Can you reset Claude's usage limit?

No, you cannot reset Claude's usage limit

You cannot reset Claude's usage limit. There is no button, no support request, no hidden setting and no trick that clears the meter. The only thing that resets a Claude usage limit is the clock: the five-hour session window rolls over on its own, and the weekly window rolls over at a fixed time each week.

Everything else people try either does nothing, or raises the ceiling without touching the timer. Each of those attempts is covered below, with what it really changes. If you are locked out right now and need the fastest way back to work, what to do when Claude says the limit is reached is the short list.

How do the session and weekly windows reset?

Anthropic's help center describes two limits on paid plans, as of September 2026. The session limit is a rolling five-hour window. The help center's wording is that "your session-based usage limit will reset every five hours". The weekly limit sits on top of it and "resets at a fixed time each week that is assigned to your account". That day and time do not move when you start using Claude or when your subscription began, and each cycle gives you the full weekly allowance again.

Both limits belong to your account, and both are shared. Anthropic's Claude Code article states that "all activity in both tools counts against the same usage limits", so a long Claude Code session on your VPS (virtual private server) counts against the same limit as chatting in the browser. The limits also count every model together. Some plans show an extra weekly bar for Opus on top of the shared one, which matters later, under switching models.

The reset time is never hidden. On claude.ai, open Settings, then Usage, and the Current session bar shows how much of the five-hour window you have used and when it resets. In Claude Code, run /usage for the same bars. When you reach the limit, the message itself carries the time:

5-hour limit reached - resets 3:45 PM.

Claude Code phrases it as You've hit your session limit or You've hit your weekly limit. Since version 2.1.234 it also waits in the open session, shows a countdown line that reads Usage limit reached · continuing automatically at 3:45pm · esc to cancel, and picks the task up by itself after the reset. Keep the terminal open and it does the waiting for you.

What the help center does not publish is a message count, because usage depends on the model, the length of the conversation, the features in use and the effort level. The ceilings also move. Anthropic doubled the five-hour Claude Code allowance on Pro and Max in May 2026 and removed the peak-hours reduction at the same time, so a number copied from an older post is probably wrong. For how the meter is measured, read how Claude's usage limits are counted. For the specific reset windows on the $100 and $200 tiers, see when the Max plan session and weekly limits reset.

Which things look like a reset, and what does each one really do?

Starting a new chat

Nothing happens to the meter. The limit is on your account, and a new conversation draws from the same five-hour and weekly budget as the old one. The one real effect is on the cost of each later message: Claude re-reads the whole conversation on every turn, so a fresh, short chat spends less of your allowance per message than a long one. That is a good reason to start new chats early. It does not give you any usage back.

Logging out, reinstalling, or switching device

Nothing. The count lives on Anthropic's side, attached to your account, so signing out of the web app, clearing the browser cache, deleting the mobile app, or moving from your laptop to your phone changes nothing. Signing in to the same account from Claude Code on a VPS draws from the same meter. A second, free account has its own meter, but it also has the free plan's much smaller five-hour allowance and none of your projects or chat history.

Upgrading your plan

An upgrade changes the size of the bucket. It does not change the clock. The five-hour window you are in keeps its reset time, and the weekly reset day stays the same, so what you gain is a larger allowance inside the same windows. Moving from Pro to Max is charged for the remaining part of the billing cycle only, so you can do it mid-month. Two things to know before you pay in the middle of a lockout. First, the app sometimes keeps showing the old limit for a while after the payment goes through, and Claude Code in particular does not notice an upgrade you made in a browser until you sign in again with /login. The fix for a Claude upgrade whose limits have not updated covers that. Second, an upgrade is a monthly commitment for a problem that ends in a few hours, so decide whether Max is the right answer before you are locked out, not during.

Buying usage credits

Usage credits extend a limit. They do not reset it. On Pro and Max you can turn them on under Settings, then Usage, or from Claude Code with /usage-credits, which opens that same page. Once you pass your included allowance, each further message is billed from the credit balance at standard API rates, and the help center is explicit that "your session limits reset every five hours as usual" while that happens. The claude.ai message changes to 5-hour limit resets 3:45 PM - continuing with usage credits. You can set a monthly spend limit so a long agent run cannot empty the balance, and an auto-reload option tops the balance up when it falls below a threshold you pick. Credits are the honest way to keep working through a window you have already used up. They cost real money per token, so set a spend limit the same day you turn them on.

Switching models

This one is a partial exception. The session and weekly limits count every model together, so when the message says session or weekly, /model will not get you back in. The model-specific messages are different. You've hit your Opus limit and You've hit your Sonnet limit mean only that model family is used up, and switching to a model outside it keeps you working. If the Opus bar is full and the shared bar is not, switch to Sonnet and continue.

What do you really want when you search for a bypass?

The person typing "bypass Claude limit" or "Claude Limit umgehen" is almost never trying to break a rule. They want to keep working. Every honest route to that goal is a different meter, never a reset of the one you hit.

Spend less per message. The same window holds far more work on a smaller model at a lower effort level. In Claude Code, /model switches the model and /effort sets the effort level for the next request. Sonnet handles most coding tasks well and draws much less per turn than Opus, and Haiku is enough for short questions and for subagents. Opus versus Sonnet versus Haiku for each kind of job gives the practical split, and how the effort levels change usage covers the effort side. Clearing context between unrelated tasks helps for the same reason a new chat does, and managing Claude Code's context shows the habits that keep each turn small.

Move the work to the API. The API (application programming interface) is a separate account at platform.claude.com, billed per token, with no five-hour window and no weekly window. Its limits are rate limits, measured in requests and tokens per minute, and they grow with your spend tier. As of September 2026 the list prices are $1 per million input tokens and $5 per million output tokens for Haiku 4.5, and $2 and $10 for Sonnet 5. Claude Code can sign in with an API key instead of a subscription, so the same tool keeps working after the subscription meter is empty. Whether that is cheaper than Max depends entirely on how much you use, and the API versus subscription cost comparison has the arithmetic.

Run an open model on your own server. Claude itself cannot be self-hosted. The weights are not published, so the API and the cloud platforms are the only ways to run it. An open-weight model behind a runner such as Ollama on a VPS has no usage window at all, because the only meter is the server you already pay for. The quality is lower than Claude on hard tasks, and you need enough memory (RAM) for the model you pick, so treat it as a fallback for routine work rather than a replacement. If you are building agents, cost control for AI agents on a VPS shows how to route cheap steps to a local model and keep Claude for the steps that need it.

Whatever you choose, the meter you hit today resets on its own. Read the time in the message, and if you cannot wait that long, extend with credits or move the work to a different meter. Nothing else changes the clock.

FAQ

Does the Claude usage limit reset at midnight?

No. The session limit is a rolling five-hour window, so it resets five hours after it opened, at whatever time of day that falls. The weekly limit resets at a fixed day and time that Anthropic assigns to your account, and that time does not move when you start using Claude or when your billing date changes. The exact reset time is shown in the limit message itself and under Settings, then Usage on claude.ai. Claude Code shows it with /usage.

Does starting a new conversation reset my Claude limit?

No. The limit belongs to your account, so a new chat draws from the same five-hour and weekly allowance. A new chat does make each following message cheaper, because Claude re-reads the whole conversation on every turn and a short conversation costs less to re-read. That saves usage going forward. It does not return any usage you have already spent.

If I upgrade to Max, does my limit reset right away?

The upgrade gives you a larger allowance inside the same windows. The current session keeps its reset time and the weekly reset day does not change. The new allowance should apply once the plan change goes through; if the app still shows the old limit, sign out and back in, and in Claude Code run /login, because Claude Code does not check for an upgrade made in a browser on its own.

Do usage credits reset the five-hour limit?

No. Credits let you keep working past the included allowance, billed per token at standard API rates, while the session and weekly windows keep resetting on their normal schedule. The message changes from a hard stop to a note that Claude is continuing on credits. Set a monthly spend limit when you enable them, because an agent loop can spend the balance quickly.