It feels like you secretly rolled out an update. The model was working perfectly fine for me, but suddenly its intelligence has dropped drastically. It keeps forgetting and dropping context, even forgetting or altering content from the last chat turn, or even the current conversation. This issue started this morning, and it’s happened before too—it got better after a few days. I was enjoying using it, but now it’s so dumbed down that it can’t handle long-context conversations at all.
I’ve been experiencing the same issue since the weekend. It’s worse in long-context conversations, but even in short-context I’m seeing hallucinations that weren’t there before—things like assuming or fabricating entities that were never provided, or getting detailed naming conventions wrong. This is a clear regression from how it was performing just days ago.
I don’t know wheter it’s related, the UI of my Gemini PRO account updated recently (some buttons were changed and an option called extended reosing appeared); ever since then the model feels worse. It’s answers are very generic and it no longer follows instructions on complex tasks
Same issue. It cant follow basic instructions..keeps hallucinating and referringto old stuff