The Gemeni 3.1 modle is sooo dumb, that i dont have words to explain its stupidity !

[x7dl8p]

they changed the architecture or something, now it worst then gpt 5.1 mini, in vs code gpt 5.1 mini, a free model is more intelligent then gemeni, anybody else experiencing the same ?

[aniruddhaadak80]

From my point of view, posts like this become much easier for others to engage with when they include one or two concrete regressions instead of only a general quality judgment. If you share the exact model, the prompt, the response that looked wrong, and what an earlier version did better, maintainers and other users can actually compare behavior. Right now it is clear that you are frustrated, but it is hard to turn that into something actionable.