Remove the 5-hour usage window for Extra High / Max / Ultra reasoning modes

What variant of Codex are you using?

App

What feature would you like to see?

For long-running Codex tasks, the 5-hour usage window becomes a major limitation when using GPT-5.6 Sol or Daybreak Blue at higher reasoning levels such as Extra High, Max, and Ultra.

In intensive agent workflows, the entire 5-hour allowance can sometimes be exhausted in only 15-25 minutes of actual work. A single deep research task then has to be interrupted and resumed across 3-4 separate 5-hour windows.

The problem is not only the delay. After a multi-hour interruption, Codex can recover persisted context, but it often loses some of the exploratory momentum of the previous run: it tends to reconstruct the task, close the currently visible branch, and move toward summarization instead of continuing the same deep search trajectory. This can materially reduce the quality of long-running research, debugging, security analysis, and other exploratory agent tasks.

I suggest that Extra High / Max / Ultra should not be constrained by the 5-hour usage window at all. Instead, keep the overall weekly allowance and let users consume that allowance continuously when needed.

The total amount of compute available to the user would not have to increase - this would simply remove the artificial 5-hour fragmentation.

In other words:

Current: weekly quota + 5-hour sub-limit → one deep task is repeatedly interrupted.

Suggested: weekly quota only for Extra High / Max / Ultra → the same quota can be spent continuously on one task.

This would make the highest reasoning modes substantially more useful for the kinds of long-running tasks they are best suited for.

Additional information

_No response_