Limits

Claude usage limits — and when they last changed

Claude usage limits are published by Anthropic but never dated — this page tracks the weekly limit, session limits and every change with a timestamp.

Session limit

5 hours, rolling

Starts with your first message and resets five hours later. It moves with your usage, not with the clock.

Weekly limit

Paid plans, fixed reset

Sits on top of the session limit and resets at a time assigned to your account, not at a rolling interval.

Published figures

None

Anthropic publishes ratios between plans, not amounts. No token or message count is official for any plan.

Last verified August 16, 2026

What are Claude's usage limits?

Claude usage limits combine a rolling five-hour session limit with a weekly limit on paid plans, plus a separately metered API. Anthropic documents the structure and the ratios between plans, but publishes no token or message count for Free, Pro, Max, Team or Enterprise.

There is no official token count or message count for Free, Pro, Max, Team or Enterprise, and any page quoting one has either estimated it or is repeating a figure that has since changed.

What Anthropic documents

  • A rolling five-hour session limit. It starts with your first message and resets five hours later, so it moves with your usage rather than with the clock.
  • A weekly limit on paid plans, on top of the session limit. It resets at a fixed time assigned to your account, not at a rolling interval.
  • Plan multipliers expressed as ratios: Pro relative to Free, Max 5x and Max 20x relative to Pro. The ratio is what is promised, not an amount.
  • One shared allowance. Claude Code draws on the same pool as the web, desktop and mobile apps.
  • Usage credits on paid plans, which let you continue past the included limit at standard API rates.
  • The API is metered separately, with per-minute request and token rate limits and usage tiers, and is not subject to the five-hour or weekly structure.

What Anthropic does not document

  • Any token count, for any plan.
  • Any message count. Message-based estimates depend on message length, model, attachments, tools and reasoning effort, all of which vary.
  • The baseline the multipliers are applied to. A ratio can stay constant while the amount it resolves to changes.
  • Advance notice of adjustments. Anthropic's terms reserve the right to adjust limits.

Where to see your own limits

Settings → Usage, inside Claude. It shows your current session, your remaining weekly allowance and your reset time for your account. That is the only authoritative figure for you, and it is the one number on this subject that is worth trusting.

What is the Claude weekly limit?

The weekly limit is an allowance on paid plans that sits on top of the five-hour session limit. It resets at a fixed time assigned to your account rather than at a rolling interval, and Claude Code draws on the same pool as the web, desktop and mobile apps.

Because the weekly limit and the session limit run at the same time, hitting one does not reset the other. A heavy Claude Code session draws down the same weekly allowance as chatting in the app.

How do Claude Pro and Max limits differ?

Max is described as a multiple of Pro — 5x and 20x — and Pro as a multiple of Free. The multiplier is the promise; the baseline it is applied to is not published, so a change can be real while the advertised ratio stays the same.

Why estimates disagree

Every published estimate is derived, not disclosed. Two people with the same plan can see very different amounts of work from the same allowance, because model choice, message length, tool calls, file reads and reasoning effort all draw differently on it. This is also why a change can be real and invisible at the same time: if the baseline moves, the ratio you pay for still holds.

When did Claude's limits last change?

See the full record in the limits changelog. We began monitoring Anthropic's public limits, pricing and plan pages on August 16, 2026; changes before that date are not in our archive.

More questions

How many messages do I get on Claude Pro?
Anthropic does not publish a number. Usage is metered as a share of a rolling five-hour session and against a weekly limit, and how far it goes depends on message length, model and features. Settings → Usage shows your actual remaining allowance.
When does my weekly limit reset?
At a fixed time assigned to your account, not at a rolling interval and not at a universal weekly boundary. Your reset time is shown in Settings → Usage.
Does Claude Code have its own separate limit?
No. Claude Code draws on the same plan allowance as the web, desktop and mobile apps.
What is the difference between the session limit and the weekly limit?
The session limit is a rolling five-hour window that moves with your usage. The weekly limit sits on top of it and resets on a fixed schedule. Waiting for a session reset does not restore weekly allowance.
Does the API use the same limits?
No. The API is metered per organisation with per-minute request and token rate limits and its own usage tiers, and is not subject to the five-hour session or the weekly cap.
Why do different websites give different numbers?
Because none of them are quoting a published figure. Every number in circulation is an estimate derived from observation or from API list prices, and estimates go stale whenever the underlying allowance changes.

How this page is made

This page describes what Anthropic publishes on its own documentation and support pages. We do not publish token or message counts, because Anthropic does not publish them and any figure we printed would be an estimate presented as a fact. Where we describe a change, we link to the source and date it. Last verified: August 16, 2026.