Controlled tests on GitHub found different five-hour quota use across client labels, including a roughly 2.9x gap in one Team-plan test.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [github.com](https://github.com/anthropics/claude-code/issues/97074)
Why it matters #
If the reported differences generalize, developers using Claude Code automation may exhaust the same subscription allowance sooner than users doing comparable work through another entrypoint; the tests do not yet establish why. Developers running Claude Code through its headless CLI and Agent SDK report that the same work can use substantially more of a Claude subscription's five-hour allowance than it does through other entrypoints. The evidence is a series of user-run tests, not an Anthropic finding, and it does not establish whether the difference is intentional or a metering bug.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.