Google AI Pro Weekly Usage Limit Is Being Exhausted Too Quickly

Google Antigravity
,
1

I am a Google AI Pro subscriber, and I would like to raise a concern about the current usage limits.

A 5-hour usage limit is understandable, and I also accept having a weekly usage limit. However, the issue is that the weekly limit is being exhausted long before I actually reach the 5-hour limit. If I use the service for the full 5-hour window only twice in a day, my weekly quota is almost completely consumed. This makes the advertised 5-hour usage window practically unusable.

What concerns me even more is how quickly the weekly quota is depleted. To test this, I simply entered “Hi” as a prompt. The response was just a basic greeting asking how it could help, yet when I checked my usage immediately afterward, my weekly limit had already increased by 2%. A single, trivial interaction should not consume such a significant portion of my weekly quota.

As a paying Google AI Pro customer, I expect the usage limits to be transparent, fair, and proportional to actual usage. The current behavior makes it feel as though the weekly quota is being consumed much faster than expected, even for very simple prompts.

Please review how usage is being calculated and clarify why such minimal interactions consume a noticeable percentage of the weekly limit. If this is the intended behavior, I believe it should be communicated much more clearly to subscribers before they purchase the service.

I hope this issue can be investigated and resolved, as the current experience does not align with the expectations of a paid Pro subscription.

2

Consumption depends on:

  • Selected model
  • Selected thinking/reasoning level

So it highly varies depending on what you choose. A 2% for a “Hi” prompt sounds normal, because the model pretty much has to guess what you mean by that. It will first load the system prompt, then your rules, then all your skills descriptions, full skills depending on the content of your rules & prompt and finally codebase in its context; then it will not just assume to greet you back. Perhaps “Hi” is a concept in the domain of your product/solution and will start searching for what it really means, maybe loading even more code from your codebase in the process. At high reasoning levels it will probably iterate over itself about five times and then answer you. So in other words, a “Hi” prompt isn’t straightforward or “small”.
Regarding the subscription: The Pro subscription isn’t suited for Agentic development, but as a way for you to test the waters.

Cheers.