Gemini felt dumb on October 8, 2025.

I’m frustrated with Gemini’s current limits—context handling drops off, output isn’t stable, and speed feels sluggish. I wish it could call tools more reliably, especially for math where it still ignores my explicit request to run code, and improve its Google search integration instead of just regurgitating training data. Faster flash and pro would be a huge win.

I was experimenting with different prompts and suddenly Gemini started spitting out a massive, nonsensical message that filled the screen with countless photos. It was the first time I’d seen anything like that, and I wasn’t sure if it was a bug or just a weird glitch. The output felt jumbled and out of context, leaving me frustrated and questioning whether this was normal behavior.

I tried using guided learning and it turned out to be the best experience I’ve had with an AI tool so far. The responses were spot‑on and helped me grasp the material quickly. The only downside was that chatGPT sounded a bit condescending when I asked certain follow‑up questions, which left me slightly irritated, but overall the help was solid and encouraging.

I tried using Gemini on a simple derivation that was tied to a famous open problem, only to have it refuse outright. It said it was “dangerous” to do novel work, even though I’d already shown the result in my context. The block felt arbitrary—​the model can handle far harder math elsewhere, yet it won’t help with this trivial task, which was frustrating and felt like a missed opportunity.

I set Gemini as the default assistant on my S24 Ultra and turned on the “Google assistant features,” but every time I ask it to open an app it says it can’t. It works fine on my old A53, so the failure on the new phone is really frustrating.