Usage & Limits
Claude usage and limits
Everything about being stopped: how the windows are measured, why the same plan feels different week to week, and what to do in the moment you get cut off.
Usage limits are the most confusing part of Claude because they are not counted in anything you can see. You are not billed per message, so the only signal you get is the message telling you to come back later — usually mid-task, usually on a deadline.
The underlying mechanic is simple even if the numbers are not published: your allowance is consumed by tokens, and the tokens you send include everything already in the conversation. Two people sending the same twenty messages can consume wildly different amounts. That is why the fix is usually structural rather than a bigger plan — see the playbook.
In this hub
- How many tokens with Claude ProHonest answer: no published hard token allotment — session/weekly capacity and how to estimate throughput.
- How to check your Claude usageSettings > Usage, Claude Code /usage, API headers, and Console — every meter, every plan.
- When Claude's usage limit resetsSession (~5h) vs weekly account-assigned reset times — and how to plan heavy work around both.
- Team spend trackingAttribute spend to seats and projects, set alert thresholds, and handle concentrated usage.
- Team seat limitsStandard vs Premium seats (1.25× / 6.25× Pro), per-member bars vs the shared-pool myth, and Team credits.
- Max usage limitsHow the Max 5x/20x session and weekly bars actually work — and when Max still hits walls.
- Why you keep hitting Claude's usage limitDiagnose what burns the window fastest, then fix each habit before upgrading.
- Extra usageBuying through a bad week without upgrading your plan every month.
- Increasing your limitsEvery route to more headroom, cheapest first.
- Rate limits vs usage limitsAPI RPM/ITPM/OTPM vs consumer usage bars — and what a 429 really means.
- LLM budgets & spend alertsOrg and workspace spend caps, Console notifications, and Usage and Cost API monitoring.