Claude Code usage: check your limits and when they reset
See how much of your Claude plan Claude Code has used: /usage, Settings, then Usage, when the five-hour and weekly limits reset, and what happens at the limit.
Type /usage in Claude Code (/cost and /stats open the same screen). On a Pro, Max, Ultra, Team or Enterprise plan it shows your plan’s limits, when they reset, and a breakdown of what counted against them; on an API key it shows the session’s tokens and an estimated dollar cost instead. The same figures, with your next reset time, are on claude.ai under Settings, then Usage. The five-hour limit starts counting at your first message and resets in full five hours later; the weekly limit resets at a fixed day and time assigned to your account. Chats in the Claude apps and Claude Code draw from the same limits.
Inside Claude Code
Claude Code’s command reference describes /usage as “Show session cost, plan usage limits, and activity stats”, with /cost and /stats as aliases (/stats opens on the Stats tab). It runs immediately even while Claude is in the middle of a response. What it shows depends on how you signed in, per the costs page:
- Signed in with a plan: “plan usage bars, activity stats, and a usage breakdown on the same screen.” In the breakdown, press
dorwto switch between the last 24 hours and the last 7 days. The breakdown covers this machine only: “usage from other devices or claude.ai is not included,” though the bars themselves are your plan’s. - With usage credits on: a usage-credits row with this month’s spend against your monthly spend limit, or
Unlimitedif you set none. - On an API key: the Session block, which shows the session’s token usage and a dollar figure Claude Code computes locally at list price. It resets to $0 when
/clearstarts a new session, and it is an estimate, not your bill.
If /usage says something like rate limited until 3:14am, that is about the usage lookup, not your work: Claude Code can’t fetch your plan limits again until then and shows the bars it last loaded on this machine within the past hour.
Anthropic’s help article on Claude Code with Pro, Max or Ultra still says to “Monitor your remaining allocation using the /status command.” The command reference, as of November 2026, describes /status as the Status tab with “version, model, account, and connectivity”, which is the place to see whether you are on your plan or an API key, not how much is left. For the figures, use /usage.
In your Claude settings
On claude.ai, in Claude Desktop or in the mobile app, Settings, then Usage shows how much of the current session you have used and the time remaining under Current session, and your next weekly reset time, per Anthropic’s article on usage limits. Because the apps and Claude Code share one allowance, this is the view that counts both. The same page holds:
- Usage credits: the toggle that turns them on, and Adjust limit for a monthly spend limit.
- Limit resets: a Reset for free button when Anthropic has given your account one. Anthropic says resets are “given occasionally to eligible plans”, and the limit-reset article notes the button “isn’t currently available in Claude Code in your terminal or IDE”, so you press it here.
- On Ultra, whether your account is verified: “If you don’t see a Current session bar, your account is verified,” and only the weekly limit applies.
When each limit resets
The five-hour session limit is not a fixed clock: “The period starts when you send your first message, and your limit resets in full five hours later.” Pro and Max have it; verified Ultra accounts do not. On Team, Claude Code’s costs page calls each member’s allowance one that “resets on a rolling five-hour window and a weekly window.”
The weekly limit resets “at a fixed time each week that is assigned to your account”, and that day and time do not move with when you start working. A free limit reset refills a limit early but leaves the weekly reset time where it was.
Fable on Max and Ultra is described two ways on Anthropic’s pages as of November 21, 2026. The Claude Code help article says “Max and Ultra plans also have a separate weekly limit for Fable”; the pricing page lists Fable on Max as “included, up to half your weekly limit” and at “50% of weekly limits” for Max and Ultra in its table, and Anthropic’s article on Fable models says they “draw from your plan’s regular weekly usage limits.” On Pro, Fable is paid with usage credits.
Anthropic publishes no hour or message counts for any window, only multiples of Pro. Those, the prices, and what usage-based Enterprise does instead (no per-seat limits, usage billed at API rates) are in which Claude plans include Claude Code.
When you hit the limit
Claude Code warns first, with a line such as You’ve used 85% of your session limit · resets 3:45pm. If the limit lands mid-response you may see Usage limit reached · wrapping up, which applies only to the answer already in progress. Then it blocks requests until the reset, with one of the messages Anthropic’s error reference lists:
You've hit your session limit · resets 3:45pm
You've hit your weekly limit · resets Mon 12:00am
You've hit your Opus limit · resets 3:45pm
You've hit your Sonnet limit · resets 3:45pmCopiedThe session and weekly limits cover every model, so switching models does not help. The Opus and Sonnet limits apply only to that model family, and /model to a model outside it keeps you working. Both windows count at once, and the reference warns that “a single burst of heavy activity, such as a large workflow fanout, can exhaust the weekly allowance before the session window resets”, which is worth knowing before you start many subagents at once.
In an interactive session signed in with a plan, Claude Code 2.1.234 and later waits and picks the task back up by itself, per the interactive-mode docs:
Usage limit reached · limit resets 3:45pm
Continuing automatically at 3:45pm · Esc to cancelCopiedIt does not resend your last message, and it gives up after three automatic continuations. To turn it off, switch off Continue automatically at usage limit in /config, or set autoContinueAtUsageLimit to false in your user settings. The wait is for interactive sessions; the docs offer it neither in background sessions nor in claude -p runs (what the limit means for unattended work is in Claude Code on a schedule). For those, the error reference’s retry settings, which it suggests turning on with CLAUDE_CODE_RETRY_WATCHDOG=1 “in unattended sessions such as CI jobs”, include one for this case from version 2.1.301: CLAUDE_CODE_RETRY_WATCHDOG_MAX_USAGE_LIMIT_WAIT_MS, “the longest time in milliseconds one request waits for your plan’s usage limit” to reset.
If the limits rule Claude Code out for you, Claude Code alternatives lists the agents that take other plans and models, including the ones that still run Claude. Your other routes from inside Claude Code:
/rate-limit-optionslists them all: wait, add usage credits, or upgrade./usage-credits(called/extra-usagebefore) opens your usage-credits settings, or on Team and Enterprise without billing access sends your admin a request. Credits are billed at standard API rates and cover Claude Code as well as chat./upgradeopens the page for a higher plan.
With usage credits on, the stop can come from their spend limit instead: You’ve hit your monthly spend limit on Pro, Max and Ultra, raised in Settings, then Usage, or an individual, team or organisation limit on Team and Enterprise that an admin raises. When a plan window is also what ran out, the message adds when it resets.
Over SSH, where no browser can open, both print the URL instead, from version 2.1.205 for /usage-credits and 2.1.323 for /upgrade. On an older version, open Settings, then Usage on any device.
On an API key instead of a plan
A Claude Console API key has no five-hour or weekly window: it is billed per token. Inside Claude Code, /usage shows only the local estimate described above, and /usage-credits isn’t available. The figures that are billed are in the Console, on the Usage and Cost pages, and a workspace called “Claude Code” is created there automatically, on which you can set a spend limit.
If you meant to use your plan and /usage shows a dollar cost and no bars, an ANTHROPIC_API_KEY in your environment is probably winning over your sign-in, and Anthropic warns that this means “API usage charges rather than using your subscription’s included usage.” /status shows which credential is in use; the fix is in the credit-balance-too-low guide.
Keeping an eye on it from a script
Claude Code passes its status line script a JSON object that, for Pro, Max and Ultra subscribers, carries both windows: rate_limits.five_hour and rate_limits.seven_day, each with used_percentage (0 to 100) and resets_at in Unix epoch seconds. The object appears only after the session’s first response from the API, and either window can be missing, so read it the way the docs suggest:
jq -r '.rate_limits.five_hour.used_percentage // empty'CopiedThat puts the session and weekly percentages on screen all the time, which helps on a long unattended session over SSH where nobody is running /usage.