More control over the queue system

ChatGPT Feature requests
1

I would love more control over the queue system.

1. Automatically pause the queue when usage limits are reached

If I’m working on a project with multiple queued tasks and my usage limit is exceeded, Codex currently continues processing the queue, but each response simply becomes “No usage available.”

As a result, the queued tasks are effectively lost, and I have to manually copy and paste them back into the chat once I have more credits or my usage resets.

Instead, Codex should automatically:

  • Pause the queue when usage is exhausted.

  • Keep all remaining queued tasks intact.

  • Allow the queue to resume once additional credits are purchased or usage resets.

2. Pause the queue without stopping the current task

I would also love the ability to pause the queue while a task is actively running.

Desired behavior:

  • The current task continues running normally.

  • No queued tasks are removed.

  • When the current task finishes, Codex does not automatically start the next queued task.

  • I can manually resume the queue when I’m ready.

This would be especially useful when I want to review the output of a task before allowing the next queued actions to run, without interrupting work already in progress.

Together, these features would make long-running projects and multi-step workflows significantly more reliable and user-friendly.

2

Thanks for joining the community, @LOGAN26!

This is a solid feature request. The current behavior can be frustrating when queued tasks are effectively lost after hitting usage limits, and having more control over when the queue advances would be especially helpful for larger multi-step projects.

I've logged this feedback and will pass it along to the team. In particular, the ideas around automatically pausing the queue when usage is exhausted and allowing users to pause future queued tasks while letting the current one finish seem like valuable workflow improvements.

Appreciate you taking the time to lay out the use cases so clearly.

-Mark G.