I’ll keep this focused, because I think the issue deserves a clear articulation rather than a wall of frustration.
I’m not here about quota reductions. I’m here about how the quota refresh is triggered.
Right now, the refresh window starts only after the first token is used, not on a fixed schedule. In practice, this means:
- If I don’t interact with the model the moment my window resets, I lose that time permanently.
- To maximize my quota, I’d need to set alarms and initiate sessions at odd hours, including overnight.
- This punishes users with async workflows, irregular schedules, or — simply — jobs and lives outside of AI tooling.
To be clear about my priorities: I wouldn’t mind if the refresh window increased from 5 hours to 8 hours — as long as it runs on a fixed clock. Predictability matters far more to me than raw frequency. A fixed-clock refresh (every N hours from a set anchor time) would solve this entirely and wouldn’t require users to structure their lives around a model’s usage timer.
This is the primary reason I’m reconsidering the Ultra plan. Not the quota size, but THE LOGIC. I’d strongly encourage the product team to revisit this before it becomes a meaningful driver of churn. I suspect I’m far from the only subscriber affected.
