The short answer
When you sign in to Claude Code with a Pro or Max subscription, it draws from the same limits as claude.ai: a rolling 5-hour session and a weekly limit, plus any per-model weekly cap. Chat and coding share one pool, so a long Claude Code session can leave you blocked on claude.ai too. If you use Claude Code with an API key instead, there are no subscription limits: you pay per token under your API account's rate limits.
Which limits apply to you
| How you use Claude Code | Limits | Where to see them |
|---|---|---|
| Pro or Max subscription | Shared 5-hour session and weekly limit with claude.ai, plus per-model weekly caps | /usage inside Claude Code, claude.ai/settings/usage, or Claude Monitor in your browser |
| API key (Claude Console) | No session or weekly limit; pay per token, within your API rate limits | The Claude Console; ccusage for token and cost totals from local logs |
For how each limit works and when it resets, see Claude usage limits explained.
Why Claude Code uses your limit faster than chat
- Agentic loops. One request can turn into many model calls as Claude Code reads files, runs commands and edits code.
- Large context. Each call carries the conversation so far plus the files it has read, so long sessions cost more per step.
- Stronger models. The most capable model uses more of your limit per message than a lighter one.
How to make your limit last
- Clear or compact between tasks. Use
/clearwhen you switch to an unrelated task and/compactto shrink a long conversation, so each call carries less context. - Match the model to the job. Switch with
/modeland keep the strongest model for the hard parts. - Plan around the reset. If the session window is nearly used up and resets soon, wait for the reset before starting a big refactor.
- Keep the numbers in view. Running
/usageor checking the usage page by hand is easy to forget mid-task. A tracker in the toolbar shows the session and weekly percentages and warns you at 80% and 95%.
Tools that help
Claude Monitor (ours, free and open source) reads the official limits from claude.ai and shows the session, weekly limit, per-model caps, extra usage credits and routine runs in your browser toolbar, with a countdown to each reset. Because the limits are shared, the numbers you see there include your Claude Code usage.
ccusage (open source, not ours) reads Claude Code's local logs and reports tokens and estimated cost per day, session or 5-hour block. It answers "how much did I use", while the official percentages answer "how much is left"; many people run both. To compare browser trackers, see Claude usage trackers compared.
Questions
Does Claude Code share usage limits with claude.ai?
Yes, on Pro and Max plans. Claude Code and chat on claude.ai draw from the same 5-hour session and weekly limit, so usage in one reduces what is left in the other.
How do I check my Claude Code usage limit?
Inside Claude Code, run /usage for a quick check of your 5-hour session and weekly usage with reset times. On claude.ai, the same numbers are at claude.ai/settings/usage. Neither keeps history or warns you in advance; Claude Monitor shows the same numbers in your browser toolbar and alerts you before you run out. For token counts and estimated cost from your local Claude Code logs, ccusage is a good complement.
When does the Claude Code limit reset?
The session limit resets five hours after the window started, and the weekly limit on its own 7-day schedule. Because the limits are shared with claude.ai, the reset times are the same ones shown on the usage page.
Does using an API key avoid the limits?
An API key from the Claude Console is billed per token and is not subject to the subscription's session and weekly limits, only to your API rate limits. It is a different cost model, so heavy use can cost more than a subscription.
Do Claude Code routines count against my limits?
Routines have a daily run allowance of their own, shown on the usage page next to the other limits. Claude Monitor shows it as a used and limit count.