Gemini felt dumb on January 5, 2026.

I added a personal context telling Gemini to prefer Vim over Nano, hoping it would just use Vim in the terminal. Instead, it now pops up with “since you’re a Vim user…” in any Linux‑related chat, even when we’re talking about unrelated topics. The constant reminders are irritating, and I’m looking for a way to phrase the context so the model stops bringing it up everywhere.

I’ve been chatting with Gemini 3 Pro for extended sessions, and it keeps crashing or freezing after a while. The interruptions are random, and I lose the context I built up, which is really annoying. I expected it to handle long conversations smoothly, but the tool’s instability makes it hard to rely on for anything beyond short queries.

I tried using AI Studio to summarize a YouTube video, only to hit a dead end with the message “An internal error has occurred.” The tool that normally helped me quickly digest content now stalls completely, leaving me frustrated and unable to get any summary at all. This glitch feels like a major setback, especially when I was counting on it for a tight deadline.

I tried using Gemini AI Studio’s “Restore” button expecting it to roll back both my code and the AI’s memory. It does revert the code, but the model still remembers later prompts, causing it to reference renamed functions or apply changes I never asked for. I end up having to manually tell it to ignore future prompts, which feels clunky and frustrating.

I tried asking Gemini 3.0 Pro Preview for a simple chess move, expecting a quick, correct answer. Instead it stalled for over four minutes and still handed me the wrong move. The delay was exasperating, and the incorrect suggestion made the whole interaction feel useless, leaving me frustrated and doubtful about its reliability.