Feature Request: Priority / Standard / Flexible Compute for ChatGPT Pro

ChatGPT Feature requests
1

Feature Request: Flexible Compute Mode for ChatGPT Pro

I would like to suggest a simple feature for ChatGPT Pro:

Priority | Standard | Flexible

The idea is that not every task needs to run immediately.

For example, if I start a large coding, research or analysis task in the evening, I often do not care whether it runs at 10 PM, 2 AM or 7 AM. I only care that it is finished by the next morning.

With Flexible Mode, the user could choose something like:

  • finish within 3 hours

  • finish within 6 hours

  • finish by tomorrow morning

  • use the cheapest available compute window

In exchange, Flexible Mode could use fewer credits or provide more usage than Priority Mode.

This could benefit both sides:

  • Priority users still get the fastest available performance

  • Flexible users get better value for their credits

  • OpenAI can move non-urgent workloads away from peak periods

  • compute capacity could be used more efficiently

  • users with hobby projects or larger coding tasks could afford more compute

Instead of giving every task the same urgency, let the user decide whether the task is urgent or flexible.

I think this would be especially useful for Pro users.

A simple switch like:

Priority | Standard | Flexible

could make a big difference.

2

Great idea, flex is already exposed in the API. So adding a timing option to ChatGPT and sending it flex should be doable.

3

The ChatGPT desktop app, which is based on Codex, has a config setting for selecting the service tier, including flex, priority, and default (standard).

Unfortunately, this is currently bugged because the in-app settings override the config. However, the fact that this setting already exists also means this is very likely to become an option soon.