Guide

Claude Code usage limits explained

Claude Code enforces usage limits based on your Anthropic plan. This guide covers what counts toward your limits, how the different windows work, and how to get the most out of your quota.

How Claude Code limits work

Anthropic tracks your Claude Code usage on two separate timers:

• 5-hour session window — starts when you send your first message. Resets 5 hours later regardless of activity. See reset details.
• 7-day rolling weekly cap — a total budget over a 7-day period. As old usage rolls off, new capacity becomes available.

Hitting either limit pauses Claude Code. The 5-hour window is what most users encounter first. The weekly cap mainly affects developers who use Claude Code for many hours every day.

Limits by plan

Anthropic does not publish exact token numbers for each tier. The multipliers (5x, 20x) are relative to the Pro plan baseline. Actual limits may vary as Anthropic adjusts pricing and capacity.

What counts toward your limit?

• Input tokens — your messages, system prompts, file contents that Claude reads, and tool results.
• Output tokens — Claude's responses, code edits, and tool call arguments.
• Tool calls — each Bash command, file read, file edit, grep, etc.
• Context window — longer conversations accumulate more input tokens because the full history is sent with each turn.

How to check your current usage

Inside Claude Code, run:

This shows your remaining quota and when the current 5-hour window resets. For a more detailed breakdown across sessions:

VibeUsage reads your local Claude Code data to show session counts, token usage, tool call breakdowns, and personalized tips.

Tips to maximize your limits

• Run /compact when conversations get long — it summarizes history to reduce token usage.
• Start fresh sessions for new tasks instead of continuing one giant conversation.
• Be specific in prompts — vague requests cause Claude to explore broadly, using more tokens and tool calls.
• Use targeted tools — Grep for searching is much cheaper than having Claude read entire files.
• Check your patterns with npx vusage — you might discover that most of your tokens go to a specific type of operation that can be optimized.

Rate limiting vs usage limits

Rate limiting and usage limits are different things:

• Usage limits — total budget per 5-hour/7-day window. Once exhausted, wait for reset.
• Rate limiting — requests-per-minute throttling. Temporary, resolves within seconds. Usually happens during peak hours.

If Claude Code says "rate limited," wait a few seconds and retry. If it says "usage limit reached," check your reset time.

Frequently asked questions