Field note
Claude Usage Limit Reached: What Teams Can Do
Claude's usage limit is a rolling five-hour window plus a weekly cap, shared across chat and Claude Code, so a small team should plan around it rather than treat it as a fault.
Claude's usage limit is a rolling five-hour window plus a weekly cap, shared across chat and Claude Code, so a small team should plan around it rather than treat it as a fault.
How the limit works
Paid Claude plans meter usage in two windows. The five-hour session limit starts when you begin working and resets five hours later. The weekly limit sits above it and resets on a schedule you can see in Settings, Usage. Some plans show a separate weekly bar for a specific model family, and the message you see names the one you have hit.
Three facts change how people should plan:
- The allowance is shared. On Pro and Max, Claude Code and Claude on the web, desktop and mobile draw from the same pool. On Team and Enterprise, each seat's allowance is shared across chat, Cowork and Claude Code, and its size depends on the seat tier.
- It is not a message count. Anthropic lists what consumes usage: message length, file attachment size, current conversation length, tool use such as research and web search, model choice, effort level, artifacts, and multi-step tasks like running code or browsing. A short question in a very long chat costs far more than the same question in a new one.
- Cached content counts less. Material in a Project that is referenced again while still cached is cheaper than fresh text. Stable reference files are the cheap way to give Claude context.
Model-specific limits behave differently from the shared ones. "You've hit your Opus limit" allows a switch to another model family with /model in Claude Code or the model selector in chat. "You've hit your session limit" or "weekly limit" applies across all models, so switching does not help.
How to see usage
- In claude.ai: Settings, Usage on Pro, Max, Team and Enterprise plans shows progress bars for the five-hour session and the weekly limit, the time remaining in the current session, and when the weekly limit resets.
- In Claude Code:
/usageshows the same plan bars plus a breakdown of what is consuming usage, attributed to skills, subagents, plugins and MCP servers, with flags for behaviour such as long context or cache misses when one accounts for 10 percent or more. Pressdorwto switch between the last 24 hours and the last seven days./statusconfirms which account and authentication you are using. - For Team and Enterprise admins: the analytics dashboard shows adoption, and the spend report covers usage credits if they are on.
If /usage says the usage endpoint is rate limited, it shows the last known bars from the past hour and offers a retry with r.
Planning for a small team
Limits are set by Anthropic and change, so plan on patterns rather than numbers. The table gives a starting point for a UK business with a handful of Claude users.
| Who | Typical use | Risk of hitting the limit | Plan |
|---|---|---|---|
| Owner or manager | A few chats a day, occasional document | Low | Pro; no special measures |
| Marketing, sales, operations | Daily drafting, long documents, Projects | Medium | Pro or Team seat; one chat per task; Projects for reference material |
| Analyst using research and files | Long multi-step tasks, big uploads | High in bursts | Max or Premium seat; schedule heavy runs so they do not collide with the session window |
| Developer on Claude Code | Long sessions, large repositories | High | Max or Premium seat; /clear between tasks; Sonnet for routine work |
| Two or more developers | Parallel sessions and automation | Very high | Team plan with usage credits and per-member caps, or API for unattended jobs |
Spread heavy work across the day. If two people in a two-person Team both run big Claude Code tasks at 9am, both hit the five-hour wall by lunch. Staggering costs nothing.
What to do when someone hits it
- Read the reset time in the message. If it is under an hour away, wait; it is the cheapest option.
- Switch to a task that does not need Claude, or to a different model family if the message was model-specific.
- In Claude Code on v2.1.234 or later, use
/rate-limit-optionsto wait and continue the interrupted task automatically. Admins can control whether this starts on its own with theautoContinueAtUsageLimitmanaged setting. - If the work cannot wait, use usage credits. On Pro and Max the person enables them in Settings, Usage; on Team and Enterprise an admin turns them on and sets caps. Our note on budgeting usage credits explains the controls.
- Afterwards, fix the habit that caused it. The
/usagebreakdown will usually name it: a session left open all day, an MCP server returning huge results, or Opus as the default for everything.
Habits that stretch the allowance
- One chat per task, and a new chat when the task changes.
- Put stable reference material in a Project rather than pasting it.
- In Claude Code,
/clearbetween unrelated work,/compactwith a focus when continuity matters, and keep CLAUDE.md short. - Lower effort or turn off extended thinking for routine drafting.
- Turn off connected apps and web search when a task does not need them.
- Choose the model for the job. Sonnet handles most routine work at lower cost.
For the business continuity side, including manual fallbacks for customer-facing work, see AI chatbot usage limits and business continuity.