Skip to content

Guides

How to check your Claude Code usage, and see it over time

Checked on 5 November 2026 against the sources below. Updated on 12 November 2026.

Just ran out? What you can do now covers when your window comes back, how to keep working until it does, and how to make the next one last, for Claude Code and Codex.

If you have just run out, or think you are about to, type /usage in Claude Code now. It shows how much of your 5-hour window and your weekly window you have used, when each one resets, and, below the bars, what has been using them. If both bars are full, the reset time is when you can carry on without paying; if the message you got named one model, such as Opus, switching model with /model keeps you working. Everything else on this page is about seeing the same figures without asking, and keeping a record of them.

The quickest check is /usage inside Claude Code. On a Pro, Max, Ultra, Team or Enterprise plan it shows two bars, one for your 5-hour window and one for your weekly window, with the time each resets, plus a breakdown of what has been using them. The same two bars are on claude.ai under Settings > Usage. If you want the figures in front of you on every turn, Claude Code passes them to a custom status line as rate_limits.five_hour.used_percentage and rate_limits.seven_day.used_percentage, and /statusline will write the script for you. To see the figures over time rather than right now, you need something that records them: a local tool such as ccusage or claude-session-budget, which read Claude Code's files on your machine, or OpenReason Intel, which reports both windows from your machine to a dashboard that keeps the history.

Each option below is described by what it reads and what it cannot tell you, with the vendor pages it was checked against at the bottom.

Run /usage in Claude Code

Type /usage at the Claude Code prompt. The docs describe what a subscriber sees: "Subscribers see plan usage bars, activity stats, and a usage breakdown on the same screen." The bars are your 5-hour and weekly windows. /cost and /stats open the same screen; they are aliases now.

Below the bars is the part most people miss. On a paid plan, /usage shows "a breakdown of what counts against your plan limits, with recommendations": recent usage attributed to skills, subagents, plugins and individual MCP servers, each as a share of the total, and a recommendation for anything behind ten per cent or more of your recent usage, such as trimming a large CLAUDE.md or lowering the effort level. Press d or w to switch between the last 24 hours and the last 7 days.

Two limits of the screen, in the docs' own words. The breakdown figures "are approximate and computed from local session history on this machine, so usage from other devices or claude.ai is not included." And when the request for your plan limits fails, usually because the usage endpoint is rate limited, /usage shows the last bars it loaded in the past hour with a "Showing last-known usage" note; press r to retry.

The Session block at the top shows a dollar figure. For a subscriber it is not a bill: the docs say the session cost "isn't relevant for billing purposes" because usage is included in the subscription. It is computed locally at list price.

Check Settings > Usage on claude.ai

Open claude.ai/settings/usage. Anthropic's help centre: "You can see how much you've used and the time remaining under Current session in Settings > Usage", and "You can see your next reset time" on the same page. This is the account-wide view, so it includes chat and Claude Desktop as well as Claude Code, which /usage on one machine does not.

The same page is where usage credits live, if you have turned them on: the balance, this month's spend and the monthly spend limit. Running /usage-credits inside Claude Code opens it.

Show it in the status line

Yes, you can see your usage in the status line, and it is the only way to have the figures in view without asking. Claude Code's status line is a shell script that receives JSON on every turn and prints whatever you like. The status line docs list these fields:

  • rate_limits.five_hour.used_percentage and rate_limits.seven_day.used_percentage: "Percentage of the 5-hour or 7-day rate limit consumed, from 0 to 100".
  • rate_limits.five_hour.resets_at and rate_limits.seven_day.resets_at: "Unix epoch seconds when the 5-hour or 7-day rate limit window resets".

The easy way is to ask Claude Code to build it: run /statusline with a sentence such as "show my 5-hour and weekly usage as percentages", and it writes a script into ~/.claude/ and sets the statusLine key in your settings. If you would rather write it yourself, a script this short does the job:

#!/bin/bashinput=$(cat)five=$(echo "$input" | jq -r '.rate_limits.five_hour.used_percentage // empty')week=$(echo "$input" | jq -r '.rate_limits.seven_day.used_percentage // empty')echo "5h ${five:-?}% · week ${week:-?}%"

Save it as ~/.claude/statusline.sh, make it executable, and point statusLine in ~/.claude/settings.json at it, as the docs show. The // empty fallback matters: the docs say rate_limits "appears only for claude.ai Pro, Max, and Ultra subscribers, or behind a Claude apps gateway that sets a spend limit for you. Even then, it appears only after the first API response in the session." Team and Enterprise seats are not in that sentence, so if you are on one, try it before relying on it. Claude Code also drops a window from the JSON once its reset time passes, until the next response.

In the VS Code extension there is no status line, but the Account & usage dialog shows the same attribution shares and recommendations with a Day and Week toggle.

Track it over time with a local tool

Anthropic's screens show the current windows and the last seven days on one machine. For anything longer, or for a record of when you hit the limit, you need a tool that keeps one. Four local tools are widely cited, each reading something different, and OpenReason's free Session Forensics page is a fifth.

ccusage

ccusage describes itself as a tool to "analyze coding (agent) CLI token usage and costs from local data". It reads the session files Claude Code writes on your machine and produces daily, weekly, monthly, session and blocks reports. The blocks report groups usage into Claude Code's 5-hour windows, each "triggered by your first message" and lasting "exactly 5 hours from start time", with the tokens used in each. What it cannot know is your plan's limit: the --token-limit option is a number you set yourself, or max to use your largest previous block as the ceiling. So it shows how much you used, not how close to the limit you are, and its dollar figures are token counts priced at list rates rather than a bill. Its status line shows cost and context, not the 5-hour and weekly percentages Claude Code hands to a status line.

tokscale

tokscale describes itself as "A high-performance CLI tool and visualization dashboard for tracking token usage and costs across multiple AI coding agents". Like ccusage, it reads the session files of Claude Code, Codex and many other tools and prices the tokens at API rates, which its README calls "API-equivalent estimates, not subscription charges". Unlike ccusage, it can also show your current 5-hour and weekly percentages, by asking Anthropic for them with the sign-in Claude Code keeps on your machine, and it does the same for Codex. That is where you stand now; it does not keep those percentages over time. It is free and open source, and it uploads nothing unless you choose to submit your daily totals to its public leaderboard.

claude-session-budget

claude-session-budget is a Claude Code hook that will "track Claude Code's 5-hour session usage from local JSONL and pause tool calls before hitting the limit". It reads the same local files, works out where your 5-hour window started by reading the reset time from /usage, and when usage reaches a threshold (93 per cent by default) it stops Claude Code from dispatching tool calls until the window resets. It is installed from PyPI, Homebrew or the Claude Code plugin marketplace. It is a brake rather than a record: it covers the 5-hour window on one machine, not the weekly one, and it does not keep a history.

Usagebar

Usagebar is "a macOS menu bar app that tracks your Claude Code 5-hour and weekly usage limits". Rather than reading session files, it reads the sign-in Claude Code keeps in the macOS Keychain and calls Anthropic's usage endpoint, so it shows the same percentages Anthropic does, in the menu bar. It is a $9 one-time purchase with a free tier, and it is Mac only. It shows where you are now; it is not a history.

Session Forensics

Session Forensics is a free page from OpenReason that reads your Claude Code and Codex session files in your browser, with no install and no sign-up, and your files are never uploaded. It looks back seven days and answers a different question from the tools above: which limits you hit, and which turns used each window up, each with its estimated share. For Claude Code its 5-hour figure is an estimate, because Claude Code's files do not record the meter, and it does not show the weekly window.

If you are on a team

Everything above is per person and per machine. The person who pays for the team's plan does not see any of it: Anthropic's admin screens show adoption and, once usage credits are on, spend past the seat, a day behind. They do not show each member's 5-hour or weekly window. If the question is what the whole team uses and who is hitting the limit, see what your team spends on Claude Code and who went past their plan.

See your 5-hour and weekly windows over time in OpenReason Intel

OpenReason's reporter, or-otel, uses the same status line mechanism described above. One install command sets statusLine in your Claude Code settings to or-otel statusline, which reads the rate_limits fields Claude Code passes it, sends the used percentage and reset time of both windows at most once a minute and only when they changed, and then runs your own status line if you had one, so you keep whatever you were showing. In the dashboard you can see your 5-hour and weekly history, when each window filled, and the sessions and turns that were running when it did. It does the same for Codex, reading the limits Codex writes to its session file.

What it adds over the screens above, for one person: a record across days and weeks rather than now or the last seven days; every machine you use in one place, where /usage counts only the machine it runs on; each turn's estimated share of the window, so you can see what used it up; your Codex windows beside your Claude Code ones; and a record you can show your lead if your company pays and you need more.

What it does not do. It does not warn you or stop anything; it is a record you look at. What you can do instead is keep a status line showing your percentages as you work, which or-otel leaves running, and look afterwards at what filled the window. It reports sessions where the status line runs, which means Claude Code in a terminal: Claude Desktop has no status line, and a statusLine set in a project's own .claude/settings.json overrides it. The figures leave your machine and go to OpenReason, so it is not a local tool; the engineers' page lists exactly what is sent and what never is.

To start: sign up with a Google account or an email address, no card, create a team (a team of one is fine), and run the command from the or-otel quickstart on your Mac, Windows or Linux machine. Quit and reopen Claude Code, send a prompt, and the first report arrives within about a minute. or-otel uninstall puts your status line back and removes everything.

What to do next

  1. See what used up your limit

    Session Forensics is a free page that reads the Claude Code and Codex session files already on your computer, in your browser. It lists the limits you hit and the turns that used each one up. There is nothing to install and no sign-up, and your files are never uploaded. For Claude Code its 5-hour figure is an estimate, because Claude Code's files do not record the meter.

    Would you rather have your own agent do it? Paste this prompt into Claude Code or Codex.

    Open Session Forensics (opens in a new tab)
  2. Get an email when the limits change

    Get an email when Claude Code or Codex limits change. One email per change, at most one a week, from OpenReason. Unsubscribe from any email in one click.

  3. Track it over time

    Sign up with your Google account, no card, and run one install command. OpenReason Intel then records your 5-hour and weekly windows from every machine you use, with the turns that used each one up, and your Codex windows beside them. It is free for one person.

    It shows you what happened; it does not stop a session as you work, and it cannot change your vendor's limits.

    Sign up free with Google

Questions people ask

How do I check how close I am to my Claude Code limit?
Run /usage in Claude Code. It shows a bar for your 5-hour window and one for your weekly window, each with its reset time, and a breakdown of what has been using them. The same bars are on claude.ai under Settings > Usage, which also covers chat and Claude Desktop.
Can I see my Claude Code usage in the status line?
Yes. Claude Code passes rate_limits.five_hour.used_percentage and rate_limits.seven_day.used_percentage, with each window's resets_at, to your status line script on every turn. Run /statusline and describe what you want, and Claude Code writes the script. The docs say the fields appear for Pro, Max and Ultra subscribers after the session's first response.
How do I track my Claude Code usage over time?
Anthropic's screens show the current windows and up to seven days on one machine. For a longer record, ccusage reads Claude Code's local session files into daily, weekly, monthly and 5-hour block reports of tokens used; claude-session-budget watches the 5-hour window and pauses tool calls near the limit; OpenReason Intel reports both windows from every machine you use to a dashboard that keeps the history, with the turns that used each window up.
What counts against the Claude Code 5-hour window?
Every request on your account, in Claude Code and in chat, on every device. The cost of each depends on the model, the effort level and above all how much conversation is being sent with it, which is why a long-running session uses the window faster than a fresh one. The guide on how Claude Code's limits work has the full list.

Sources

Every figure on this page was read from these pages on 5 November 2026. Limits and prices change; if a vendor page now says something different, the vendor page is right and this one is out of date.