Your Claude usage limit, tracked live

How Claude's usage limits work

Every paid Claude plan runs two kinds of limits in parallel. You can be fine on one and blocked by the other, which is why the cutoff so often feels random. It is not: it is two clocks ticking at different speeds.

The 5-hour session window

All your usage inside a 5-hour window draws from one shared session budget, and that budget "will reset every five hours" in Anthropic's own words. Heavy turns drain it faster: Opus, long context, big attachments and extended thinking all cost more per message. Anthropic does not publish per-message numbers, because consumption depends on what each message actually makes the model do.

The weekly caps

On top of the session window, weekly limits cap your total usage. Pro has one weekly limit across all models. Max plans have two: one across all models, and a second one that Anthropic's Max page currently scopes to Sonnet models. Unlike the session window, weekly limits are not rolling: they reset at a fixed day and time assigned to your account, and "your reset day and time stay the same regardless of when you start using Claude", per the same page.

Claude Code spends the same budget

If you use Claude Code on a Pro or Max plan, it does not get its own allowance: Anthropic confirms that usage across claude.ai, Claude Code and Claude Desktop "counts towards the same usage limit". An afternoon of agent-driven coding can quietly consume the evening's chat capacity. The good news: on May 6, 2026 Anthropic doubled Claude Code's five-hour rate limits for Pro, Max, Team and seat-based Enterprise plans, and removed the peak-hours limit reduction on Claude Code for Pro and Max accounts.

Claude plans and their limits, side by side

Anthropic publishes relative capacity rather than message counts. Here is what each plan officially gives you; every cell links to the page it comes from.

Prices are Anthropic's published US prices as of June 2026 and can change. The hour-count estimates that circulate online ("40 to 80 hours of Sonnet per week" and similar) do not appear on any current official page, so we do not repeat them here.

When you hit the wall: official overage options

Hitting a limit no longer just means waiting. Anthropic now offers two documented ways to keep going, both strictly opt-in:

• Usage credits: pay-as-you-go usage billed at standard API rates once your plan windows are exhausted, with a monthly spend cap you set yourself (official details).
• Usage bundles: prepaid credits at up to 30 percent off, applied only after your included limits run out (official details).

Both make tracking more valuable, not less: overage is exactly where an invisible limit turns into a visible bill. Knowing where your windows stand tells you whether to wait for the reset, buy credits, or move the task to another model.

Reset cheat sheet

Where to check: in Claude, Settings then Usage shows your current session and when your weekly limits reset (Anthropic's own guidance). That works, but it is a settings page you have to remember to open. tokenkarma keeps the same numbers in your menu bar.

How to see your Claude usage

Claude warns you late by design: a notice when you are close to the cap, a settings page if you go looking. tokenkarma closes that gap. For the session and weekly windows it shows:

• the percentage used and how much is left, per window,
• the exact time each window resets,
• alerts at 75 and 90 percent, before you hit the wall,
• your usage next to every other AI you pay for.

For the full background on what changed in 2026, read Claude usage limits explained and why Claude Code limits doubled.

Track Claude next to every other AI you pay for

Most people who pay for Claude also pay for ChatGPT, Gemini, Cursor or Perplexity, each with its own limits and its own blind spots. tokenkarma puts them all in one view, so you can move work to whichever model still has room. See the full AI usage tracker or the side-by-side limit comparison.