Codex Mac feedback upload times out and reasoning fails to display
What version of the Codex App are you using (From “About Codex” dialog)?
Version 26.818.61809
What subscription do you have?
chat GPT plus
What platform is your computer?
Darwin 25.5.0 arm64 arm bash-3.2$
What issue are you seeing?
The feedback upload keeps loading and eventually times out. The reasoning section sometimes fails to load or display.
What steps can reproduce the bug?
When I submit feedback in the Codex Mac app, the upload eventually times out. I also sometimes see that the reasoning section fails to load or display.
Steps to reproduce:
- Open the Codex Mac app.
- Start or open a conversation.
- Submit feedback or a request.
- Wait for the upload/response to complete.
Actual result:
The upload times out, and the reasoning content may fail to display.
This issue occurs intermittently. I do not have a code snippet because the problem happens in the Codex app interface.
What is the expected behavior?
The feedback should upload successfully without timing out, and the reasoning content should load and display normally.
Additional information
The issue persists after retrying. I can provide screenshots or a session ID if needed.
4 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Correction and clarification of my original report:
#39892 matches the feedback upload failure, but #40553 does not match the
reasoning issue.
The reasoning issue is not about raw reasoning content failing to display.
It concerns a failure to detect architectural drift during a long,
multi-document task.
The task involved an email, a PDF, and several dependent examples. Codex
handled each human correction locally, but did not reconstruct the global
document architecture. Each correction produced another plausible local
repair that created a contradiction elsewhere. I had to correct Codex four
times before it returned to the original plan and restored the intended
roles and ordering of the documents.
Expected behavior:
After multiple corrections affecting the order, medium, or function of
dependent deliverables, Codex should detect a risk of global architectural
drift, pause local execution, reconstruct the canonical plan, and consider
whether a deeper reasoning pass is needed.
The feedback submission failure occurred while I was trying to report this
reasoning issue. Three submissions through Help > Feedback failed after
approximately 11 seconds with:
failed to upload feedback to Sentry: operation timed out
The failure reproduced on two different networks and after restarting the
app.
Therefore:
Le mar. 25 août 2026 à 12:08, github-actions[bot] @.***>
a écrit :
Independent Windows ARM64 confirmation on the same Desktop release family:
26.818.8289.026.818.61809The detailed sanitized evidence is recorded in #39892:
https://github.com/openai/codex/issues/39892#issuecomment-5410774900
This confirms that the feedback failure is cross-platform and that the current UI does not provide an auditable delivery state. No feedback text, logs, account data, local paths, hostnames, or project names are included.
To keep the two reports separate, the architectural-drift issue is now tracked independently in #40849.
This issue (#40584) should be treated as concerning the feedback upload and delivery-state failure, which overlaps with #39892. The distinct failure to reconstruct a global plan across dependent deliverables is documented in #40849.
Please use #40849 for discussion and triage of the architectural-drift behavior.