Claude Usage Limits Explained: What Resets and When

Claude usage limits come in two layers on every paid plan: a session limit that resets every five hours, and a weekly limit that resets at a fixed day and time assigned to your account. Both are shared by everything you do with Claude, on the web, in Claude Desktop, on mobile, in Cowork and in Claude Code, so a morning of chat leaves less for the terminal. Anthropic does not publish a fixed number of messages. How fast you use the allowance depends on how long your messages and conversations are, the files you attach, the model, the effort level and the tools you turn on. You see where you stand under Settings > Usage, and when you hit a limit you can wait for the reset, turn on usage credits, or move to a larger plan. Everything below is as Anthropic documents it as of September 30, 2026.

Usage limits are not length limits

Two different limits get called “the limit”, and they have different fixes. Anthropic’s help article on how usage and length limits work↗ (opens in a new tab) draws the line: a usage limit controls how much you can use Claude over a period of time, across all your conversations, while a length limit is the context window, how much one conversation can hold. On paid plans the newest models support up to a 1M-token window and others 500K or 200K. Hit a length limit and you start a new conversation. Hit a usage limit and a new conversation will not help; you wait, pay, or upgrade.

The two do interact. A long conversation spends more of your usage on every message, because Claude re-reads it each time, and when Claude automatically summarizes a chat that is approaching its window, that summarizing also counts against your usage. The window itself, and what fills it, is covered in the AI context window.

The session limit: a five-hour window

The session limit is the one you meet most often. Anthropic’s plan articles say your session-based usage limit resets every five hours, and the Claude Code documentation describes it for Team and Enterprise seats as a rolling five-hour window. Settings > Usage shows how much of the current session you have used and how much time is left in it. When the session runs out you see the reset time in the message, and everything that shares your account, including Claude Code, waits for the same reset.

The weekly limit: a fixed reset time

On top of the session sits a weekly limit that applies across all models. According to the Pro plan article↗ (opens in a new tab), weekly limits reset at a fixed time each week that is assigned to your account; the reset day and time stay the same whatever day you start using Claude or when your subscription began, and you receive your full weekly allowance each cycle. The Max and Team plan articles say the same. Your next weekly reset is shown in Settings > Usage.

• Both clocks count at once. Every request draws on the session and the weekly allowance together, and the Claude Code error reference warns that one heavy burst, such as a large workflow fanout, can use up the weekly allowance before the session window resets.
• Some limits apply to one model family. Claude Code can report “You’ve hit your Opus limit” or “You’ve hit your Sonnet limit”, which apply only to requests to that family, so switching to a model outside it keeps you working. The session and weekly limits are shared across all models, so switching does not help there.
• Fable models have their own share. On Max plans and premium Team and seat-based Enterprise seats, you can use up to 50% of your weekly limit on Fable models, and they draw on it faster than other models. On Pro and standard seats, Fable runs on usage credits from the start.
• Anthropic keeps room to do more. Each plan article adds that it may limit usage in other ways, such as weekly and monthly caps or model and feature usage, to manage capacity.

How the plans compare

Anthropic states each plan’s allowance as a multiple of Pro’s per-session allowance, not as a number of messages or hours, and it does not publish how the weekly limits compare between plans:

• Pro: more usage per session than the Free plan, a five-hour session limit and a weekly limit across all models. Claude Code is included.
• Max 5x and Max 20x: five or 20 times Pro’s per-session allowance, with a weekly limit as well. The Max plan article↗ (opens in a new tab) calls it “5x or 20x the Pro plan’s usage allowance” in one place and “per-session usage allowance” in another; the second is the precise claim.
• Team Standard and Premium seats: 1.25 times and 6.25 times Pro’s per-session allowance, each with a weekly limit. Limits are per member, so one person running out does not stop anyone else. The Claude Team plan covers seat types.
• Enterprise: seat-based plans have per-seat limits and optional usage credits. Usage-based Enterprise plans have no per-seat limits; usage is billed on consumption at API rates.
• An API key in Claude Code: no plan limits at all. You pay per token, and the only ceilings are your organization’s rate limits and any spend limits you set.

What uses the allowance faster

The number of messages you get is not fixed because each message costs a different amount. Anthropic’s usage limit best practices↗ (opens in a new tab) lists what moves it:

• Message length and the size of attached files.
• The length of the current conversation, which is re-read with every message.
• Tools such as Research and web search, and connected apps, which Anthropic calls token-intensive.
• The model you choose and the effort level. Higher effort uses more tokens.
• Creating and using artifacts, and multi-step tasks such as running code, creating files or browsing websites.

Caching works in your favor. Content in a project is cached and counts less against your limits when it is reused, but caches expire after a period of inactivity, so the first message after a long break counts that content in full again.

Why limits can feel tighter than they used to

If the same work runs out sooner than it did, a few documented changes are the likely reasons. Weekly limits for Pro and Max arrived in late August 2025, on top of the five-hour session. Claude Code now starts on Opus 5.5 by default on Pro, Max, Team and Enterprise; before v2.1.280 it started Pro and Team Standard on Sonnet 5, and Anthropic’s own cost guidance is that Sonnet handles most coding tasks well and costs less than Opus. A promotion that let Pro and standard Team seats use Fable 5 inside their weekly limit ended on July 19, 2026, so Fable now bills to usage credits on those plans. And current models run with a 1M-token context window, so a conversation can grow far larger before anything trims it. None of these is a hidden cut to the allowance itself; each one makes the same allowance go faster.

Checking your usage

On the web or in Claude Desktop, open Settings > Usage. The Plan usage limits section at the top shows the current session, with the time remaining, and your weekly limits for all models and, where your plan includes them, Fable models. If usage credits are on, the same page shows your credit balance and this month’s spend. In Claude Code, run /usage for the same bars plus a breakdown of what used them; Claude Code usage limits covers that screen, the status line fields and the warning messages in detail.

When you hit a limit

1. Wait. The limit message says when the window resets. Claude Code can wait in the open session and continue the task on its own after the reset.
2. Turn on usage credits. Under Settings > Usage, usage credits let you keep working past the limit at standard API rates, billed separately from your subscription, with an optional monthly spend limit. The usage credits article↗ (opens in a new tab) notes that your session limits still reset every five hours as usual.
3. Use a limit reset if you have one. Anthropic occasionally gives eligible plans a reset that puts either the session or the weekly limit back to full. It is used from the Reset for free button on the web or in Claude Desktop, not from mobile or Claude Code, and once used it cannot be undone.
4. Upgrade, if you hit limits week after week. Pro moves to Max 5x, Max 5x to Max 20x, and on Team an admin can give a heavy user a Premium seat.

Staying under the limit

• Start a new conversation for a new topic, so each message does not carry an old one.
• Put documents you reuse in a project instead of attaching them again; cached content counts less.
• Batch related questions into one message, and give background up front instead of in follow-ups.
• Match the model and effort to the job. A lower effort level, or extended thinking switched off where your model allows it, saves tokens on routine work.
• Turn off connectors and web search in conversations that do not need them.
• Plan heavy work around the five-hour windows, which Anthropic itself suggests.
• In Claude Code, the biggest savings come from clearing between tasks and keeping the context lean. How to reduce Claude Code token usage has the settings and habits.

Do not lose the work when a limit lands mid-task

A limit stops the conversation, not the work, but if the only record of where things stand is that conversation, a reset hours or days away costs you the thread. Keep the state somewhere outside it. On fenbs, the task Claude is working on sits in In Progress with its note, which says what the problem is, and its plan, which says how it will be done. When you see a usage warning, ask Claude to rewrite the plan with what is finished and what is next, and to comment where it stopped. Whichever session picks it up, yours after the reset or a teammate’s on another plan, starts with fenbs_get_context and that task, not from nothing. fenbs does not see your Claude usage or your reset times, and it has no due dates, so it will not schedule around a limit for you; it only keeps the handover. Handing work between AI agents and people goes further.