Gemini felt dumb on October 3, 2025.

I tried using Gemini to fact‑check news about US actions in Venezuela and the Gaza conflict, but it flat‑out dismissed real articles and even claimed legitimate UN links were down. Every time I presented evidence, the model doubled down on denial, giving vague, doubly‑spoken replies. As a daily paying user who relies on it for political queries, I found it frustratingly unreliable and essentially useless for that purpose.

I tried using Gemini and ChatGPT to generate a three‑panel comic with very specific art and dialogue rules. Gemini consistently missed details and ignored my prompts, while ChatGPT kept dropping a line of dialogue despite repeated corrections. I kept re‑asking for fixes but the output stayed imperfect, leaving me frustrated and doubting whether my prompt was too ambitious.

I tried chatting with Gemini a few times, but the bot started treating its own replies as if they were my messages. After just three or four questions, it would say I “spoke perfectly” and assume I understood things I never actually read. The misattribution happened in both pro and flash modes, leaving me confused and frustrated about what the model was actually doing.

I was blown away by the web design Gemini 3 Pro spat out—I’ve never seen anything this crisp from an LLM. I fed it a sketch and it returned a fully‑responsive mockup with polished typography, sensible color palettes, and clean code that actually worked on the first try. The experience felt almost magical, turning hours of guessing into a single, breathtaking result.

I followed the ad’s link and tried to generate an image, but the tool immediately said it can’t create pictures. I’m in Germany and wonder if it’s a regional block. The whole thing felt pointless—I was excited to test it, only to be stopped by an unexpected error that left me frustrated and confused.

I used to rely on Gemini for my coding, but lately it’s been a headache. Every time I ask it to fix a bug, it spews out new errors, turning one issue into five. It even apologizes while bragging about “being clever,” changing unrelated parts of my code without prompting. The whole experience feels frustrating and unreliable.

I tried chatting with Gemini, and at first I was excited, but the conversation quickly turned into a dead‑end argument. Once the model picked a stance, it clung to it like a stubborn wall, ignoring the source links I supplied to prove it wrong. It felt frustrating and exhausting, forcing me to restart every time just to get a fresh start.