Claude Code 5-hour limit, explained

This is the one that stops a task mid-flight. Here's how the rolling 5-hour window actually behaves, and why it can feel unpredictable if you don't know the mechanics.

It's rolling, not a fixed clock

The 5-hour limit tracks usage over the trailing 5 hours, not a fixed block like "9am to 2pm." As your oldest activity in that window ages past the 5-hour mark, the allowance it used frees back up. There's no single moment it "resets" for everyone; your own reset time depends on when you were busiest.

What burns it fastest

Long agentic runs, heavy thinking/reasoning modes, and bigger models all draw down the 5-hour window faster than quick single-file edits. A single "refactor this and run the tests" prompt on a large codebase can spend a meaningful chunk of the window in one go, especially with extended thinking switched on.

What happens when you hit it

Once you're at the cap, Claude Code stops taking new requests until enough of the window rolls off to free up allowance again. If your plan supports switching model or effort level, dropping to a lighter setting is often the fastest way back to working, rather than waiting out the full window.

Why it feels unpredictable

Because it's rolling, the same task can behave differently depending on what you did in the previous few hours. Two similar refactor sessions, one right after a quiet morning and one stacked on top of a heavy session, can hit the wall at very different points, even though the work looks the same on your end.

Seeing it coming

ReserveGauge shows your live 5-hour usage and reset countdown in the menu bar, and projects your current pace forward so you know before you start a long task whether you'll actually finish it inside the window you've got left.