Claude Weekly Limit Reached
Claude's paid plans have two separate caps running at the same time, and most confusion comes from mixing them up. The session limit is a rolling five-hour window that starts with your first message and clears on its own a few hours later. The weekly limit sits on top of it and tracks total usage across a seven-day period, so you can be well inside the session limit and still be locked out for days. Heavy Claude Code users, people who default to the largest model, and anyone working in very long chats with big file attachments burn the weekly cap fastest, which is why reports of 'my whole week gone in two days' are common. The fix is to read your actual reset time instead of guessing, change what is consuming the allowance (model choice, conversation length, repeated uploads, agent runs), and decide in advance whether extra usage or a higher plan is worth paying for. Anthropic adjusts these limits periodically (including temporary boosts and one-off resets during 2026), and it does not publish exact message counts, so treat the Usage page in your own account as the source of truth.
Affiliate link, we may earn a commission at no extra cost to you.
Symptoms You Might See
A banner says you have reached your weekly limit and shows a reset day and time
Claude Code stops mid-task with a usage limit message even though the 5-hour window should have cleared
The weekly progress bar in Settings > Usage fills far faster than the session bar
Waiting a few hours does not restore access, unlike a normal session limit
Larger-model usage runs out while a smaller model is still available
Teammates on the same plan seem to get much more use than you
Common Causes
Using the largest model for everything
Bigger models consume the allowance much faster per message than smaller ones. Leaving the top model selected for quick questions, formatting, and simple edits drains the weekly cap on work a lighter model would handle just as well.
Claude Code and agent runs
Agentic coding sessions read files, run tools, and loop through many model calls for a single request. One long refactor or debugging session can use as much as days of ordinary chat, which is why Claude Code users report hitting the weekly cap first.
Very long conversations
Each new message is processed alongside the conversation so far. Deep into a long chat, every reply costs far more than it did at the start, so staying in one thread all week is one of the fastest ways to exhaust the limit.
Re-uploading the same large files
Attaching big PDFs, codebases, or datasets to fresh chats repeatedly counts that content again each time. Anthropic's own guidance notes that content kept in a Project is cached and does not count fully against limits when reused.
Confusing the session limit with the weekly limit
The 5-hour session limit resets automatically and quickly. The weekly limit resets on its own schedule shown on the Usage page. People who expect the weekly lockout to clear in a few hours assume something is broken when it is working as designed.
Step-by-Step Fixes
Check the exact reset time in Settings > Usage
When to try: First, as soon as you see the limit message
Open Settings, then Usage, on claude.ai or the desktop app. Paid plans show separate progress bars for the current session and the weekly limit, along with when each resets. In Claude Code, the usage command shows the same information from the terminal. Plan around that time instead of retrying.
Switch to a smaller model for routine work
When to try: Immediately, and as a permanent habit
Use the model picker to move everyday tasks (summaries, rewrites, quick code questions) to a lighter model and reserve the largest model for genuinely hard reasoning. On Max plans there can be a separate weekly limit for some models, so check which bar is full before assuming you are fully locked out.
Start fresh chats instead of extending long ones
When to try: Whenever a thread has run for dozens of turns
When a conversation gets long, ask Claude for a short summary of decisions and open questions, then paste that into a new chat. You keep the context that matters while dropping the expensive history that every new message would otherwise reprocess.
Put reference material in a Project
When to try: If you work with the same files across many chats
Upload recurring documents, style guides, or code context to a Project once rather than attaching them to every chat. Project knowledge is cached on reuse, so repeated questions about the same material cost less of your allowance.
Scope Claude Code sessions tightly
When to try: If Claude Code is what is draining your week
Give Claude Code a specific, bounded task, point it at the relevant files, and clear or compact the conversation between unrelated tasks. Avoid open-ended instructions like 'improve this codebase' that trigger long exploratory loops through many files.
Turn on extra usage with a spending cap
When to try: When a deadline cannot wait for the reset
In Settings > Usage, paid individual plans can enable extra usage so work continues past the limit, billed separately at standard API rates. Set a monthly cap with Adjust limit so a runaway session cannot produce a surprise bill. Team and Enterprise owners control this at the organization level.
Upgrade only if the pattern repeats
When to try: After two or three weeks of consistent lockouts
If you hit the weekly cap most weeks even after changing model choice and chat habits, compare Pro against Max 5x and Max 20x on Anthropic's pricing page. Upgrading for a single bad week is usually more expensive than extra usage for that week.
How to Prevent This
Glance at the weekly bar in Settings > Usage at the start of each working day
Default to a smaller model and switch up only when a task needs it
Summarize and restart long chats rather than letting them grow all week
Keep recurring files in a Project instead of re-uploading them
When to Contact Support
Contact Anthropic through support.claude.com if the Usage page shows the weekly limit as full when you have barely used Claude, if the reset time passes and access does not return, or if you were billed for extra usage you did not enable. Include your plan, the time you hit the limit, and a screenshot of the Usage page. Support cannot raise individual limits on request, but it can investigate metering errors.
Frequently Asked Questions
When does the Claude weekly limit reset?▼
It resets on its own schedule, which is shown on the Usage page in Settings (and via the usage command in Claude Code). It is a seven-day window, separate from the rolling five-hour session limit, so waiting a few hours does not clear it. Check the exact day and time in your account rather than assuming a fixed calendar day.
What is the difference between the session limit and the weekly limit?▼
The session limit covers a rolling five-hour window that starts with your first message and resets automatically. The weekly limit tracks total usage over seven days and sits on top of it. You can be under the session limit and still be blocked by the weekly one.
Why did I hit the weekly limit in one or two days?▼
The usual culprits are using the largest model for everything, long Claude Code or agent sessions, very long conversations where every message reprocesses the history, and repeatedly uploading large files. Changing those habits typically makes the same plan last several times longer.
Can I keep using Claude after hitting the weekly limit?▼
Yes, if you enable extra usage in Settings > Usage on an eligible paid plan. Usage past the limit is billed separately at standard API rates, and you can set a monthly spending cap. Otherwise you wait for the reset shown on the Usage page.
Does Anthropic publish exact weekly message counts?▼
No. Limits depend on model, message length, attachments, and conversation length, and Anthropic has adjusted them several times, including temporary increases during 2026. The progress bars on your own Usage page are the only reliable measure of how much you have left.
Sources and verification
Product behavior and limits can change. These primary sources were used to verify this guide.
Other Common AI Errors
ChatGPT Not Working
Service Issues
🚦Rate Limit Exceeded
Usage Limits
📏Context Length Exceeded
Input Limits
🔥Model Overloaded
Service Issues