They changed the quota yesterday so it immediately takes a week to refresh instead of 5h

https://nitter.net/antigravity/status/2031835833716625883#m

> Google AI Pro is the home for the practical builder, hobbyists, students, and developers who live in the IDE and don't necessarily rely on an agent. This plan features generous limits for Gemini Flash, with a baseline quota included to "taste test" our most advanced premium models.

In the followup post they link to

https://antigravity.google/docs/plans

Which claims

    > Users on Google AI Pro receive:
    >    High, generous quota, refreshed every five hours until weekly limit reached
    >    Higher weekly rate limit

I don't know why there aren't more people on here talking about it.

I just checked using $ sudo smartctl -a /dev/nvme0n1

    Percentage Used:                    2%
    Data Units Read:                    7,173,143 [3.67 TB]
    Data Units Written:                 22,666,414 [11.6 TB]

I don't think its mathematically equivalent or even close because the context/logprobs will be very different, since you only produce 1 token per pass. I'd say the token itself has a lot less information than the signal propagating through the residual stream of transformer blocks.

Gemini CLI shows how much was saved through caching each session, and it's usually somewhere around 90%