So Gemini 3.1 dropped huh?
Trying to use it in Antigravity - you need to update. But auto update doesn’t see update. Okay downloaded new from website and installed.
Gemini 3.1 - after 5 minutes I stopped the loop of “waiting/generating” with no reasoning or content output.
Gemini 3 Pro - “Not available anymore use Gemini 3.1”
Opus, I assume will have the same issues like the past month - 80% failure rate? At least Sonnet works for now.
But not being able to use Gemini in a Google product is hilarious.
Also, where Gemini 3.1 does actually work (outside of Antigravity only) it is not much of an improvement over Gemini 3 which we all know is not very good. This “better than Opus 4.6” on benchmarks.
The reasoning content is still pure BusinessSpeak - there is no reasoning there just the model telling you what it would like to do if it was able to. Or irrelevant noise. And it comes in fast, in chunks, too fast. I think you are hiding the real reasoning in all Gemini 3+ models for some reason and stream. If this really is the real reasoning then this model has no future.
It is still unable to perform complex tasks anyway. Nowhere near Opus 4.5, let alone 4.6. Just tweaked it to score on benchmarks and called it a day. Idk maybe it is just a model for frontend code.
But since it is unusable in Antigravity anyway, I guess I am staying with Claude Code for now lol.
2 months of Antigravity being pretty much unusable.