Claude does not publish one fixed public daily token number for consumer plans. As of April 8, 2026, Free usage resets every five hours, Pro and Max give you more capacity plus weekly limits, and your real headroom still changes with conversation length, attachments, tools, model choice, and current demand.
What many readers call a Claude token limit per day is usually one of four different questions: Free plan usage, Pro or Max session headroom, a context-window question, or Claude Code / API usage. If you keep hitting the limit, identify which system you are actually in first, because that tells you whether the right move is to wait for a reset, buy extra usage, upgrade plans, or switch to a more specific guide.
Verification note: this guide was rechecked against Anthropic pricing and help-center usage pages on April 8, 2026. Where Anthropic uses relative or variable wording instead of a fixed quota, this article keeps that wording intact rather than inventing a cleaner number.
| If you actually mean | Start here |
|---|---|
| Free Claude on the web, desktop, or mobile app | Think in five-hour session resets and variable message capacity, not in one public daily token number. |
| Claude Pro or Max | Think in larger per-session headroom plus weekly limits, not in an unlimited plan. |
| Context window | Think about how much text Claude can hold at once. That is a different question from how much usage you have left today. |
| Claude Code or API billing | Use the Claude Code usage guide or the Claude API key free tier guide, because those surfaces use different meters and billing rules. |
What Free, Pro, and Max actually reset
The easiest way to stop getting misled by quota folklore is to replace the idea of a universal daily token bucket with the public contracts Anthropic actually publishes.
For Free Claude, Anthropic describes a session-based usage limit that resets every five hours. It does not promise one stable public message count or one stable token allowance. Anthropic also says free usage can vary with demand, and that some features can consume capacity faster than ordinary chatting. That is why one person can get through a light afternoon without trouble while another burns through the same plan quickly with long conversations, uploaded files, or web-heavy prompts.
For Claude Pro, the contract gets broader but not cleaner. Anthropic says Pro offers at least five times the usage per session compared with Free during peak hours, still works on a five-hour reset, and also includes a weekly all-model limit. That weekly limit matters more than many third-party guides admit, because a plan can feel fine in short bursts and still start constraining you if the same heavy pattern repeats through the week.
For Claude Max, Anthropic publishes the price and the relative multiplier, not a neat quota. Max 5x costs $100/month, Max 20x costs $200/month, and Anthropic says those tiers provide 5x or 20x more usage per session than Pro. Anthropic also says Max has weekly limits, including one across all models and another for Sonnet models. In other words, Max is much larger than Pro, but it is still not a public promise of unlimited daily tokens.
The cleanest summary looks like this:
| Plan | Official public wording | What resets | What stays variable | What to conclude |
|---|---|---|---|---|
| Free | Session-based usage limit | Five-hour session window | Capacity changes with demand, conversation length, files, tools, and features | Good for normal consumer use, but not for guaranteed heavy daily workloads |
| Pro | At least 5x Free usage per session during peak hours | Five-hour session window plus weekly all-model limit | Exact headroom still changes with workload and current conditions | Better for regular heavy use, but still not a fixed daily quota |
| Max | 5x or 20x more usage per session than Pro | Larger session headroom plus weekly limits | Still dynamic, and Anthropic can apply additional caps | Best for high-interruption users, not for people looking for "unlimited" |
One more detail matters if your Claude usage feels inconsistent: Anthropic says usage across claude.ai, Claude Code, and Claude Desktop counts toward the same shared usage limit. If you spend a lot of time in chat, then jump into Claude Code or a desktop workflow, the product can feel like it "used too many tokens" when the real story is simply that several surfaces are drawing from the same pool.
Why Anthropic does not publish one public daily token number
People keep searching for a single number because that is how many API products train us to think. Consumer Claude plans do not currently work that way.
Anthropic's help pages repeatedly frame consumer usage as variable by design. The factors it names are the ones that actually change the outcome: message length, attached files, current conversation length, model choice, tool or feature use, and overall demand. That is why any article that confidently promises one exact public "Claude daily token limit" should be treated carefully. It may be quoting anecdotes, older plan language, or a rough planning band as if it were a contract.
There is also a smaller source of confusion on Anthropic's own surfaces. The web-search help page says Free accounts have daily usage limits for web search and web fetch. That statement is real, but it does not add up to a universal public daily token budget for all consumer Claude usage. It is better read as a feature-surface constraint layered on top of the broader Free-plan system. If you use long link fetches or search-heavy workflows on Free, you may hit that feature's daily ceiling long before ordinary chat would have forced the same experience.
That is why the useful question is not "How many tokens do I get per day?" but "Which activity is consuming which kind of capacity?" Claude's public consumer answer is shaped by resets and workload, not by a stable token table.
The practical consequence is straightforward. If you want a dependable heavy-use contract, you should not build your expectations around folklore numbers from forums or SEO quota pages. Build them around the current official rules:
- Free resets every five hours and stays variable.
- Pro increases session headroom, still resets every five hours, and adds a weekly cap.
- Max expands that headroom again, still keeps weekly limits, and still does not become unlimited.
That model is less satisfying than a fake exact count, but it is much closer to the product you are actually using.
Usage limit vs context window vs API billing
The phrase "token limit" keeps causing trouble because it points to at least three different systems.
The first system is consumer usage limits. This is the one most readers mean when they ask why Claude stopped them. Anthropic describes it with five-hour resets, weekly limits, shared usage across some Claude surfaces, and plan-relative capacity. This system answers the question, "How much consumer Claude use can I do before I have to wait, pay overflow, or upgrade?"
The second system is the context window. This answers a different question: "How much text can Claude hold in the conversation at once?" Anthropic's current help center says paid Claude plans normally use a 200K context window. That does not mean you have 200K tokens to spend per day. It means the model can work with roughly that much conversation context at a time. A larger context window can help with longer threads, but it is not your daily allowance, and it does not protect you from usage resets.
The third system is API billing and API rate limits. This is the right frame when you are using Anthropic through Claude Console or an API key, not when you are talking about consumer Free, Pro, or Max. API usage is measured through token pricing, rate buckets, and spend limits. That is why a developer question like "How many tokens can I send?" belongs to the API docs or a Console guide, not to a general Claude plan explainer.
Here is the distinction in one glance:
| System | What it measures | Current public contract | Wrong shortcut to avoid |
|---|---|---|---|
| Consumer usage limits | How much Claude plan capacity you have left | Five-hour resets, weekly limits, variable headroom | Treating it like one exact public daily token table |
| Context window | How much text Claude can keep in working memory | Paid Claude plans normally use 200K context | Treating context size as if it were daily usage |
| API billing and rate limits | How API requests are priced and throttled | Token pricing, spend controls, and API rate buckets | Assuming Pro or Max includes API budget |
If your real problem is not general consumer Claude but a coding workflow, go narrower. Use the Claude Code usage guide if you are dealing with shared plan usage, /status, or /cost, and use the Claude API key free tier guide if your question is really about Console credits, funded API use, or per-key expectations.
That separation is more than semantics. It is the difference between choosing the right fix and spending the next week chasing the wrong meter.
What to do when you hit the limit
Once you know which system you are in, the next move becomes much simpler.
If you are on Free Claude, start with workload shape, not quota math. Long conversations, big attachments, and search-heavy or fetch-heavy use can make Free feel much smaller than a plain-text chat session. The first move is usually to wait for the next five-hour reset, shorten the conversation, or reduce what you are asking Claude to inspect in one pass. If that pattern keeps breaking normal work, the real answer is not a better folklore number. It is that Free may no longer match your workload.
If you are on Pro, watch for repeated interruption rather than one bad afternoon. Pro is the right default for many serious users, but the upgrade signal is not "I hit the limit once." It is "five-hour resets and the weekly cap are breaking valuable work often enough that the interruptions now cost more than the price difference." Anthropic's current paid-plan structure also gives you extra usage as an overflow route. If your heavy periods come in spikes instead of every day, Pro plus extra usage can be smarter than living on Max full time.
If you are on Max, do not diagnose every stop as a broken promise. Max is deliberately larger, but Anthropic still documents weekly limits and additional discretionary caps. When Max feels unexpectedly tight, first rule out the simpler explanations: a week with unusually heavy cross-surface use, longer conversations than usual, larger files, more feature use, or a task that really belongs on API billing because it behaves more like an automation workload than a consumer chat workload.
If your real question is about Claude Code, leave this page now. Anthropic says consumer usage is shared across Claude and Claude Code on paid plans, but the practical diagnosis inside Claude Code is different because /status, /cost, context growth, and authentication path all matter. Use the Claude Code usage guide for the mental model and the Claude Code Pro vs Max guide if the real decision is plan choice for coding work.
If your real question is about API tokens, stop using consumer-plan language. Pro and Max are not API budgets. If you need predictable token economics, explicit spend controls, or a funded developer contract, the useful route is the Claude API key free tier guide, then Anthropic's pricing and rate-limit docs after that.
The decision rule is simple: wait when the current limit fits the product you chose, pay overflow when the spike is temporary, upgrade when the interruption has become routine, and switch surfaces when you are asking consumer Claude to behave like an API product.
FAQ
Does Claude have a daily limit?
Sometimes Anthropic uses daily language for specific feature surfaces, especially around Free web search. But Anthropic does not currently publish one universal public daily token number for consumer Free, Pro, and Max. The main public contract is still five-hour resets plus plan-relative and workload-dependent limits.
What resets every five hours?
Anthropic's Free and Pro help pages say the session-based usage limit resets every five hours. That is the short reset cycle most readers should care about first. Pro and Max also have weekly limits, which is why looking only at one busy session can hide the larger pattern.
Is Claude Max unlimited?
No. Max gives much more headroom than Pro, but Anthropic still documents weekly limits and reserves the right to impose additional caps. The right way to think about Max is "larger consumer capacity," not "no more limits."
Does web search count against Claude limits?
Yes. Anthropic's web-search help page says web search and web fetch use capacity on Free accounts and can consume a meaningful share of that feature's limit. If search-heavy use is what is making Free feel small, the problem may be your feature mix rather than a mysterious drop in general chat capacity.
Is the context window the same as the Claude daily limit?
No. The context window is about how much text Claude can hold in the conversation at once. Usage limits are about how much plan capacity you have left before reset or overflow. They are related in the sense that bigger workloads can consume more usage, but they are not the same meter.
What if I really mean Claude Code or API tokens?
Then this page is only the first correction. Use the Claude Code usage guide if you are asking about shared paid-plan usage in coding workflows, or the Claude API key free tier guide if you are asking about API billing, Console credits, or token-priced developer usage.
The practical takeaway
The clean answer to "Claude token limit per day" is that consumer Claude does not publish one fixed public daily token number. Free, Pro, and Max are better understood as variable usage systems with reset rules, not as neat public token buckets.
If you remember only three things, make them these: five-hour resets matter more than folklore counts, weekly limits matter more than people expect, and "token limit" often hides the fact that you are actually asking about a different system entirely. Once you sort those three layers correctly, Claude's current pricing and usage surfaces become much easier to navigate.