[CRITICAL FEEDBACK] The Shared Quota Pool is Ruining the Gemini Pro Developer Workflow

Hi everyone and Google Product Teams,

As a Google Advanced/Pro subscriber using the Gemini ecosystem as a primary development partner, the latest quota update has introduced a massive regression that completely ruins the developer workflow.

Here is the bitter truth about the current state of Gemini Pro and Flash limits:

  • The Shared Quota Penalty: Forcing Gemini 3.5 Flash and Gemini 3.1 Pro into a shared, restrictive quota pool makes absolutely no sense. We are being heavily penalized for using a lightweight model (3.5 Flash) that was heavily marketed as ultra-cheap, fast, and efficient.
  • Massive Regression in Usability: With the older Gemini 3 Flash model, the 5-hour rolling quota was generous enough that it rarely interrupted a continuous development session. Now, the shared quota burns out within 20–25 minutes of active coding/debugging, forcing a grueling 4.5-hour lockout. This is completely unacceptable for anyone trying to build, audit, or deploy production-ready systems.
  • Paid Tier Contradiction: Pro subscribers are paying a premium to build, iterate, and develop without arbitrary operational roadblocks. The current execution of these limits feels like a step backward for paying users.

What Google Needs to Address and Implement Immediately:

  1. Decouple the Pools: Separate the Gemini 3.5 Flash quotas from the heavier Pro models entirely.
  2. Restore Development Velocity: Give Gemini 3.5 Flash a highly generous, high-frequency rate limit for Pro subscribers—mirroring the operational freedom we enjoyed with the previous version.

If Google expects developers and architects to rely on the Gemini ecosystem as a primary infrastructure partner, these restrictive short-term and weekly caps on high-speed models must be overhauled immediately.

Please share your thoughts below. Let’s bring this to the product team’s attention. Remove the weekly/short-term shared limits!

I agree with your points and the top up token burn is insane now.

I think they have unhooked the models for now. Probably working on it. Go devs go!