Claude Usage Limits Explained: Pro, Max and Claude Code

Legacies is a software and web studio from Romania, founded by Horia Stan and Alexandru Talnaci, and we run Claude and Claude Code every working day. We hit the same walls you do, including the weekly cap that shows up on a Thursday.

Here is the short version. Paid Claude plans give you a usage budget that refills every five hours, plus a weekly cap on top, and Claude Code draws from the same budget as the chat app. Anthropic does not publish a fixed number of messages, because usage depends on how much context each request carries, which model you use and how hard it thinks. The API is a separate system with per-minute rate limits and monthly spend caps.

Quick answer

• Pro costs $20 a month ($17 a month billed yearly) and includes Claude Code. Max is $100 (5x Pro usage) or $200 (20x Pro usage). Free does not include Claude Code.
• Two windows apply: a session limit that resets every five hours and a weekly limit across all models that resets at a fixed time assigned to your account.
• Claude Code and the Claude app share one budget. A long coding session eats into the same allowance as your chats.
• The API is different: rate limits per minute (requests, input tokens, output tokens) and a monthly spend cap per usage tier, billed per token.

Claude plans and limits at a glance

All prices below come from the official Claude pricing page and the Max plan help article, checked at the time of writing.

Notice what is missing: a message count. Anthropic describes limits as multiples, because a one-line question and a request carrying a whole codebase cost very different amounts.

How Claude usage limits actually work

The five-hour session limit

Every paid plan has a session-based limit that resets every five hours, according to the Pro plan help article. Think of it as a bucket. Heavy work empties it faster, light work slower, and five hours later it is full again.

The weekly limit

On top of the session window, Pro and Max plans have a weekly limit that applies across all models. Anthropic says it resets at a fixed time each week that is assigned to your account, so it is not necessarily Monday morning. You can see your exact reset time in Settings > Usage on claude.ai.

The weekly cap is the one that surprises people: you can stay under every five-hour window and still run out on day five.

Claude Code uses the same budget

Anthropic states that Pro and Max usage limits are shared across Claude and Claude Code. There is no separate coding allowance. If you spend the morning on a long research chat, you start your coding session with less in the tank.

One recent change helps coders. After a temporary promotion that ran from May to September 2026, Anthropic made a permanent adjustment: from September 14, 2026, weekly limits in Claude Code are 25% higher than before the promotion for Pro, Max, Team and seat-based Enterprise plans. The five-hour limit did not change.

Model-specific limits

The Claude Code cost docs explain that session and weekly limits are shared across all models, so switching with /model does not help. A model-specific message, like "You've hit your Opus limit", is different: switching to another model family keeps you working.

The top model, Claude Fable 5.1, is a special case: depending on your plan, its usage can bill to usage credits instead of your included limits.

API rate limits are a separate system

With an API key, subscription limits do not apply. You pay per token and face two other limits, described in the API rate limits docs.

Spend caps by tier: Start allows $500 a month, Build $1,000 and Scale $200,000. When you reach the cap, requests return HTTP 429 until the first day of the next month, unless you request a higher limit.

Rate limits per model, measured in requests per minute (RPM), input tokens per minute (ITPM) and output tokens per minute (OTPM). On the Start tier, Claude Opus 5.5 and Sonnet 5.5 each allow 1,000 RPM, 2,000,000 ITPM and 400,000 OTPM. Exceed one and you get a 429 with a retry-after header.

For most current models, tokens read from the prompt cache do not count toward ITPM, so a well-cached app gets far more throughput than the headline number.

Current API prices per million tokens, from the models overview:

For a sense of scale, Anthropic reports that across enterprise deployments, Claude Code averages around $13 per developer per active day at API rates, and stays below $30 a day for 90% of users.

What eats your quota

Anthropic lists the factors in its usage limit best practices: message length, attachment size, conversation length, tool use like web search and Research, model choice, effort level, artifacts and multi-step tasks. In Claude Code, these are the ones that bite hardest.

• Long conversations. Claude Code sends the whole conversation with every request, plus every tool result. Caching makes that history cheaper, not free.
• Cache misses after a break. On a subscription the prompt cache lives for an hour. Come back after lunch and your first message reprocesses the full context.
• Big models and high effort. Opus 5.5 is the default model in Claude Code for Pro, Max, Team and API users, per the model configuration docs. Thinking tokens bill as output, so high effort on a simple task burns budget for nothing.
• Parallel agents. Subagents, scheduled /loop tasks and routines all send their own requests. Anthropic says agent teams use about 7x more tokens than a standard session when teammates run in plan mode.

How to tell you are about to hit the limit

Three places tell you where you stand:

• Settings > Usage on claude.ai shows progress bars for the current session and the weekly limit, plus when each resets.
• /usage in Claude Code shows the same bars plus a breakdown by skills, subagents, plugins and MCP servers, and flags habits like long context or cache misses when one passes 10% of recent usage.
• API response headers such as anthropic-ratelimit-input-tokens-remaining tell your app how close it is to a rate limit before it fails.

We check /usage before any long task. If the weekly bar is past 70% on a Wednesday, we plan differently.

Practical ways to work within Claude limits

These are the habits that changed our weekly numbers the most. We go much deeper, with configs and reasoning, in our guide on how to reduce Claude Code token usage.

1. Clear between unrelated tasksRun /clear when you switch topics. Stale context costs tokens on every message that follows, and clearing costs nothing.
2. Match the model to the jobUse Sonnet for routine edits and keep Opus for architecture and hard debugging. Set Haiku for simple subagents.
3. Lower effort on easy workUse /effort low or medium for renames, copy changes and small fixes. Save high effort for work where edge cases matter.
4. Plan before you buildPlan mode lets Claude explore and propose an approach before it writes code, which avoids paying twice for a wrong direction.

If you are choosing between tools partly because of limits, our comparisons of Claude Code vs Codex and Cursor vs Claude Code look at how each one meters usage.

The hidden cost is your own time

A founder on Max 20x who spends three weeks building their own product is not spending $200. They are spending three weeks, plus the quota, plus the hours fixing what the agent got almost right. That is fine while you are validating an idea. It stops being fine when the app needs proper authentication, a real database and deployment before anyone can pay for it, the gap we cover in taking a vibe-coded app to production.

We use the same AI tools, but we review every line they write and price the result as a fixed project, not an open-ended subscription burn. Our services page lists starting prices, from 1,499 lei (about EUR 285) for a landing page to 3,499 lei (about EUR 665) for a web app.

When to do it yourself and when to bring in a team

Do it yourself if you are prototyping, learning or building internal tools where a bug costs you an afternoon, not a customer. A Pro or Max plan and good habits will carry you far. Our post on how AI coding agents change small teams covers the review discipline that makes it work.

Bring in a team when the product has users, payments or personal data, when you keep hitting the weekly cap on work that is not your core business, or when the last 20% before launch keeps slipping. Start with our free website audit if you already have something live, look at what we have built, or send us your project through the services page. We reply in writing with a fixed price.

Frequently Asked Questions

How many messages can I send on Claude Pro?

Anthropic does not publish a fixed message count for Claude Pro. It says Pro gives at least five times the Free plan's usage per session, and the real number depends on message length, attachments, conversation length, model and effort level. Short chats stretch it much further than long coding sessions.

When do Claude usage limits reset?

The session limit resets every five hours. The weekly limit resets at a fixed time each week that Anthropic assigns to your account, which you can see in Settings > Usage on claude.ai or with the /usage command in Claude Code.

Does Claude Code have separate limits from the Claude app?

No. On Pro and Max plans, Claude Code and the Claude app share the same usage limits, so everything you do in both counts against one budget. Since September 14, 2026, weekly limits in Claude Code specifically are 25% higher than they were before Anthropic's summer promotion, but the five-hour limit is unchanged.

Is Claude Max worth it for coding?

Claude Max is worth it if you code with Claude Code most days and regularly hit the Pro weekly limit. Max 5x costs $100 a month and Max 20x costs $200 a month, against $20 for Pro. If you only hit limits occasionally, turning on usage credits on Pro is often cheaper than upgrading.

What happens when I hit my Claude limit?

You see a message that you have hit your session or weekly limit, with the reset time. You can wait, enable usage credits to continue at API rates, upgrade your plan, or switch models if the message is about a specific model like Opus. Recent Claude Code versions can also resume the task automatically after the reset.

Are Claude API rate limits the same as subscription limits?

No, they are separate systems. The API bills per token and limits requests and tokens per minute for each model, plus a monthly spend cap that depends on your usage tier, from $500 on the Start tier to $200,000 on Scale. Subscription plans have five-hour and weekly windows instead.