Claude Code usage limits: how they work and how to check them

How do Claude Code usage limits work?

If you sign in to Claude Code with a Pro, Max or Ultra subscription, you do not pay per token. You get an allowance, and the allowance is rationed over time. Anthropic's help center article on how usage and length limits work describes it as a conversation budget. Paid plans have two types of usage limit: a session limit that resets every five hours, and weekly limits that reset once each week.

Three points save confusion. First, the limits are shared: usage across claude.ai, Claude Code and Claude Desktop counts toward the same limit. Using Claude Code in VS Code, Cursor or a JetBrains IDE counts toward the same shared limit too. Second, the weekly reset is not rolling. You can see your next reset time in Settings > Usage on claude.ai. Third, Anthropic does not promise fixed numbers. Its help center says it may adjust usage limits at its discretion to manage capacity and ensure fair access, and it does not guarantee uninterrupted service.

This page does not repeat what each plan costs or give message counts, because neither is stable. For plan prices and how Claude Code is billed, see Claude Code pricing. For how Pro and Max compare, see Claude Pro vs Max.

What counts toward Claude Code usage?

The help center lists the factors that affect usage: the length and complexity of your conversations, the features you use, which model you are using, and the effort level you selected. Claude Code's own documentation explains why a coding session adds up faster than a chat. Claude Code sends your full conversation with every request, and each time Claude uses a tool it sends another request carrying that batch of tool results.

One number to ignore: the docs say the dollar figure in /usage is computed locally from token counts at list price, and that for Pro, Max and Ultra subscribers, whose usage is included in the subscription, it is not relevant for billing. Treat it as a size indicator for your session, not a bill.

How do you check your Claude Code usage?

• `/usage` in a session. Shows session cost, plan usage limits and activity stats. On Pro, Max, Ultra, Team and Enterprise plans it adds a breakdown of what counts against your plan limits: recent usage attributed to skills, subagents, plugins and individual MCP servers as a percentage of the total, plus recommendations. Press d or w to switch between the last 24 hours and the last 7 days. The figures are approximate and computed from local session history on this machine, so usage from other devices or claude.ai is not included. /cost and /stats are aliases.
• Settings > Usage on claude.ai. Shows the current session and the time remaining, and your next weekly reset. This is the number that includes every device and surface.
• The status line. Subscribers' sessions can receive a rate_limits object with the percentage used for the five-hour and weekly windows. The status line guide has a script that shows it.
• Warnings. You get a warning message as you approach a limit.

Anthropic's own support article on using Claude Code with a plan also names the /status command for checking your remaining allocation. In the current docs, /status opens the Status tab with version, model, account and connectivity, and /usage is the one with limit bars, so try /usage first.

A related message: rate limited until 3:14am in /usage means Claude Code cannot fetch your plan limits again until then, not that your work with Claude is limited. Press r after that time for fresh data.

When /usage shows a hint such as f to fix 2 of 3, pressing f has Claude Code act on the recommendations it can, such as running an /effort command. It requires Claude Code v2.1.305 or later. An example row from the docs reads 25% Trim ~/code/api/CLAUDE.md · 49k tokens per request. Each row's share tells you how much of your recent usage it applies to, not how much the change would save.

What happens when you hit the limit?

You will see a message such as “You've hit your session limit” or “You've hit your weekly limit”, with the time the window resets. These are shared across all models, so switching models with /model does not restore access. A model-specific message, such as “You've hit your Opus limit”, is different: switching to a model outside that family keeps you working.

Two behaviors soften the stop.

1. Wrap-up allowance. If you reach your five-hour limit while Claude Code is partway through a response, it may keep working briefly to finish the step in progress. You will see “Usage limit reached · wrapping up”. The help center says the allowance is small and capped, counts toward your weekly limit, and applies only to a response already in progress. On Pro it applies up to once in each weekly usage period. It is rolling out gradually, so your account may not have it yet.
2. Automatic continue. On Claude Code v2.1.234 or later, in interactive sessions signed in with a claude.ai subscription, the docs say Claude Code waits in the open session and continues the task on its own after the limit resets. You will see Continuing automatically at 3:45pm · Esc to cancel. Keep the session open. Cancel with Esc at an empty prompt, or with /rate-limit-options.

Automatic continue still asks for permissions as usual, so the task can stop on a prompt while you are away. If it hits the limit again, Claude Code re-arms the wait at most twice in a row and then stops. If you plan to walk away, a mode that prompts less, such as auto mode, makes it less likely the task stalls on a question.

What are usage credits, and are they the same as API credits?

No, and the help center is careful to separate them. Usage credits let Pro, Max and Ultra subscribers keep working after hitting a limit by switching to consumption-based pricing at standard API rates. You enable them in Settings > Usage, add a payment method, and can set a monthly spend cap and auto-reload. Claude Code's /usage-credits command opens the same page in your browser. When one of your included limits is fully used, the next request goes through on credits and the notice changes to “Now using credits.”

The monthly API credits that come with Max and Ultra plans are a different thing. The help center states they cannot be used in Claude Code or the Claude apps, including for usage credits. If you subscribed through a mobile app, you can only enable usage credits on the web version of Claude. In most cases usage credits do not expire, but in some jurisdictions they do; the help center names Japan, with expiry six months after purchase for purchases from September 10, 2026.

If you want to stay strictly within your plan, decline the API credit option when it is offered, wait for the reset, and check your allocation. To stop the API credit prompt from appearing at all, run claude logout, then claude login with only your plan credentials and no Console credentials.

How do limits work on Team and Enterprise plans?

Each member's Claude Code usage draws from a per-seat allowance that resets on a rolling five-hour window and a weekly window. The Claude Code docs say the allowance is shared with Claude chat and Cowork, and its size depends on the member's seat tier. Administrators can turn on usage credits and set spend limits at the organization, group or individual level. Usage inside the seat allowance is not metered in dollars.

Admins see different messages from individuals. “You've hit your individual spend limit”, “org's monthly spend limit” or “team's shared budget” mean a request would be billed to usage credits and those credits reached a limit the administrator set. In that case a member runs /usage-credits to request more from an admin, and an admin raises the limit named in the message under Organization settings > Usage.

A context or auto-compact warning is not a usage limit. It means the conversation is close to the session's auto-compact window, where Claude Code summarizes older history.

How do you make your Claude Code limits last longer?

Everything below comes from the Claude Code cost documentation. Tokens scale with context size, so most of it is about keeping context small.

1. Clear between tasks. /clear starts fresh and costs nothing. Use /rename first so you can /resume later.
2. Compact with focus. /compact Focus on code samples and API usage tells Claude what to keep. You can set standing instructions in your CLAUDE.md.
3. Pick the model for the job. The docs say Sonnet handles most coding tasks and costs less than Opus; reserve Opus for complex architecture or multi-step reasoning.
4. Lower the effort level with /effort for simple tasks, since higher effort uses more tokens.
5. Trim what loads every session. Keep CLAUDE.md under about 200 lines, move workflow-specific instructions into skills, and disable MCP servers and plugins you are not using. claude plugin details <name> shows a plugin's always-on cost.
6. Write specific prompts. “Add input validation to the login function in auth.ts” triggers fewer file reads than “improve this codebase”.
7. Plan before big changes. Plan mode prevents expensive re-work when the first direction is wrong.
8. Delegate noisy work to subagents, knowing their own requests still draw on your usage. Choose a smaller model for them.

# Compact instructions
When you are using compact, please focus on test output and code changes

Background features use tokens too. Session recap, idle compaction and prompt suggestions can send requests while you are idle, and each can be turned off. The guides to CLAUDE.md files, plugins and Claude Code memory cover the pieces that load into every session.

If a limit is all that stands between you and a finished build, one workable approach is to spend your budget on the version that matters. Pick a small idea from the board, take its generated build prompt, and run one focused session instead of several wandering ones.