Is anyone else experiencing a significant performance drop with Gemini models in the recent Antigravity updates?
Everything ran smoothly back in v3.5, and v3.6 was still decent. However, ever since updating to v3.7 and v3.8, Gemini has become frustratingly slow and constantly gets stuck on the working... status. Toggling to High mode makes zero difference, even though my internet connection is completely fine.
When switching to Claude (Sonnet/Opus) within the same IDE environment, it responds instantly without any delay but it eats through quota way too fast and gets expensive for daily use.
Why are newer Antigravity releases making Gemini processing so much slower? Is this a server-side latency issue, hidden rate limiting, or a bug introduced in the latest IDE updates?
Same here! v3.5 and v3.6 were smooth, but v3.7/v3.8 are painfully slow. Constant “working…” stalls, and High mode does nothing or lets say low, medium , high uses are similar.
Claude is fast but burns quota too quick for daily use.
I noticed that asking “Are you stuck in a loop?” sometimes snaps it out of it — which makes me think it’s not just network lag, but an agent loop issue.
I am exeperiencing the same thing since September 4th
The Agent has been instanely slow 1 simple task like , expand a width in the UI takes 30 mins! , it’s impossible to have it as a professional tool right now. I am on google ultra plan.
This is happening to my PC and laptop it’s crazy , I thought the update had some conflicts with the cached of the old version data , but it’s not the case.
Gemini(s) do work super fast , like 10 times faster during 4am to 7 AM in the morning (Asia time) it’s crazy , I kinda think google is intentionally limit the global computing power to prevent models to crampped but this if this the case, it’s not the professional way to do it.
