You’ve reduced rate limits by 5x and on top of that you’ve switched from 3 flash good price model to 3.5 flash that burns tokens like there’s no tomorrow which means that in the end rate limits expire like 10x to 15x faster. This is absurd, unless you want to lose every user to codex, cursor and claude code, in which case it would probably be a good strategy lol..
On top of this, using opus/sonnet constantly results in “agent terminated due to error: Our servers are experiencing high traffic right now..”, I don’t even know what to think at this point..
Personally I have decided to switch to Codex because I can only afford a 20 dollars subscription, and their gpt 5.5 is far better than the gemini models, their rate limits are better than yours, and with equal amount of tokens, gpt 5.5 can achieve much better results. This I/O event was the beginning of the end of using antigravity for me