How the Claude Code usage limit works
| Limit | How it works |
|---|---|
| Session limit (5-hour window) | The window starts with your first message and the limit resets in full five hours later. Verified Ultra accounts don't have a five-hour limit. |
| Weekly limit | Applies across all models and resets at a fixed day and time assigned to your account, the same each week whenever you start using Claude. |
| Shared pool | Usage in claude.ai, Claude Desktop and Claude Code counts toward the same limits. |
| Model and context cost | Opus uses several times more per turn than Sonnet, and Sonnet more than Haiku. Every turn resends the conversation so far and the files Claude has read, so long sessions and higher effort use your limit faster; /clear and /compact keep it down. |
| API key billing | Signed in with an API key (Console, Bedrock, Vertex or Microsoft Foundry), Claude Code is pay-as-you-go, billed per token to that account instead of drawing on your plan's limits. /cost shows the session's spend. |
Check your usage with /usage in Claude Code, which shows your plan usage limits, or in Settings > Usage on claude.ai, which shows the current session and your next weekly reset.
What uses up your Claude Code usage limit fastest
The biggest drains are a heavier model than the task needs and a long conversation, because every turn resends the whole session so far plus your CLAUDE.md and the files Claude has read. Anthropic's Help Center article on models, usage and limits in Claude Code lists five habits that stretch your usage furthest:
- Match the model to the job. Opus uses meaningfully more of your quota than Sonnet, and Sonnet more than Haiku. Switch with /model, or use /model opusplan, which uses Opus while planning and Sonnet for execution.
- Clear between tasks. Every previous message is resent on every turn, so run /clear when you start a new task and /compact when you're continuing a long one.
- Point at files instead of pasting them. Anything you paste stays in context for the rest of the session; a bare file path lets Claude read selectively, while an @ mention pulls in the entire file plus its CLAUDE.md tree.
- Keep CLAUDE.md lean, under roughly 200 lines. It is sent with every turn; prompt caching makes turns after the first cheaper, but it still takes up context on every message.
- Ask for a plan before big changes. For anything touching more than two or three files, use Plan Mode or ask Claude to list the files it will touch first, then correct the plan before it executes.
Keep working when Claude Code hits its limit
See which limit you hit and when it resets
Claude Code's message names the limit, such as "You've hit your session limit" or "You've hit your weekly limit", and shows when the window resets. Run /usage for your plan usage bars, or open Settings > Usage on claude.ai for the current session and your next weekly reset.
Choose a Claude option
Run /rate-limit-options to wait and continue automatically when the limit resets, add usage credits, or upgrade your plan. /usage-credits opens your usage-credits settings; Team and Enterprise members without billing access send a request to their admin instead. After a model-specific Opus or Sonnet limit, switching to a model outside that family with /model keeps you working.
Save where the task stands
Commit your work in progress on a branch and push it, and note the goal, what is done, what is next and how to check it. Reason can't see your previous agent's conversation, so the branch and your notes carry the task.
Open Reason on the web or the Mac app
Sign in at reasonmachines.com, or download the Reason Mac app for macOS from reasonmachines.com/download and sign in there. Both open the same Sessions.
Give Reason the task in the composer
Tell Reason which repository and branch to work on, or in the Mac app which folder on your computer, then paste your notes. Pick a GPT model on your ChatGPT plan or another connected model in the composer, so the task keeps moving while your Claude Code limit resets.
What Anthropic offers when you hit a limit
Anthropic lists four options for Pro, Max and Ultra plans: enable usage credits to keep going after the five-hour or weekly limit, switch to a Claude Console account and buy API usage credits, wait until your limits reset, or upgrade (Pro to Max, Max to Ultra). If you upgrade, the new plan's limits apply right away.
If Claude Code is partway through a response when you reach the five-hour limit, it may keep working briefly to finish the step, showing "Usage limit reached · wrapping up". That allowance is small, counts toward your weekly limit, and on Pro applies up to once per weekly period; Anthropic says it is rolling out gradually. Verified Ultra accounts have no five-hour limit, so it doesn't apply to them.
Watch for an ANTHROPIC_API_KEY environment variable: if one is set, Claude Code uses that key instead of your subscription, and you pay API rates. The monthly API credits that come with Max and Ultra can't be used in Claude Code.
Checking what's left
/usage shows session cost, plan usage limits and activity stats; /cost and /stats are aliases. Its breakdown of what counts against your limits is computed from session history on that computer, so usage from other devices or claude.ai isn't in it. Settings > Usage on claude.ai shows the current session and your next weekly reset across every Claude surface.
/status opens the Status tab with your version, model, account and connectivity. Anthropic's help center also suggests /status to monitor your remaining allocation, but the commands reference describes plan limits under /usage.
Carrying the task on in Reason
A handoff moves the state, not the conversation: Reason starts from your branch and your notes. In the web app it works on the repository you name; in the Mac app it can also work in a folder on your computer, uncommitted changes included. It runs on the model accounts you connected in Settings › Models, such as your ChatGPT plan, and on a Reason plan's credits after that.
If the account a Reason turn runs on hits its own usage limit, Reason moves the turn to the same model on your other connected accounts, then to Reason credits, and tells you in the conversation.
Questions
Does Claude Code have its own usage limit?
No. On a paid Claude plan, Claude Code draws on the same five-hour session limit and weekly limits as claude.ai and Claude Desktop. If you sign in with an API key instead, Claude Code is billed per token to that account and doesn't use your plan's limits.
How long until my Claude Code limit resets?
The five-hour session limit resets in full five hours after the first message of the window. Weekly limits reset at a fixed day and time assigned to your account. Claude Code's limit message shows the reset time, and Settings > Usage shows the next weekly reset.
Does using claude.ai count against my Claude Code limit?
Yes. Anthropic says usage across claude.ai, Claude Code and Claude Desktop counts toward the same limits.
How do I make my Claude Code usage last longer?
Anthropic's Help Center suggests five habits: use Sonnet or Haiku for routine work and save Opus for hard problems (or /model opusplan to plan with Opus and execute with Sonnet), run /clear between tasks and /compact on long ones, reference files by path instead of pasting them, keep CLAUDE.md under roughly 200 lines, and ask for a plan before changes that touch more than two or three files.
Can I keep using Claude Code after the limit?
Yes, if you pay for it: turn on usage credits, or switch to a Claude Console account billed at API rates. Otherwise wait for the reset, or carry the task on in Reason on another model account you already pay for, such as your ChatGPT plan.
Can Reason continue a Claude Code task?
Yes, if you hand it the state: commit and push the work in progress, then tell Reason the branch and what's next in the web or Mac app. Reason can't read the Claude Code conversation itself.
What does Reason cost?
Reason's plans are Pro at $50 a month and Max at $200 a month. Models you run on your own connected accounts count against those accounts, not your Reason credits.