Dear Google:
Gemini 3.1 pro is so absolutley … when it comes to code… I am working on a very larg multi-piece engine. with 3, it was a breeze, worked fine high limits, good context and understanding. Occasionally I’d have to restart to a new conversation to keep it fresh after an error…
But my god.. 3.1 is top tier xxxx. You nerfed it majorly. Same with Antigravity, both in one fell swoop. First it will tell me my limit is reached and will recharge in 3 hours, then i go check and it now says 7 days. I know you lowered limits, but with a VAGUE announcement . No warning or anything. However, my limits are still NOT as implied in that update doc.
Flash is still OK, but it gets so much wrong overtime its annoying to use (but the most non-broken model on antigravity.)
Next, claude? Why do you even have claude in there if the limits are so small i get two prompts with it till its blocked. Yeesh. Thing is, those two prompts save my butt many times when Gemini goes full idiot mode.
And finally Nano Banana 2…. How did you break that so bad. It is unusable for someone in a niche art workflow. Constantly not understanding style basis, I always have to redo with pro. Nano banana pro was great. Is great. Why is there no default setting where I can just do “ALWAYS USE NANO BANANA PRO”
And why not let us just keep using Gemini 3? Your new stuff is horrible, maybe its fresh for new project out of the box, but hwne it comes to big systems built by multiple models 3.1 falls short heavily.
Thank you for coming to my ted talk.