GitHub issue #97074 in anthropics/claude-code, opened 2026-09-25 by qvarker, records a report, based on a controlled A/B test, that on a Max subscription the same workload uses about 1.8 times more of the 5-hour window when it runs headless via claude -p (entrypoint sdk-cli) than in the interactive TUI (entrypoint cli), with token usage the same in both modes.
Verbatim Excerpts from Cited Records
“On a Max subscription, the same workload uses about 1.8× more of the 5-hour window when it runs headless (`claude -p --input-format stream-json`, recorded as `entrypoint: "sdk-cli"` in the transcripts) than when it runs in the interactive TUI (`entrypoint: "cli"`).”
·
anthropics/claude-code (issue 97074 opening body, What is Wrong section)
·
2026-09-25T11:00:35Z
·GitHub Issue ↗
“The same tokens on the same model should use the same share of the 5-hour window, whether they come from `claude -p` or from the interactive TUI.”
·
anthropics/claude-code (issue 97074 opening body, What Should Happen section)
·
2026-09-25T11:00:35Z
·GitHub Issue ↗
“After moving the same work from the TUI to `claude -p`, we hit the 5h limit several times a day with no change in workload, and we paid for extra usage (`rejected` + `isUsingOverage: true`) that the same work in the TUI would not have needed.”
·
anthropics/claude-code (issue 97074 opening body, What is Wrong section)
·
2026-09-25T11:00:35Z
·GitHub Issue ↗
Counter-record: Issue remains open as of 2026-10-28 with 54 comments. Surveyed comments show multiple independent reproductions (different plans and models, controlled A/B harnesses, Windows, macOS and Linux). Commenter frostfire575 (2026-10-26) reports a workaround of setting CLAUDE_CODE_ENTRYPOINT to a custom tool name, with qvarker confirming it with a controlled A/B the same day; boudra notes a fix in Paseo beta 0.14. No maintainer reply or official statement appears in the surveyed comments; the claim documents the original report and its measurements, not any later resolution.
Falsification probe: If issue #97074 is deleted without archive, if the author edits the opening body so the quoted sentences no longer match verbatim, or if the author or maintainers post a definitive retraction showing the described measurements were mischaracterized, this claim drops to Withdrawn.
Audit verification: Verified 100% continuous verbatim substring against official GitHub REST API by
scripts/verify_claim_verbatim.py.