Claude limits explained: how do the 5-hour window, weekly limit and credits work?

Claude has a 5-hour and a weekly limit, shared across chat and Claude Code. How they work per plan, what usage credits are and what to do at the limit

Claude limits explained: how do the 5-hour window, weekly limit and credits work?

Summary

  • Claude runs two limits at once: a session limit that fully resets five hours after your first message, and a weekly limit that resets at a fixed time per account
  • Anthropic publishes no fixed message count: how much you can do depends on the model, conversation length, files, tools and effort
  • Max gives 5x Pro usage per 5 hours and per week; Ultra (formerly Max 20x) gives 10x Pro per week and has no 5-hour limit for verified accounts
  • Claude on web, desktop, mobile and Claude Code draw from the same pool; heavy chatting in the morning leaves less room to build in the afternoon
  • Usage credits let you keep working after the limit; the API credits included with Max, Ultra and Team do not cover Claude Code or the Claude apps

You are halfway through a report, and Claude tells you that you have hit your limit. When can you continue? And why did it happen now, when yesterday you could work all afternoon?

The answer lies in two limits that run at the same time, a shared pool across chat and Claude Code, and a few switches that let you buy extra usage. Anthropic explains all this across a dozen help pages, without a single concrete number per plan. This article puts everything in one place, as of 19 November 2026, with a source for every claim.

How do Claude’s usage limits work?

On paid plans, Claude runs two limits at once: a session limit that fully resets five hours after your first message, and a weekly limit that resets at a fixed time per account. There is no fixed message count. How much you can do depends on the model, the length of your conversations, your files, the tools and the effort level.

Anthropic puts it like this in its Help Center: “Paid plans have two types of usage limits: a session limit that resets every five hours, and weekly limits that reset once every week.” The pricing page adds that there is “no fixed message count”.

That last point matters most. Lists with a fixed number of messages per five hours still circulate, but they appear on no current Anthropic page. A short question to the lightest model costs a fraction of a long analysis with attachments on Opus.

Besides the usage limit there is also a length limit: the context window, the amount of text a language model can process in one conversation. On paid plans, Anthropic says this goes up to 1 million tokens for the newest models, and 500,000 or 200,000 for others. These are two different limits, even if they feel the same to the user.

ClickForest guides SMEs in Flanders through the AI implementation of Claude, and anyone rolling Claude out to a team needs to know how these limits count.

When does Claude’s limit reset?

The session limit resets five hours after your first message, not at a fixed time on the clock. The weekly limit resets on a fixed day and time assigned to your account, regardless of when you start or subscribe. You can see both moments in Claude under Settings > Usage.

The 5-hour window

The window starts with the first message you send. According to Anthropic: “The period starts when you send your first message, and your limit resets in full five hours later.” Send your first question at 9 am, and your session limit is full again at 2 pm. Under “Current session” in Settings > Usage you can see how much you have used and how long the window still runs.

The weekly limit

The weekly limit applies across all models. Max and Ultra also have a separate weekly limit for Fable, Anthropic’s model with the highest price per token. The reset moment is fixed per account: “Your reset day and time stay the same regardless of when you start using Claude or when your subscription begins.”

The weekly limit is fairly new. Anthropic announced it on 28 July 2025 for Pro and Max, rolling out in late August 2025, and estimated at the time that it would apply to less than 5% of subscribers (Simon Willison). The trigger was a small group running Claude Code day and night.

A verified Ultra account has no 5-hour window. Only the weekly limit applies there.

How much usage do you get per plan: Pro, Max, Ultra, Team and Enterprise?

Anthropic expresses everything relative to Pro. Max gives 5x Pro usage per five hours and per week. Ultra gives 10x Pro per week, with no 5-hour limit for verified accounts. A standard Team seat gets 1.25x and a premium seat 6.25x the Pro per-session allowance. Enterprise charges usage at API rates on top of the seat price.

Plan Limit per 5 hours Weekly limit Price (USD, excluding tax)
Free Limited Limited $0
Pro At least 5x Free per session Yes, across all models $17 per month billed annually ($200 up front), or $20 per month
Max 5x Pro 5x Pro, plus a separate Fable limit $100 per month
Ultra None, for verified accounts 10x Pro, plus a separate Fable limit $200 per month, monthly only
Team, standard seat 1.25x Pro per session Yes, per member $20 per seat billed annually, or $25 per month
Team, premium seat 6.25x Pro per session Yes, per member $100 per seat billed annually, or $125 per month
Enterprise Not published Not published $20 per seat per month (billed annually, 20-seat minimum) plus usage at API rates

Source: claude.com/pricing and the help pages on Max, Ultra and Team, accessed on 19 November 2026. The pricing page shows the visitor’s currency: from the United States you see dollars, from Belgium euros. Tax is not included, and prices in the mobile app may differ. What Claude Code itself costs and can do is covered in our explainer on Claude Code.

Do not read the ratios as one row. The Team figures apply per session, Ultra’s 10x per week, and Max’s 5x to both. On Team, limits also apply per member: “Usage limits on Team plans are per-member, rather than applied to the team as a whole.”

The confusion around Ultra has a history. Until 30 October 2026 the plan was called Max 20x, and some new users assumed that 20x also applied to the week. The product lead explained where that went wrong:

“When we launched our top plan in April 2025, it had a single limit: 5 hours. It gave you 20x Pro’s limit, so we called it Max 20x. When we added weekly limits to Max 20x, we set them to 10x Pro’s (the 5-hour limit stayed at 20x). I get that some new users assumed 20x applied to both limits. We could have done better here, and I’m sorry.”

— Mike Overell, product lead for Claude subscriptions at Anthropic, on X (30 October 2026)

Fable is included in Max and Ultra for up to half of your weekly limit, and uses it up faster than the other models. On Pro you pay for Fable with usage credits from the first message.

Do Claude, Claude Code and Cowork share the same limit?

Yes. Everything you do in Claude, on the web, in the desktop app, on your phone and in Claude Code, counts toward the same limit. Anthropic says so literally: “These limits are shared across Claude and Claude Code, meaning all activity in both tools counts against the same limits.” For Team, the documentation states that Cowork is in the same pool.

In practice this means: if you work through a long strategy document with Claude in the morning, you have less room in the afternoon to build a tool in Claude Code, and the same goes for the desktop app. On Team, the documentation also puts Cowork in the same pool. For Pro, Max and Ultra no help page mentions Cowork explicitly, but the pricing page says that all your activity “on web, desktop, mobile, and Claude Code” draws from the same pool.

One exception often catches people out. If you have set an Anthropic API key on your computer (ANTHROPIC_API_KEY), Claude Code no longer runs on your subscription. It uses that key instead, and you pay per token rather than from your included usage. There is then no limit window anymore, but no cap on the bill either.

ClickForest knows the problem first-hand. Anyone working intensively with Claude hits the limit without seeing it coming, because usage is tucked away in the settings. That is why ClickForest built the Claude Token Tracker with Claude Code, a browser and editor extension that shows your session limit and weekly limit live, with a warning before you hit the wall.

Why do you hit your Claude limit so fast?

Your usage grows with everything Claude has to process: long messages, large attachments, a conversation that keeps growing, tools such as web search and Research, an expensive model such as Opus and a high effort level. Agents that carry out steps on their own use it up quickly. Opus costs several times more per turn than Sonnet, and Sonnet more than Haiku.

Anthropic lists the factors itself in its usage limit best practices: message length, file attachment size, current conversation length, tools, model choice, effort level, artifacts and multi-step tasks, such as running code or browsing websites.

Three mechanisms explain most surprises:

  • The whole conversation is sent again every time. According to the Claude Code documentation, Claude sends your full conversation with every request. Message 40 in a long conversation therefore costs much more than message 1.
  • A break makes the cache go cold. Content in projects is cached and counts less when reused. If you come back after a long break, your first message counts that content in full again.
  • Agents work in many steps. According to Anthropic, agent teams in Claude Code use approximately 7x more tokens than a standard session when teammates run in plan mode.

A more expensive plan gives you more of the same budget; as Ethan Mollick puts it, “The more expensive plans are mostly buying you more hours of AI labor, not smarter AI.” The Wharton professor added a warning in his guide of July 2026, about the entry plans of both Claude and ChatGPT:

“One warning: the $20 tiers include real but limited agent usage, and agents burn through those limits quickly.”

— Ethan Mollick, professor at the Wharton School, on his blog One Useful Thing (23 July 2026)

He wrote the same about Cowork in February 2026: “Cowork is still a research preview, meaning it’s early and will eat through your usage limits fast” (One Useful Thing). That the limit runs out faster when you use agents simply follows from how the system counts.

Want your team to work with Claude in a way that stops the limit from being a brake? ClickForest offers tailored AI training for SME teams, built around their own tasks and ways of working. Get in touch for a conversation.

What are usage credits and how do you buy extra usage?

Usage credits let you keep working after you reach your limit. They exist for Pro, Max and Ultra and are billed at standard API rates. You turn them on under Settings > Usage, set a monthly spend limit and buy a balance. On Team, only Owners and Primary Owners decide on usage credits.

Anthropic used to call this “extra usage”, a term you still come across in older documentation. How it works according to the Help Center:

  1. Turn it on. Open Settings > Usage and switch on usage credits. Add a payment method.
  2. Set a cap. With “Adjust limit” you set a monthly maximum. “Set to unlimited” is possible too, but rarely wise for an SME.
  3. Buy a balance. You buy an amount and can optionally turn on auto-reload. That only works on the web.
  4. Keep working. Your usage continues until you reach your monthly cap, or until your balance runs out if auto-reload is off.

Two conditions matter. Usage credits are non-refundable, including if you cancel your subscription. And individual credits purchased on or after 7 December 2026 expire 12 months after purchase; earlier purchases do not expire.

On Team, Owners and Primary Owners turn on usage credits in Organization settings > Usage, for the whole organization or, where available, for specific members with their own monthly limit (Help Center). A team member without billing rights who types /usage-credits in Claude Code sends a request to those admins. Team credits always expire after 12 months.

What are the API credits included with Max, Ultra and Team, and what do they cover?

Since 7 October 2026, Max and Team subscribers get monthly API credits, and Ultra since the plan exists: $100 on Max, $200 on Ultra, and on Team $20 per standard seat and $100 per premium seat, pooled up to $500. They are for your own applications through the Claude API, and explicitly not for Claude Code or extra usage in the Claude apps.

This is where many people get stuck, because the two kinds of credits sound alike and do very different things.

Feature Usage credits API credits with your plan
What it is A balance you buy yourself Included monthly with Max, Ultra and Team
What for Keep working in Claude and Claude Code after your limit Claude API, Agent SDK, Managed Agents and the Playground
Not for Free, which has no usage credits Claude Code and extra usage in the Claude apps
Expiry After 12 months for purchases from 7 December 2026; Team always after 12 months At the end of each billing cycle, with no rollover
Who Pro, Max and Ultra; on Team the Owners and Primary Owners Max, Ultra and Team, claimable after 7 days on the plan

Source: Anthropic, API credits for Max, Ultra, and Team plans and the release notes, accessed on 19 November 2026.

Anthropic gives its own example: a team with three standard and two premium seats receives $260 a month in API credits. The documentation is also explicit about what the credits do not do: “Credits don’t change your usage limits in Claude or Claude Code.”

For an SME, these credits are mainly interesting for experimenting with your own integration, a small AI agent or an automation through the API, without opening a separate billing account right away. If you want to build such integrations, our explainer on MCP shows how Claude talks to other systems.

For heavy users, the subscription remains good value compared with paying per token. Developer Simon Willison calculated how much his subscriptions gave him in tokens:

“If you are a heavy user of coding agents these plans are a fantastic deal.”

— Simon Willison, independent developer, on his blog (27 May 2026)

His calculation: according to the tool ccusage, he used $2,180.16 worth of tokens at API prices in 30 days, for $200 in subscriptions with Anthropic and OpenAI combined. That is one heavy user, and the figure is an estimate by ccusage.

What do you do when you hit Claude’s limit?

You have four options: wait until the window or the week resets, turn on usage credits, upgrade to a bigger plan, or move bulk work to the API. Switching models does not restore access, because the limit applies across all models. Working more economically pays off mainly up front.

Here is how to approach it:

  1. Check where you stand. In Claude, Settings > Usage shows your current session, your weekly usage and your next reset. In Claude Code, /usage shows your plan usage with a breakdown per skill, subagent, plugin and MCP server.
  2. Choose between waiting and buying. If the window refills within the hour, wait. From version 2.1.234, Claude Code can continue an interrupted task automatically after the reset. If it is urgent, turn on usage credits with a low monthly cap.
  3. Start a new conversation. Anthropic recommends starting a new conversation if you are approaching your limit in a long chat. In Claude Code, /clear wipes the conversation, and that costs nothing.
  4. Choose the model deliberately. Use Sonnet or Haiku for routine work. Anthropic puts it bluntly: “Spending Opus on routine work is the fastest way to drain a daily limit.”
  5. Lower the effort and turn off tools. A lower effort level uses fewer tokens, and tools and connectors are “token-intensive” according to Anthropic. Turn off what you do not need.
  6. Batch and reuse. Put similar questions in one message, and keep documents you use often in a project, where they are cached. Whatever you capture in a fixed workflow, as with Claude Skills, you do not have to explain again every time.
  7. Review your plan. If you hit the weekly limit several times a week, Max (or a premium seat on Team) makes more sense than buying credits again and again.

A good instruction also costs less than three half-baked ones. How to phrase a question sharply in one go is covered in our article on writing better prompts.

What changed in Claude’s limits in 2026?

Claude’s limits changed several times in 2026. Anthropic doubled the 5-hour limits in Claude Code in May, raised the 5-hour limits on most plans in September, added API credits in October and renamed Max 20x to Ultra at the end of October. A figure from earlier this year may therefore well be outdated.

Date Change
16 January 2026 Claude Code is included with every standard seat on the Team plan
6 May 2026 The 5-hour limits in Claude Code double for Pro, Max, Team and seat-based Enterprise
13 May to 13 September 2026 Temporarily 50% higher weekly limits in Claude Code
14 September 2026 Weekly limits in Claude Code are permanently 25% higher than before that promotion
22 September 2026 Higher 5-hour limits on Pro, Max, Team and seat-based Enterprise, plus a limit reset you can save
7 October 2026 Monthly API credits for Max and Team
30 October 2026 Max 5x becomes Max, Max 20x becomes Ultra, with no price change; no 5-hour limit for verified Ultra accounts

Source: Anthropic, higher limits in Claude Code, the temporary Claude Code promotion, the launch of Opus 5.5, the release notes and the announcement on X.

The September change shows how one figure has two sides. Compared with before the promotion, weekly limits in Claude Code were 25% higher. Compared with the promotion period, it was a cut, and Anthropic said so itself on 29 August 2026: “this works out to a 17% reduction in weekly limits on Claude Code” (ClaudeDevs on X). Both are true.

Anthropic also reserves the right to adjust limits: “We may adjust usage limits at our discretion to manage capacity and ensure fair access to all users.” How large the increase of 22 September was, it has not published.

Which Claude plan suits an SME team?

For an SME team, the Team plan with standard seats is the logical starting point: every member gets their own limit, Claude Code is included, and the admin decides who may use usage credits. Give premium seats only to the heavy users. For a single intensive user without a team, Max or Ultra is simpler.

A rule of thumb per profile:

  • An occasional question or text: Pro, or a standard Team seat. The limit rarely comes into play.
  • Long documents, analyses and projects every day: a standard seat with usage credits as a buffer, under a cap the admin keeps an eye on.
  • Builds with Claude Code or lets agents work: a premium seat on Team, or Max as an individual. Whoever hits the weekly limit daily looks at Ultra.
  • Builds own applications on the API: use the included API credits to test, and pay per token after that.

In ClickForest’s view, the biggest gains come from ways of working: the right model per task, short conversations per topic, projects with fixed files and agreements on who uses what. A team that learns this gets more out of a standard seat than an untrained team out of a premium seat. How to bring that knowledge into the team is covered in our guide to an AI workshop for your team. That it also matters legally is explained in our article on AI literacy in your company.

Still weighing Claude against other assistants? Our comparison of AI tools and our practical guide to ChatGPT, Claude and Perplexity can help.

ClickForest guides SMEs in Mechelen, Antwerp and the rest of Flanders through a Claude rollout, from plan choice to usage agreements. If you would rather have a tool built than use up your own limit, our approach to building with AI shows how that works.

What should you remember about Claude’s limits?

Claude has a session limit that resets five hours after your first message and a weekly limit that resets at a fixed moment per account. There is no fixed message count. Everything you do in Claude, on your phone and in Claude Code comes from the same pool. Usage credits let you keep working; the API credits with Max, Ultra and Team are for your own applications through the API.

We hit the limit often enough ourselves to build an extension for it. The limit feels like a wall until you know how it counts, and after that like a budget you divide deliberately.

ClickForest helps SMEs in Flanders roll out Claude company-wide through AI implementation and tailored AI training. Want to know which plan and way of working suits your team? Get in touch.

Grow with AI, practically and without unnecessary complexity

Ready to use AI to save time and accelerate your marketing? Discover our AI services

Discuss your challenge directly with Frederiek: Book a free strategy call or send us a message

Prefer email? Send your question to frederiek@clickforest.com or call +32 473 84 66 27

Strategy without action remains theory. Let's take your next step together.

FAQ

Frequently asked questions

There is no fixed number. Anthropic's pricing page literally says there is no fixed message count: how much you can do depends on the length of your conversations, the model and the features you use. Lists with a fixed number of messages per window are outdated.

Yes, it is the same plan under a new name since 30 October 2026, at the same $200 a month. The weekly budget is 10x Pro, not 20x: that 20x only applied to the old 5-hour window. Verified Ultra accounts no longer have a 5-hour limit.

Yes. Team plan limits apply per member, not to the team as a whole. A standard seat gets 1.25x and a premium seat 6.25x the Pro plan's per-session allowance. ClickForest helps SMEs in Flanders choose between standard and premium seats as part of an AI implementation.

No. According to Anthropic's documentation, those monthly API credits explicitly do not cover Claude Code or extra usage in the Claude apps. They are meant for your own applications through the Claude API, the Agent SDK, Managed Agents and the Playground, and they expire every billing cycle.

That depends on when you bought them. Individual usage credits purchased on or after 7 December 2026 expire 12 months after purchase; earlier purchases do not expire. On Team they always expire after 12 months, and usage credits are never refunded.

Switching with /model does not restore access, because the session and weekly limits apply across all models. Choosing a lighter model up front stretches your budget, since Opus costs several times more per turn than Sonnet.

ClickForest guides SMEs in Flanders through a company-wide Claude rollout: plan choice, usage agreements and tailored training for the team. In ClickForest's view, the biggest gains come from ways of working that keep usage low.

Sources and references

Limits and plans: Anthropic Help Center:

Usage credits and API credits:

Claude Code:

Anthropic announcements:

Sources of quotes:

Ready to use AI practically in your business? Discover AI for growth
Ready to grow

One call and it becomes clear.

No noise, short lines, no annual contracts. Book a no-obligation intro call with Frederiek.