Claude Code Rate Limit Exceeded: What It Means and How to Fix It

When Claude Code says your rate limit is exceeded, you've hit Anthropic's usage cap for the current 5-hour rolling window. There's no fix other than waiting for the window to reset. The real solution is knowing when you're approaching the limit before you get cut off. Usagebar does exactly that: it sits in your macOS menu bar, tracks your active window, and sends notifications at 50%, 75%, and 90% so you can pace yourself.

• Claude Code limits are tied to a 5-hour rolling window, not a calendar day
• There is also a weekly usage cap on Pro and Max plans
• Usagebar costs $9 one-time and tracks both the window cap and the weekly cap

What does "rate limit exceeded" mean in Claude Code?

As of July 2026, Anthropic enforces usage limits on Claude Code based on a 5-hour rolling window rather than a fixed daily reset. When you exhaust the token or request budget within that window, Claude Code returns a rate limit error and stops responding until the window rolls forward. This applies to both Pro ($20/mo) and Max ($100/$200/mo) plan subscribers, though the caps differ by plan.

The error can appear mid-task, which is particularly disruptive during long agentic runs. Unlike API rate limits, which are measured in requests-per-minute, Claude Code's limits are tied to cumulative usage over the rolling window. Heavy context, large file reads, and multi-step tasks burn through the budget faster than simple prompts.

Why do you keep hitting the rate limit?

Several patterns cause developers to hit the ceiling faster than expected:

• Large context windows: Claude Code reads entire files and directory trees. Each token of context counts toward your limit, not just your prompts. See the Claude Code context window explainer for details.
• Agentic loops: Multi-step tasks where Claude Code runs commands, reads output, and iterates consume far more tokens than one-shot prompts.
• Opus 4.5 or Sonnet as default model: More capable models use the same token budget but complete tasks in fewer turns. Switching models doesn't give you more budget, but choosing a lighter model for routine tasks can slow consumption.
• Forgetting the weekly cap exists: Heavy usage early in the week can deplete the weekly limit, blocking you even if your 5-hour window is fresh. Check the difference between the 5-hour lockout and the weekly cap.

How do you check your current Claude Code usage?

There are three ways to check how much of your budget remains before hitting the limit:

1. /usage slash command: Type /usage inside an active Claude Code session. It returns a snapshot of tokens used in the current window.
2. claude.ai/settings/usage: The web dashboard shows usage history but doesn't give a real-time view of your rolling window status.
3. Usagebar: A macOS menu bar app ($9, one-time purchase) that monitors your 5-hour window, weekly cap, and context usage continuously. It sends alerts at 50%, 75%, and 90% thresholds so you get early warning before work stops mid-task.

For a detailed walkthrough of each method, see how to check your Claude Code usage limits.

When does the Claude Code rate limit reset?

The rate limit resets on a rolling 5-hour basis from the point your usage started climbing, not at midnight or on a fixed schedule. If you hit the ceiling at 2pm, the earliest relief is around 7pm, depending on when token consumption began. There's also a separate weekly cap that resets on a fixed schedule. Full details are in the Claude Code usage reset guide.

Anthropic documents the rate limit behavior on the Claude Code Pro and Max plan support page. The exact token numbers are not publicly disclosed and vary by plan.

How do rate limits differ across Claude Code plans?

Plan	Monthly Cost	Relative Limit	Notes
Pro	$20/mo	Baseline	Sufficient for moderate daily use
Max (5x)	$100/mo	~5x Pro	Designed for power users
Max (20x)	$200/mo	~20x Pro	For heavy agentic workloads

Upgrading your plan is the only way to get a structurally higher cap. The limits are enforced server-side; there's no client-side setting to bypass them. For a breakdown of whether Claude Code usage counts against your broader Claude Pro allowance, see does Claude Code usage affect Pro limits.

What can you do when the rate limit is exceeded?

There are a limited number of options once you're blocked:

• Wait for the 5-hour window to roll over. Check approximately when your heavy usage began, and estimate when the window clears.
• Switch to the Claude.ai web interface. This runs on a separate usage bucket and may have remaining capacity if you haven't used it heavily.
• Use the API directly. Claude Code is a subscription product; the API is billed per token. If you have API credits, you can continue work outside the Claude Code client. See the Claude Code vs API cost comparison.
• Upgrade your plan. Moving from Pro to Max unlocks a substantially higher ceiling.

If you're hitting the limit regularly, the underlying issue is usually lack of visibility into consumption. By the time the error appears, it's too late to adjust behavior for the current session.

How to avoid rate limit errors with Usagebar

Usagebar is a macOS menu bar app that connects to Claude Code's usage data and displays your current consumption in real time. It tracks three metrics: your 5-hour rolling window usage, your weekly cap, and per-session context size. Alerts trigger at 50%, 75%, and 90% of each limit, giving you enough time to wrap up the current task or defer lower-priority work.

At $9 one-time, it pays for itself the first time it prevents a long agentic run from failing mid-execution. There's no subscription, and it works across all Claude Code plan tiers.

If you're seeing the rate limit error repeatedly, Usagebar gives you the operational awareness to pace heavy work across sessions rather than discovering the ceiling at the worst possible moment.

Key takeaways

1. Claude Code rate limits are enforced on a 5-hour rolling window, not a daily reset
2. There is a separate weekly cap that can block you even if your hourly window is fresh
3. Check usage with /usage inside Claude Code, or at claude.ai/settings/usage
4. Upgrading to Max (5x or 20x) is the only way to raise the cap structurally
5. Usagebar ($9 one-time) adds proactive notifications before you hit the limit, not after

Sources

• Anthropic: Using Claude Code with your Pro or Max plan
• Anthropic Support
• Usagebar: When does Claude Code usage reset
• Usagebar: Weekly limit vs 5-hour lockout explained