Feature Request: Replace Hard Weekly Usage Resets with Gradual Quota Replenishment
I would like to suggest changing the way weekly usage limits for ChatGPT features such as Work and Codex are replenished.
Instead of restoring the entire weekly allowance to 100% at a specific reset time, I believe it would be better to replenish the allowance gradually over time.
For example:
-
Replenish 1/28 of the total weekly allowance every 6 hours
-
Or replenish it hourly, every few minutes, or continuously
-
Keep the maximum available allowance capped at 100%, as it is today
Since one week contains 168 hours, dividing the weekly allowance into 28 six-hour periods would mean that approximately 3.57% of the total weekly allowance is restored every 6 hours.
If the allowance were completely exhausted, it would therefore take exactly seven days to return from 0% to 100%.
This request is not intended to increase the total amount of usage available per week.
The goal is simply to change the replenishment mechanism while keeping the same overall weekly capacity.
Problems with the Current Hard Reset Model
The current model, where the entire weekly allowance is restored at once at a specific time, can make usage planning unnecessarily difficult.
This is especially noticeable when using Work or Codex for longer-running development or research tasks.
For example, if an immediate usage-limit reset causes the next weekly reset date to move, a user who had planned their workload around the previous reset schedule may suddenly have to reorganize their usage plan.
This can create situations such as:
“I had planned my usage around the original weekly reset date, but after resetting the limit, the next full reset moved to a different date.”
For users who manage their weekly allowance carefully, this makes long-term planning less predictable.
There is also another issue with hard resets.
If a user still has a significant amount of unused capacity shortly before the weekly reset, the current system can create a strong incentive to think:
“My allowance is going to reset soon anyway, so I should use the remaining capacity now before I lose it.”
This creates a “use it or lose it” incentive.
That may encourage users to consume capacity simply because a reset is approaching, rather than because they actually need to use ChatGPT at that moment.
Advantages of Gradual Replenishment
A gradual replenishment system could provide several benefits.
1. More predictable usage planning
Instead of having to think about a specific date and time when the entire allowance suddenly returns to 100%, users could simply know that a certain amount becomes available again over time.
For example:
-
6 hours: approximately 3.57% restored
-
24 hours: approximately 14.29% restored
-
7 days: 100% of the weekly capacity restored
This would make it easier to plan longer-running Work or Codex tasks.
2. Less disruption from reset-date changes
If capacity is replenished continuously or periodically, the exact concept of a weekly “reset date” becomes much less important.
This would reduce the impact of situations where an immediate reset changes the date of the next full reset.
3. Less “use it before you lose it” pressure
Users would have less reason to rush to consume their remaining allowance shortly before a reset.
Unused capacity would simply remain available up to the maximum limit, while previously consumed capacity gradually returns over time.
4. Potentially smoother usage distribution
A hard reset can encourage users to concentrate usage around certain points in their reset cycle.
With gradual replenishment, there would be less reason to intentionally increase usage immediately before a reset.
This could potentially result in more naturally distributed usage over time.
5. The total weekly allowance would not need to increase
This change could be implemented without increasing the total amount of compute or usage available to each user over a seven-day period.
The same overall weekly capacity could be retained while changing only the way that capacity becomes available again.
Example Model
One possible implementation would be:
-
Maximum capacity: 100%
-
Replenishment: 1/28 of the weekly allowance every 6 hours
-
Replenishment stops when the available allowance reaches 100%
Under this model:
-
6 hours restores about 3.57%
-
12 hours restores about 7.14%
-
24 hours restores about 14.29%
-
3.5 days restores 50%
-
7 days restores 100%
The exact replenishment interval does not have to be six hours.
An hourly model, a minute-by-minute model, or even continuous replenishment could work equally well.
For example, an hourly model could replenish approximately 1/168 of the weekly allowance per hour, or about 0.595% per hour.
The important part is not the exact interval, but replacing the hard weekly reset with a rolling replenishment mechanism.
Requested Change
I would like OpenAI to consider changing weekly usage limits for features such as Work and Codex from:
A hard reset where the entire allowance returns to 100% at a specific time
to:
A rolling system where used capacity is gradually replenished over time, up to a maximum of 100%.
I believe this would make usage limits easier to understand and plan around, reduce the “use it or lose it” incentive before reset times, and potentially encourage more naturally distributed usage.
Most importantly, this could be done without increasing the total weekly allowance.
Thank you for considering this suggestion.