Hit the Claude Code usage limit? What to do, and how to keep working
Options when Claude Code says you hit your session or weekly limit: auto-continue, switch models, usage credits, API billing, or another account.
- CLI
- 0.2.17
- Preview ref
- a2c45c8
TL;DR
When Claude Code says You've hit your session limit · resets 3:45pm, the options are:
- Wait, and let Claude Code continue on its own. On v2.1.234 or later with a claude.ai subscription, it waits in the open session and picks the task up after the reset.
- Switch models, if it's a model limit. An Opus or Sonnet limit applies to that family only, so
/modelkeeps you going. A session or weekly limit covers all models. - Turn on usage credits. Pro and Max can keep working past a limit at standard API rates. Run
/usage-credits. - Move the work to API billing. A Claude Console account bills per token, with no five-hour window.
- Upgrade the plan. Max 5x and 20x raise the per-session allowance.
- Continue on another account you own, or with another agent you already pay for.
Which one makes sense depends on which limit you hit, how close the reset is, and whether you'd rather pay or wait.
The problem
The limit tends to arrive at the worst moment, an hour into a refactor with the tests half green. The session stops, and the context you built up sits there. Starting over somewhere else means explaining it all again, and waiting means the task is idle until the reset.
The first step is to read the message carefully, because the different limits have different fixes.
Step 1: Work out which Claude Code limit you hit
From Anthropic's error reference:
| Message | What it is | What restores access |
|---|---|---|
You've hit your session limit · resets … | The rolling five-hour window, shared across models | The reset, usage credits, or an upgrade |
You've hit your weekly limit · resets … | The weekly limit, shared across models | The weekly reset, usage credits, or an upgrade |
You've hit your Opus limit or Sonnet limit | A limit on one model family | /model to a model outside that family |
You've hit your monthly spend limit and variants | Usage credits reached a spend limit | Raising the limit, or waiting for the plan reset named in the message |
Server is temporarily limiting requests | A server-side throttle, not your quota | Retrying after a moment |
Credit balance is too low | API billing with no balance | Adding Console credits |
Run /usage to see both windows and their reset times. When does Claude Code usage reset? explains how the windows work.
Step 2: Choose how to keep working
Wait and let Claude Code continue automatically
In an interactive session signed in with a claude.ai subscription, automatic continue is on by default from v2.1.234. You see:
Usage limit reached · limit resets 3:45pm
Continuing automatically at 3:45pm · esc to cancelKeep the session open. At the reset, Claude Code tells Claude to pick the task up where it stopped. It doesn't resend your last message. Details from Anthropic's interactive mode page:
- If the computer slept for more than about 30 minutes and the limit reset during that time, it asks you to press Enter.
- It re-arms the wait on its own at most twice in a row. After that it stops with
Automatic continue stopped after repeated usage-limit hits. - Permission prompts still appear as usual, so the continued task can stop on one while you're away.
- To turn it off, use Continue automatically at usage limit in
/config. The Desktop app has its own Auto-continue when limits reset checkbox on the session-limit card, which is separate.
This is the zero-cost option. It's a good fit when the reset is close, or when the task can wait.
Switch models for an Opus or Sonnet limit
/modelPick a model outside the family that hit its limit. Each model has its own prompt cache, so the next request re-reads the whole conversation without cache hits. It's a heavier turn than usual, and it doesn't help with a session or weekly limit.
Turn on usage credits
On Pro and Max, usage credits let work continue past a limit at standard API rates. Run /usage-credits after signing in with your subscription. It opens Settings → Usage on claude.ai, where you turn credits on and set a monthly spend limit. On Team and Enterprise, it asks your admins for more.
Anthropic's help center describes what you see once credits take over: 5-hour limit resets [time] - continuing with usage credits.
Set a spend limit before you need it. A long agent run can spend quickly at API rates.
Move to pay-as-you-go API billing
A Claude Console account bills per token and has no five-hour or weekly window. Anthropic's help article lists this as an option for intensive coding sprints. Switching is a sign-in change, so it doesn't carry the current conversation's billing with it. Set a workspace spend limit in the Console first.
Upgrade the plan
If you hit the five-hour window most days, a bigger plan is the fix that lasts. Max 5x gives five times Pro's usage per session and Max 20x gives twenty times. Anthropic says a new plan's limits apply right away.
Continue on another account you own, or another agent
Some people hold more than one subscription, for example a personal Pro plan and a work seat. Others pay for both Claude and a ChatGPT plan with Codex. In either case the work can continue somewhere you've already paid for. Stay within each provider's terms. Anthropic's Consumer Terms say you may not share your account credentials with anyone else or make your account available to anyone else.
Doing this by hand means a new session and a re-explained task. The next section covers doing it without starting over.
Keep the same session going across your own accounts
If you want a session to carry on when one of your accounts reaches its limit, and follow it from your phone, Happier is an open-source app that lets you run and orchestrate Claude Code, Codex and other agents cross-device, from your phone, desktop or the web.
Account pools hold accounts you own
A pool groups accounts you own, such as your personal subscription and your work subscription, both signed in as you, on your own computer. When one hits its usage limit, Happier can switch the session to the other and continue.
How a pool behaves:
- Nothing switches until you build a pool. Creating one is a deliberate step under Settings → Connected services.
- A new Claude or Codex pool starts with Automatic fallback on. Turn Automatic fallback off in the pool's options if you'd rather switch by hand.
- A switch inside a pool doesn't restart Claude Code or Codex. The new login applies to the running session.
- Selection strategy defaults to Least limited first. Priority order and Manual switching are the alternatives.
- Preventive switching. The Soft switch threshold defaults to 15% left, when another member has fresher quota. Set it to 0 to switch only at a hard limit.
- Switch limits. By default, up to 1 automatic switch per turn and 3 per session hour.
- Recovery mode defaults to switching when possible, otherwise waiting for the reset. The other modes are wait until reset, switch then resume, and off.
Live fallback within a session is for Claude and Codex. Gemini, OpenCode and Pi can hold a pool, but their controls say the service can't be used for runtime account fallback.
The usage limit banner
When a provider asks a session to wait, the session shows Usage limit reached with the reset time and actions such as Resume when limit resets, Stop waiting, Check limit now and, in a pool, Switch account now. Happier sends a continuation prompt when it resumes, not a replay of your original request.
Continue with another Agent
If the other subscription you hold is for a different agent, Continue with another Agent changes the agent running a session in place, keeping the same session and transcript. It works while the current agent is idle, and the handover is a bounded text-only brief, so images from earlier turns don't carry. Switching from Claude Code to Codex without losing context walks through it.
What it doesn't do
Pools don't raise a cap, add allowance, or shorten a reset window. With a single account there's nothing to switch to. Nothing continues while the computer that runs the session is off.
Connected services and their quota meters are experimental and off by default in Happier 0.2.17. Turn on Connected services, and Connected services quotas if you want meters, under Settings → Features. Setup is in Run Happier from your phone. Happier is on GitHub, you can download it here, and there's a Discord.
Prevent the next one
- Watch the meter. A status line with the
rate_limitsfields shows the five-hour and weekly share all the time. See how to check Claude Code usage. - Clear between tasks.
/clearcosts nothing. Every request in a long session re-sends the history. - Match the model to the job. Anthropic's costs page suggests Sonnet for most coding and Opus for complex reasoning.
- Lower effort for simple work. Thinking tokens count as output.
/effortlowers it. - Keep
CLAUDE.mdshort. Move workflow instructions into skills, which load only when used. - Check what's eating it. The
/usagebreakdown names skills, subagents, plugins and MCP servers by share.
Troubleshooting
Automatic continue didn't start
It needs Claude Code v2.1.234 or later, an interactive session and a claude.ai subscription. It doesn't start on API billing. Check that Continue automatically at usage limit is on in /config.
Switching models didn't help
You hit the session or weekly limit, which covers all models. Model switching only helps for an Opus or Sonnet limit.
Usage credits are on but work still stopped
Check the message. A spend-limit message means credits reached the monthly limit you set, or your balance ran out with auto-reload off.
FAQ
How long is the Claude Code usage limit?
The session limit runs on a rolling five-hour window. The weekly limit resets at a fixed time per account. The message shows the exact reset.
Can you bypass the Claude Code usage limit?
There's no way to remove a plan's limit. You can wait, pay for more with usage credits or API billing, upgrade, or continue on another account or agent you hold.
Do you lose the session when you hit the limit?
No. The conversation stays in the open session. If you close it, claude --resume brings it back after the reset.
Do usage credits cost more than the plan?
They bill at standard API rates. Compare a month of credit spend against the price of the next plan up to see which costs less for you.
Get Happier
Happier is an open-source app that lets you run and orchestrate Claude Code, Codex and other agents cross-device, from your phone, desktop or the web.