Claude Pro usage limits are not a fixed number of tokens. Anthropic does not measure Pro usage in tokens but through a session limit that resets every five hours, plus a weekly limit, and it publishes no token count for either.
How quickly you reach that limit depends on the length of your messages and attachments, the length of the conversation and the model or feature you use. The usage limit is separate from the context window, which determines how many tokens a single conversation can hold.
Claude Code shares the same limits as the chat. You can see your current usage in the claude.ai settings. Max gives five times the usage of Pro per five hours and per week, Ultra ten times Pro’s weekly usage, and for predictable usage measured in tokens the API is an alternative.
How many tokens do you get with Claude Pro?
Claude Pro does not come with a fixed number of tokens, because Anthropic caps usage per session rather than in tokens. With a subscription, you buy access to the model within that limit, not a balance of a set number of tokens.
The only specific figure Anthropic gives is roughly 45 messages per five hours. That figure applies to short conversations with a less compute-intensive model, and according to Anthropic it also varies with Claude’s available capacity. Anthropic also promises that during peak hours Pro gives at least five times as much usage per session as the free version.
What matters most for the Pro limit is that longer input and longer replies use up more of your session. A short message in a new conversation costs little; a long document in an ongoing conversation costs a lot.
How do Claude Pro usage limits work?
Claude Pro usage limits work in two layers: a session limit that resets every five hours, and a weekly limit on top of that. Both apply to all your usage combined, across the chat, the apps and Claude Code.
What is the five-hour session limit?
The five-hour session limit is the amount of usage you have within a five-hour window. According to Anthropic, this session-based limit resets every five hours, after which you have your full allowance again.
If you hit the limit in the middle of a task, you wait for the reset or switch to another route, such as the API.
What is the weekly limit?
The weekly limit is a second cap on your total usage over a week, even if you stay under the limit in each session. Anthropic confirms both a five-hour session limit and a weekly limit for Pro and Max, but publishes no figure for the weekly limit.
If you use Claude intensively, especially with long documents or Claude Code, you will reach the weekly limit sooner than someone who asks short questions spread across the week.
What determines how quickly you run out?
Four factors determine how quickly you use up your limit: the length of your messages, the length of attached files, the length of the current conversation and the model or feature you use. These are the factors Anthropic itself names.
With every new message, Claude processes the entire conversation again, so the twentieth message in a long conversation costs more than the first. Starting a new conversation for each topic therefore saves usage.
A heavier model or a feature such as extended research costs more per message than a short question to a lighter model.
How is this different from the context window?
The context window differs from the usage limit in that it determines how much information a single conversation can hold, not how much you can do over a period. Anthropic calls the context window the length limit and the session and weekly caps the usage limit.
On paid plans, Claude in the chat has a context window of 1 million tokens for current models such as Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable 5.1. Older models such as Opus 4.8 and Sonnet 4.6 have 500,000 tokens, and other models 200,000 tokens, which Anthropic describes as roughly 500 pages of text. On paid plans with code execution, Claude automatically summarises earlier messages when a conversation gets too long, so it can usually continue. Only in rare cases, such as a very large first message, do you still run into the limit.
A warning about conversation length therefore does not mean your usage limit has run out. You can carry on in a new conversation, as long as your session and weekly limits allow it.
Does Claude Code count towards your Pro limit?
Yes, Claude Code counts towards your Pro limit. According to Anthropic, the limits are shared between Claude and Claude Code, and all activity in both tools counts against the same limits, including use from a code editor.
Claude Code often uses far more per task than a chat message, because it reads files, writes code and carries out several steps on its own. An afternoon of programming with Claude Code therefore usually drains a session faster than an afternoon of writing in the chat. Within Claude Code, the /usage command shows how much room you have left within your plan limit.
How can you see how much usage you have left?
You can see how much usage you have left in claude.ai through your account at the bottom left, then Settings and Usage. That page shows your usage in the current session and for the week as a percentage of your limit, plus the time remaining until the next reset. It does not show a token count.
Claude also warns you when you are approaching your limits.
How much more usage do you get with Claude Max?
With Claude Max, you get five times the usage of Pro, per five hours and per week. Anthropic does not give a fixed number of messages for it. Ultra, formerly Max 20x, gives ten times Pro’s weekly usage and has no five-hour limit for verified accounts.
Like Pro, Max has a weekly limit, plus a separate weekly limit for the Fable model. The monthly prices of Pro and Max are in the overview of what Claude costs.
When is the Claude API a better fit than a subscription?
The Claude API is a better fit than a subscription when you want to pay per token used rather than work within a limit. The same applies when you call Claude from your own software or an automation. A Pro subscription does not give access to the API: the API has its own account in the Claude Console and is billed separately. Max, Ultra and Team subscribers do get a monthly API credit, which Anthropic states in dollars: $100 on Max, $200 on Ultra and $20 or $100 per seat on Team.
Claude is my main everyday tool, both in claude.ai and through the API. I use the chat for writing and analysis, and the Claude API for workflows on my own VPS that run without intervention. With the API, the session limit disappears, but you pay for every token and rate limits apply per account.
For most professionals, Pro remains the logical starting point, provided you know where your usage goes. How to get more out of your Pro limit with projects, short conversations and the right model choice is covered in the Claude Course, with ways of working for your own tasks.
Frequently asked questions about Claude Pro
Once you know how the Claude Pro limit works, the next question is usually what you can do when it runs out.
Can you buy extra usage when your Claude Pro limit runs out?
Yes, when your Claude Pro limit runs out, you can keep working with usage credits, which Anthropic charges at the standard API rates. You switch them on in claude.ai through Settings and Usage, add a payment method and, if you like, set a monthly spending limit. If you took out Pro through the mobile app, you can only arrange usage credits in the web version of Claude.
How much does Claude Pro cost?
Claude Pro costs €18 a month on monthly billing or €15 a month on annual billing, which comes to €180 a year. According to Anthropic’s pricing page, these amounts exclude tax. For that price, you get more usage than on the free plan and access to Opus, while on Pro you pay for Fable separately with usage credits.
How do you make your Claude Pro limit last longer?
Your Claude Pro limit lasts longer if you bundle several questions into one message and do not upload the same file again in the same conversation. According to Anthropic, content you place in a project counts for less because it is cached. A lower effort level also stretches your usage, as higher levels use more tokens.
