same line, and its usage and sessions join.
The board is live while it is open, and the widget fills the next time Claude Code runs. Context Board is free for 3 days, then $6 a month, with every widget in the one plan.
What happens when you hit the limit
Claude Code tells you which limit you hit and when it resets. “You’ve hit your session limit” and “You’ve hit your weekly limit” are the two plan-wide messages.
- A short wrap-up: if the session limit lands mid-response, Claude Code may finish the step in progress and show “Usage limit reached · wrapping up”. Anthropic is rolling this out gradually.
- Automatic continue: from Claude Code v2.1.234, a session stopped by a limit waits and picks the task back up on its own when the limit resets. The bottom of the session shows the time.
/rate-limit-options: a menu for keeping going, with waiting, usage credits and upgrading in it.- Usage credits: turn them on in Settings > Usage to keep working past a limit at standard API rates. The monthly API credits that come with Max cannot be used in Claude Code.
Anthropic’s article on using Claude Code with Pro or Max covers the upgrade path too, from Pro to Max 5x and from Max 5x to Max 20x.
Which check to use when
Each one answers a slightly different question.
In the terminal, /usage, plus the status line if you want it always there. Away from the terminal, Settings > Usage. For tokens and an estimated cost, ccusage. For both of your agents, all day, next to the sessions waiting on you, a board.
If you run Codex too, the Codex usage guide covers its limits, and a Claude Code dashboard puts every session in one place.
Look at the week before you start anything big. The five-hour limit will always tell you about itself. The weekly one tells you once it is too late.