How do Claude Code usage limits work?

The two limits

• Session limit. It starts with your first message and resets in full five hours later.
• Weekly limit. It covers all models and resets at a fixed time each week, whenever you started. Your reset time is under Settings > Usage on claude.ai.
• Model limits. Claude Code can also stop at an Opus limit or a Sonnet limit, which apply only to that model family, and some plans have a separate weekly limit for Fable.

Every request counts against the session and the weekly limit at once, so a few heavy days can use up the week before any one session runs out. On Ultra, verified accounts have a weekly limit but no five-hour limit.

The limits cover everything you do with your Claude account: claude.ai, the desktop app, and Claude Code in the terminal and in your IDE. A long chat on claude.ai leaves less for Claude Code. (Anthropic: how usage limits work)

What uses it up

How fast you reach a limit depends on the model, the effort level and how long the conversation is. Opus uses your limit faster than Sonnet (its API price per token is twice Sonnet's), and Claude Code's default model on paid plans is Opus 5.5 (Anthropic: model configuration). How to reduce Claude Code token usage covers what you can change.

What each plan gets

From Claude's pricing page on Nov 2, 2026:

• Free: no Claude Code.
• Pro, $20 a month ($17 a month billed yearly): Claude Code, with a session and a weekly limit.
• Max, $100 a month: 5x Pro's usage every five hours and every week.
• Ultra, $200 a month: 10x Pro's usage every week, and no five-hour limit for verified accounts.
• Team: usage per seat, shared with Claude chat and Cowork. A premium seat gets 5x a standard seat.
• Enterprise: a seat price plus usage billed at API rates.

Anthropic doesn't publish the limits in messages or tokens, and says it may adjust them. If you upgrade, the new limits apply right away, but an open Claude Code session keeps the old plan until you sign in again.

How to check your usage

• /usage in Claude Code shows your plan's usage bars and what counts against them. /cost and /stats open the same screen. Its breakdown counts only sessions on this computer, but use on claude.ai or another machine still counts toward the limit.
• On claude.ai, Settings > Usage shows the current session, the time left and your next weekly reset.
• The desktop app has a usage ring next to the model picker.
• Claude Code warns you as you get close, with a line like "You've used 85% of your session limit".
• On Pro, Max and Ultra, a custom status line can show the five-hour and weekly windows from rate_limits.five_hour and rate_limits.seven_day. Verified Ultra accounts have only the weekly one.

When you hit the limit

Claude Code stops and names the limit and when it resets, such as "You've hit your session limit · resets 3:45pm", or your weekly, Opus or Sonnet limit. Switching models doesn't get you past a session or weekly limit. You can:

• Wait. In an interactive session, when the reset is less than 24 hours away, Claude Code waits and carries on by itself. It's on by default from v2.1.234 when you sign in with a Claude subscription. For a reset further away, such as a weekly limit, start the wait from /rate-limit-options. Esc or Ctrl+C stops the wait, and the "Continue automatically at usage limit" toggle in /config turns it off.
• Use usage credits. /usage-credits opens Settings > Usage on claude.ai, where you turn credits on and set a monthly spend limit. On Team and Enterprise it opens your organization's settings or asks an admin. Credits are billed at API rates. The monthly API credits that come with Max and Ultra can't be used in Claude Code.
• Upgrade. A bigger plan's limits apply as soon as you switch. Run /logout and /login so an open Claude Code session picks up the new plan.

(Anthropic: Claude Code errors, using Claude Code with Pro or Max)

If you use an API key

With a Claude Console API key there's no session or weekly limit. You pay per token, so there's no reset to wait for. Your organization has per-minute rate limits instead, and hitting one shows "Request rejected (429)". An ANTHROPIC_API_KEY left in your environment makes Claude Code bill the API instead of your subscription; /status shows which one you're using.

When the limit looks wrong

If the meter jumps faster than your work, you may not be imagining it. On the Claude Code RAP Sheet, people report that usage limits drain too fast and that credits are ignored at the usage limit. Is my Claude Code usage limit a bug? shows how to tell.

More guides

• How do I reduce Claude Code token usage?
• Is my Claude Code usage limit a bug?

Plan and command facts come from Anthropic's own pages, checked Nov 2, 2026. Anthropic changes limits and commands often; when this page and theirs differ, theirs is right.