Intermediary status updates are buffered and appear only at end of turn

Resolved 💬 3 comments Opened Mar 11, 2026 by chburket Closed Mar 11, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

Summary

In the Codex chat UI, intermediary assistant updates appear to be buffered and then displayed all at once at the end of a turn, instead of streaming as they are sent.

Repro

  1. Ask the assistant to run a ~30s multi-step task and send status updates during execution.
  2. Assistant sends multiple timestamped status updates over time (example below).
  3. Observe UI behavior.

Actual

All intermediate updates become visible only at the end of the task; no incremental updates are shown in real time.

Expected

Each status update should appear in the UI when emitted.

Evidence from one run

Assistant-side send timestamps:

  • 2026-03-11T11:05:31.721-0400
  • 2026-03-11T11:05:40.358-0400
  • 2026-03-11T11:05:49.268-0400
  • 2026-03-11T11:06:02.969-0400
  • 2026-03-11T11:06:16.645-0400
  • 2026-03-11T11:06:26.291-0400

User observation:

  • All of the above messages appeared at once at the end (no intermediate visibility).

Impact

This makes long-running work appear stalled and reduces trust in progress reporting.

Environment

  • Codex account: chburket_cisco
  • Date: 2026-03-11
  • Local timezone: America/New_York

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 5 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #13106
  • #13135
  • #13343
  • #13583

Powered by Codex Action

chburket · 5 months ago

Additional environment details from reproduction machine:

  • IDE: Windsurf
  • App version: 1.107.0
  • Commit: 16cc024632923bc387171d59cf5638057d4c8918
  • Arch: x64
  • OS/runtime:
  • WSL2 on Windows
  • Ubuntu 24.04.3 LTS (Linux 6.6.87.2-microsoft-standard-WSL2)
  • OpenAI/Codex integration in IDE:
  • Extension: openai.chatgpt@0.4.79 (openai.chatgpt-0.4.79-linux-x64)
  • Codex binary version: codex-cli 0.46.0

If useful, I can also attach a short screen recording showing that intermediary status messages are emitted with distinct timestamps but only rendered in the chat panel once the turn completes.

chburket · 5 months ago

Closing as duplicate of #13583.

Confirmed in Windsurf + OpenAI extension path that intermediary updates are emitted on cadence but render in a batch; a focus-loss/focus-regain event causes queued updates to appear immediately.

Tracking further updates in #13583.