Forumics logo Forumics
Issue 01

Back to issue

All Codex known issues

CLAIM C053 Supported 2026-11-07

GitHub issue #35259 in openai/codex, opened 2026-07-24 by dimasyankauskas, records a report that during intentional Ultra and multi-agent work on Codex Desktop 26.721.31836 with codex-cli 0.146.0-alpha.3.1 on macOS, the desktop repeatedly re-entered the model merely to wait for agents or poll terminal status, that in a corrected reset-to-49-percent usage window wait and status polling turns accounted for 19.8 percent of raw local token volume, and that the polling often ran at a 30- or 60-second cadence for hours while delegated work was still running.

Verbatim Excerpts from Cited Records
“During intentional Ultra and multi-agent work, Codex Desktop repeatedly re-entered the model merely to wait for agents or poll terminal status.”
— dimasyankauskas · openai/codex (issue 35259 opening body, Summary section) · 2026-07-24T23:11:07Z ·GitHub Issue ↗
“In a corrected reset-to-49%-usage window, model turns whose only tool action was wait/status polling accounted for **19.8% of raw local token volume**.”
— dimasyankauskas · openai/codex (issue 35259 opening body, Summary section) · 2026-07-24T23:11:07Z ·GitHub Issue ↗
“Each polling interval appears to re-enter the model, ingest a large cached conversation prefix, and then make another wait/status call.”
— dimasyankauskas · openai/codex (issue 35259 opening body, What issue are you seeing section) · 2026-07-24T23:11:07Z ·GitHub Issue ↗
“Waiting for an agent or terminal process should be event-driven, or handled by the harness without repeatedly invoking the model.”
— dimasyankauskas · openai/codex (issue 35259 opening body, Expected behavior section) · 2026-07-24T23:11:07Z ·GitHub Issue ↗
Counter-record: Issue remains open with 33 comments as of 2026-11-06; no maintainer comment appears in the surveyed thread (the only automated comment is a duplicate notice). The reporter added a ten-day baseline comparison on 2026-07-24, and independent commenters corroborated the pattern on macOS, Windows, and Linux through 2026-11-04, including fresh reproductions on CLI 0.149.0, 0.153.4, and 0.163.0. One counterpoint: on 2026-07-31 and 2026-08-01 phyrexia reported that empty wait returns fell sharply in CLI 0.146.0 and suggested the fix was pull request 34852, merged 2026-07-23; another commenter argued on 2026-08-03 that 34852 does not remove the residual timer-driven re-entry path and posted a community patch, which is not a maintainer fix. Reported workarounds include long wait timeouts and a developer-instructions config block. The claim documents only the original report, not a verified current state. Falsification probe: If issue 35259 is deleted without archive, if the author edits the opening body so the quoted sentences no longer match verbatim, or if the author posts a retraction of the report, this claim drops to Withdrawn. Audit verification: Verified 100% continuous verbatim substring against official GitHub REST API by scripts/verify_claim_verbatim.py.

See also: Codex vs Cursor head-to-head comparison on 9bests