Claude Code usage limits: the two windows, the plans, and what to do when a limit stops you

Claude Code is Anthropic's coding tool. You type a request in a terminal, the text window where you type commands, and an AI model reads files, runs commands and edits code for you. On a Pro, Max, Team or Enterprise subscription, that work counts against a usage limit: an allowance of work the plan includes. The work is done in tokens, the pieces of text the model reads and writes, and Anthropic states the allowance as a multiple of another plan's allowance, with no token count attached. When the allowance runs out, Claude Code stops, and a message names the time it will start again. This guide is about what that message means and what to do.

Every fact here comes from Anthropic's own documentation, support articles and pricing page, read on 4 October 2026, plus one named news report, BleepingComputer's article of 29 August 2026 on a change to the weekly limits. Anthropic's figures are Anthropic's own statements about its own product, and the pages say so each time. Reveneau, which wrote this guide, is an AI software development consultancy and is independent of Anthropic. All of Reveneau's code is written by AI and every change must pass an eval suite, a set of automated tests written from the specification, before release, so a usage limit is a running constraint on every Reveneau build. That is the reason the guide exists, and it is also the reason the facts are dated: limit rules change, and the page on the dated log of limit changes is where this guide records what changed and when.

The guide in order

The question	Where it is answered
When does a limit reset, and what does the message mean?	The five-hour window and the weekly window
What uses up the allowance, and why does coding use more than chat?	What counts against a Claude Code usage limit
Which plan includes what, and at what listed price?	Pro, Max, Team and Enterprise: what each plan includes
How do I keep working after the limit by paying?	Usage credits after the limit
Can Claude Code wait and continue by itself?	Waiting for a reset and continuing automatically
What has changed, and when?	A dated log of Claude Code limit changes
Should I stay on a subscription or pay per token?	Subscription or API key for one developer

The two windows

Anthropic's support articles on the Pro plan and the Max plan, read on 4 October 2026, use the same two sentences for both plans. "Your session-based usage limit will reset every five hours." And: "Weekly limits reset at a fixed time each week that is assigned to your account. Your reset day and time stay the same regardless of when you start using Claude or when your subscription begins, and you receive your full weekly allowance each cycle" [2][3]. The pricing page says every plan has limits "that reset on a rolling five-hour session window" and that paid plans add weekly limits [4]. Anthropic's word for the session window is rolling, and its word for the weekly reset is fixed, and this guide uses only those words.

Both windows count the same work at the same time. Anthropic's error reference says: "Usage counts against the session and weekly allowances at the same time. A single burst of heavy activity, such as a large workflow fanout, can exhaust the weekly allowance before the session window resets" [5]. A workflow fanout is a run in which Claude Code starts many agents at once. So the weekly bar in /usage is the one to watch: a session window gives its capacity back within hours, and the weekly window gives it back on one day each week.

When a window runs out, the error reference lists the messages, each with a reset time, such as You've hit your session limit · resets 3:45pm and You've hit your weekly limit · resets Mon 12:00am, and says Claude Code blocks further requests until that time [5]. Before that, Claude Code can warn you, with a message such as You've used 85% of your session limit · resets 3:45pm [5]. The page on the five-hour window and the weekly window covers the two meanings of the word session, the plan bars in /usage, and the prompt cache lifetime of one hour that applies inside plan usage.

What counts against the allowance

Everything on the subscription. Anthropic's support article on using Claude Code with a Pro or Max plan says both plans "offer usage limits that are shared across Claude and Claude Code, meaning all activity in both tools counts against the same usage limits", and that usage in the IDE extensions "counts toward the same usage limits shared across Claude and Claude Code" [1]. On Team and Enterprise plans, Anthropic's cost documentation says each member's Claude Code usage "draws from a per-seat allowance that resets on a rolling five-hour window and a weekly window", and that the allowance is shared with Claude chat and Cowork [6]. A seat is one member's place on the plan.

Coding uses the allowance faster than chat, and Anthropic says why: each Claude Code turn includes file contents, tool calls and multi-step reasoning, so, in its words, "one debugging session can consume more than a day of chat" [6]. The Enterprise consumption guide, Anthropic's planning document for administrators, gives the same list for a coding session: system prompts, file context, tool calls and multi-turn reasoning [12]. The page on what counts against a Claude Code usage limit lists the activities one by one, explains the Opus and Sonnet limits that a model switch can get past, and gives Anthropic's own tips for staying inside the allowance.

Two sister guides take the next step. Reduce Claude Code token usage covers the habits and settings that make each turn smaller, which makes an allowance last longer. Claude Code prompt caching explains the store of already-processed request text that Anthropic's service reuses, why a long session still draws usage for its whole history, and why the cache lifetime is one hour inside plan usage and five minutes on usage credits [6].

The plans

Anthropic's pricing page, read on 4 October 2026, lists Claude Code as included on every paid plan and absent from the Free plan [4]. The prices as listed that day, all of which Anthropic says "are subject to change at Anthropic's discretion" [4]:

• Pro: $20 per month billed monthly, or $17 per month with an annual subscription of $200 billed up front [4].
• Max 5x: $100 per month. Max 20x: $200 per month. The Max plan article says these are web prices, that mobile pricing may vary, and that Max 5x "includes five times the Pro plan's per-session usage allowance" while Max 20x includes 20 times it [3].
• Team: Standard seats at $20 per seat per month billed annually, or $25 monthly; Premium seats at $100 billed annually, or $125 monthly, with five times the usage of a Standard seat [4].
• Enterprise: $20 per seat per month billed annually, plus usage at API rates [4].

The pricing page states the allowance only as these multiples: Pro "gives you at least 5x more usage per 5-hour session than Free", and Max gives five or 20 times Pro [4]. Across every source read for this guide, Anthropic gives no conversion from an allowance to a number of tokens, dollars or messages, and this guide gives none either. The page on what each plan includes sets the plans in one table and says which to pick.

Past the limit: usage credits

Usage credits are paid use after the included limit. Anthropic's support article, read on 4 October 2026, says they let a Pro, Max 5x or Max 20x subscriber "switch to consumption-based pricing at standard API rates" when the included limit is reached, that they "are charged separately from your paid Claude subscription", and that the session limits "reset every five hours as usual" while credits are on [7]. You turn them on in Settings > Usage on claude.ai, add a payment method, set a monthly spend limit or choose unlimited, and prepay funds, with a daily redemption limit of $2000 [7]. Inside Claude Code, /usage-credits opens that settings page for a Pro or Max subscriber, and /usage shows a usage-credits row with this month's spend against the limit while credits are on [6].

One consequence is easy to miss. Anthropic's cost documentation says the prompt cache lifetime "is an hour on a subscription and drops to five minutes once you're drawing on usage credits", and that the promptCacheTtl setting keeps the hour [6]. The page on usage credits after the limit covers the setting, the version it requires, and a table of credits off against credits on. Team and Enterprise plans have a variant that an administrator controls, with spend limits for the organisation, a group or one person, and the guide to Claude Code costs for teams owns that subject together with the spend-limit messages an administrator sees.

Past the limit: waiting, by hand or automatically

The first option is to wait for the time in the message. From Claude Code v2.1.234, Anthropic's interactive mode documentation says, Claude Code can do the waiting: "When a claude.ai usage limit stops Claude mid-task, Claude Code waits in the open session and continues the task on its own after the limit resets." Automatic continue is on by default in interactive sessions signed in with a claude.ai subscription [8].

The conditions are exact, and the page on waiting for a reset and continuing automatically gives all of them. In short: Claude Code starts the wait on its own only when the reset is within 24 hours, so a weekly limit that resets days out needs you to pick Wait here, then continue automatically from /rate-limit-options; it restarts the wait at most twice in a row after repeated limits; it does not offer the wait in a -p run, on an API key, a cloud provider or usage-based billing, or through an LLM gateway without a saved claude.ai login; and the setting autoContinueAtUsageLimit turns it off, with the key=value form of /config able to turn it off and unable to turn it back on, because, in Anthropic's words, "the setting grants unattended execution" [8]. The Desktop app has a separate checkbox on its session-limit card [5].

The dated log

Anthropic publishes limit changes in its release notes, and its support articles state the rule in force on the day you read them. The page on a dated log of Claude Code limit changes keeps that history from the release notes, the support articles and named news reports, and says that any change not listed was not found in those sources on 4 October 2026.

The entries from Anthropic's release notes, read on 4 October 2026: Premium seats with Claude Code and extra usage for Team and Enterprise plans on 20 August 2025; Claude Code access on every Team Standard seat on 16 January 2026; and self-serve Enterprise plans with a single seat type that includes Claude Code on 12 February 2026 [9]. The entry from the news report: BleepingComputer reported on 29 August 2026 that a temporary 50 percent increase in Claude Code's weekly limits ran through 13 September 2026, and that from 14 September 2026 a permanent 25 percent increase over the pre-promotion baseline applies to Pro, Max, Team and seat-based Enterprise plans. The report's illustration is 100 before, 150 during the promotion and 125 after, and it quotes Anthropic's clarification: "Compared to today, this works out to a 17% reduction in weekly limits on Claude Code" [10]. The release notes hold no entry for that change, so the log says it rests on the report alone.

The choice: subscription or API key

An API key is the credential of a Claude Console account, Anthropic's account site for developers who pay per token. Anthropic's Pro plan article says the Pro plan "does not include API usage through the Claude Console", so a developer who wants both pays for Console usage separately [2]. On an API key, "Claude Code charges by API token consumption" at the list prices on Anthropic's pricing page, for example $2 per million input tokens and $10 per million output tokens on Sonnet 5.5 as read on 4 October 2026, with rate limits that vary by the Start, Build and Scale usage tiers in place of the plan windows [6][11]. The automatic wait is offered to a subscription only [8].

The deciding figure is your own. Anthropic's cost documentation says the Session block in /usage "shows API token usage and is intended for API users" and that Claude Code "computes the dollar figure locally from token counts at list price" [6]. Read it for a month on your subscription, and you know what the same work would cost per token. Anthropic's own average across enterprise deployments, which Anthropic puts at $13 per developer per active day, is a figure about other people's deployments, and Anthropic itself says costs "vary widely" [6]. The page on subscription or API key for one developer works an invented example at list prices and gives a table by what matters to one developer. A team making the same choice for many developers has its own page in the costs-for-teams guide, and a developer running agents, scripts or scheduled jobs, where the allowance empties fastest, has the guide to Claude Code agent costs.

Reveneau's position

Reveneau's recommendation from these facts is short. Plan around the weekly window, because it is the one that can cost you days. Leave automatic continue on for session limits and off on any machine where a task continuing without a person present would be a problem. Turn on usage credits with a monthly spend limit before you need them, so that the first limit message offers you a choice. Read the Session block in /usage for a month before deciding on an API key. And note the date of every page in this guide, because the rules were read on 4 October 2026 and the dated log is where any later change will appear.