Claude Usage Limits Explained 2026: Free, Pro, and Max Caps
Claude limits you on two layers: a 5-hour rolling window and a weekly active-hours cap, shared across Claude Code, chat, and Cowork. Full breakdown by plan, the 2026 changes, and concrete ways to stretch your budget.
TL;DR — How Claude's Usage Limits Actually Work
Claude limits your usage on two layers at once: a 5-hour rolling window for short-term activity and a weekly cap on active compute hours. How much you get depends on your plan and your model — Opus burns the budget far faster than Sonnet. As a rough guide, Pro gets ~40-80 Sonnet hours/week; Max 5x ($100) reaches ~480 Sonnet hours or ~40 Opus hours; Max 20x ($200) is 20x Pro. Crucially, Claude Code, Claude.ai chat, and Cowork all share one pool. Anthropic doubled Code's 5-hour limits on May 6, 2026 and raised weekly limits 50% through July 13, 2026. Below: the full breakdown plus concrete ways to stretch what you have.
Figures reflect Anthropic's published limits as of late June 2026
Table of Contents
- The Short Answer
- The Dual-Layer Limit System
- The 5-Hour Rolling Window
- The Weekly Active-Hours Cap
- Limits by Plan: Free, Pro, Max
- The Shared Pool: Code, Chat, and Cowork
- How to Stretch Your Usage
- Length Limits vs Usage Limits
- Frequently Asked Questions
The Short Answer
If you just want the gist: Claude caps how much you can use it on two timescales simultaneously — a 5-hour rolling window and a weekly limit measured in active compute hours. Your plan (Free, Pro, Max 5x, or Max 20x) sets the ceiling, and the model you pick matters enormously, because Opus consumes the budget several times faster than Sonnet. Everything you do in Claude Code, Claude.ai chat, and Cowork draws from the same shared pool.
That is the whole model in three sentences. The rest of this guide fills in the numbers, the recent changes, and the practical levers for getting more out of your plan. If pricing is what you are really after, pair this with our Claude Code pricing guide.
The Dual-Layer Limit System
The thing that confuses most people is that there is not one limit — there are two, and they apply at the same time:
- The 5-hour rolling window governs short-term bursts. It stops any single stretch of intense use from monopolizing capacity.
- The weekly cap governs sustained use. It is measured in "active hours" of compute and resets weekly.
You can hit either one independently. A short, extremely heavy Opus session can exhaust the 5-hour window without touching your weekly budget much. A week of steady, light Sonnet chatting can nibble away the weekly cap while you never come close to the 5-hour limit. Understanding which one you are bumping into tells you what to change.
The 5-Hour Rolling Window
When you start using Claude, a 5-hour window opens. Your activity counts against a per-session ceiling, and the window rolls forward continuously rather than snapping back at fixed clock times. Hit the ceiling and you wait for the window to roll over before you can continue at full tilt.
This is the limit Anthropic adjusted most visibly in 2026. On May 6, 2026, the company doubled Claude Code's 5-hour rate limits and removed the prior peak-hours reduction for Pro and Max Claude Code accounts — meaning heavy coding sessions during busy hours stopped getting throttled the way they used to. If you remember Claude Code feeling tighter earlier in the year, that change is why it eased up.
The Weekly Active-Hours Cap
On top of the 5-hour window sits a weekly cap measured in active compute hours. This is the limit that matters for anyone using Claude as a daily driver. Roughly:
- Pro lands around 40 to 80 active Sonnet hours per week, depending on how heavy your prompts run.
- Max 5x ($100/month) stretches up to roughly 480 Sonnet hours or about 40 Opus hours per week.
- Max 20x ($200/month) is 20 times Pro's per-session usage.
Max plans actually carry two weekly limits running in parallel: one across all models and a separate, tighter one for Sonnet specifically — plus the small dedicated Opus budget. On May 13, 2026, Anthropic raised weekly limits 50% through July 13, 2026, a promotion still live as of late June. So if your numbers look more generous than older guides suggest, that promo is the reason.
Limits by Plan: Free, Pro, Max
Two caveats on this table. First, the hour figures are approximate — Anthropic measures compute, not wall-clock, so a session full of long, tool-heavy prompts burns the budget faster than the same minutes of light chat. Second, the numbers move. Anthropic has adjusted limits multiple times in 2026 alone, so treat any specific figure as a snapshot and confirm against the current help center. Our Claude Max plan review digs into whether the top tier is worth it.
The Shared Pool: Code, Chat, and Cowork
Here is the detail that catches people off guard: Claude Code, Claude.ai chat, and Cowork all draw from the same usage bucket. Burn through your budget on a marathon refactor in Claude Code, and you will find your Claude.ai chat capacity diminished for the rest of the window — and vice versa.
The practical implication: if you rely on Claude chat for daily work and run Claude Code for development, budget them together, not separately. Power users on Pro who do both often find Pro too tight precisely because they were mentally treating Code and chat as separate allowances. They are not. If you consistently run out, that shared-pool reality is usually the first thing worth examining — and the clearest signal it might be time to move up to Max. Founders shipping daily will find our Claude Code for startups guide useful for sizing the right plan.
How to Stretch Your Usage
Concrete levers, roughly in order of impact:
- Use Sonnet for routine work; save Opus for the hard stuff. This is by far the biggest dial. Opus eats the budget several times faster. Most day-to-day coding, drafting, and Q&A runs perfectly well on Sonnet.
- Keep conversations focused. A bloated context window means every turn reprocesses more tokens. Start a fresh session for an unrelated task instead of letting one mega-thread sprawl.
- Let MCP tools do the heavy lifting. Instead of pasting a 5,000-line file into the prompt, connect a filesystem or git MCP and let Claude read only what it needs. Reasoning tools like the Sequential Thinking MCP help the model solve problems in fewer, tighter turns rather than long rambling ones.
- Batch related questions. Ask the three things you need in one well-structured prompt rather than dribbling them out across many turns, each of which re-reads the full context.
- Watch the meter. Both Claude.ai and Claude Code surface usage indicators. Glancing at them before you start a big task tells you whether you have the headroom.
Length Limits vs Usage Limits
One last distinction, because the two get conflated. Usage limits are about how much you can use Claude over time (the 5-hour and weekly caps above). Length limits are about how big a single conversation can get — the context window. When a single thread grows long enough, Claude will tell you the conversation has reached its length limit and prompt you to start a new one.
These are independent. You can hit a length limit on a giant single conversation while having plenty of weekly budget left, and you can exhaust your weekly budget across many short conversations that never individually hit a length limit. The fix for length limits is simply to start a new conversation; the fix for usage limits is the stretching tactics above — or a bigger plan.
Want to do more with less budget? Browse 75,000+ MCP servers, skills, and agents on Skiln that make every Claude turn more efficient.
Browse Now →Frequently Asked Questions
What are Claude's usage limits?
Claude uses a dual-layer system: a 5-hour rolling window that caps short-term activity, and a weekly cap on active compute hours. The exact ceilings depend on your plan (Free, Pro, Max 5x, or Max 20x) and which model you use, since Opus consumes the budget far faster than Sonnet. The same usage pool is shared across Claude Code, Claude.ai chat, and Cowork.
How many hours of Claude do I get per week?
As a rough guide using Sonnet models, Pro plans get roughly 40 to 80 active hours per week. Max 5x ($100/month) extends that substantially — up to around 480 Sonnet hours or about 40 Opus hours, depending on session concurrency and task complexity. Max 20x ($200/month) is 20 times Pro's per-session usage. These are approximate and vary with how heavy your prompts are.
Did Claude's limits change in 2026?
Yes, several times. On May 6, 2026, Anthropic doubled Claude Code's 5-hour rate limits and removed the peak-hours limit reduction for Pro and Max Claude Code accounts. On May 13, weekly limits were raised 50% through July 13, 2026, a promotion still active as of late June 2026. Limits are adjusted periodically, so always check the current Claude help center for exact numbers.
What is the 5-hour window in Claude?
Claude's short-term limit resets on a rolling 5-hour basis. When you start a session, a 5-hour window opens; your usage during that window counts against the per-session cap, and the window rolls forward continuously rather than resetting at fixed clock times. If you hit the cap, you wait for the window to roll over before continuing.
Does Claude Code share limits with Claude chat?
Yes. This trips a lot of people up. Claude Code, Claude.ai chat, and Cowork all draw from the same usage bucket. If you burn through your budget running a big coding task in Claude Code, you will have less capacity left for chat on Claude.ai, and vice versa. Plan heavy sessions accordingly.
Why does Opus run out faster than Sonnet?
Opus is the largest, most capable Claude model, and it consumes far more compute per token than Sonnet. On Max plans there is a separate, smaller weekly cap specifically for Opus-class usage (around 40 hours on Max 5x) alongside the larger all-models or Sonnet cap. If you are hitting limits fast, switching routine work to Sonnet is the single biggest lever.
How do I avoid hitting Claude's usage limits?
Use Sonnet for routine work and reserve Opus for genuinely hard problems, keep conversations focused rather than letting context balloon, start fresh sessions for unrelated tasks, and lean on MCP tools that do work efficiently instead of pasting huge blobs of text into the prompt. Tools like the Sequential Thinking MCP help the model reason in fewer, tighter turns.
Is Claude Code free?
Claude Code is included with Claude Pro and Max subscriptions at no extra cost beyond the plan price, drawing from the same usage pool. There is no separate Claude Code fee for subscribers. You can also use Claude Code with pay-as-you-go API billing instead of a subscription. See our dedicated breakdown on whether Claude Code is free for the full picture.
Last updated: June 29, 2026 · Limits change frequently — always confirm exact figures against Anthropic's official Claude help center. Skiln tracks the Claude ecosystem daily.