GPT-6 final response disappears immediately after streaming completes in Codex macOS Chat mode #52385

What version of the Codex App are you using (From “About Codex” dialog)?

26.924.22138

What subscription do you have?

gpt plus

What platform is your computer?

Darwin 24.5.0 arm64 arm

What issue are you seeing?

I can reproduce a model-specific display issue in Codex macOS Chat mode.
Observed behavior:

1. GPT-6 begins generating a response.
2. The response streams normally and remains fully visible throughout generation.
3. Immediately after streaming finishes, the entire assistant response suddenly disappears.
4. Switching conversations occasionally restores the response, but usually does not.
5. Sending another user message does not immediately restore the missing response.
6. The next assistant response streams normally.
7. Only after the next response finishes does the previous response suddenly reappear in full.
   Model-specific behavior:
   GPT-5.6 works normally in the same macOS client. The issue occurs frequently with GPT-6.
   Suspected failure boundary:
   The transition from streaming assistant content to a finalized assistant message, potentially involving final-message replacement, message grouping, or turn-state reconciliation.
   Related issue: macOS desktop: completed empty final hides the preceding non-empty final in the same turn #44106.
   Please investigate whether a valid streaming response is being replaced by an empty or incorrectly grouped final message when the turn completes.

What steps can reproduce the bug?

just use GPT-6 in the chat mode

What is the expected behavior?

No response

Additional information

No response