I am trying to get a “feel” for the new model limits. I really don’t know how to do this except through perspective. I would love to hear other people’s as well. This is my particular way of learning 
.. So here is my morning gotcha:
I opened the Antigravity app. I am still locked out of Google Pro and Claude … 21m of my 5d left on Pro and still 2d on Claude. Gemini Flash is ready with 5h available.
I typed:
”Add “task: “ to the repo symantic prefix options and document it. It is only to be used for the “global-task-finish-workflow”
ENTER
(20 seconds later … all done)
I have 80% of Flash left for the next 4h59m
It all seems to be in a state of flux. I’m not sure anyone can offer an accurate perspective.
Approximately 21m after I wrote that message my Google Pro model was due to reset its usage timer. As the time ticked down, I watched it. When it got to 0 .. it went immediately to 6d again without me having issued a single request to the model. 
I was waiting for my Claude quote to reset and when that happened, it went up by 20% lol. I don’t think anyone has any idea about how the rates work at this point.
Hello Everyone,
Welcome to the Google AI Forum!!!
Thank you for bringing these concerns to our attention. Please be assured that I have shared your feedback with our internal team for further review.