Gemini felt dumb on November 30, 2025.

I’ve been using Gemini 3 and it used to give solid answers, but lately it’s like the model’s been lobotomized. I keep noticing it lose track of the conversation, forget what it was doing, or even ignore my original question—sometimes in brand‑new chats. The experience feels sloppy and disappointing, making me waste time trying to get clear responses.

I asked Gemini 3 about antigravity, hoping for a clear explanation, but it kept jumping to the Chrome “antigravity” extension instead. Every reply felt off‑track, forcing me to rephrase repeatedly. The tool’s misunderstanding was irritating and made the conversation feel like a wasted effort, leaving me frustrated and amused at the same time.

I was expecting a relevant answer, but Gemini went completely off‑track and started talking about female reproductive organs even though I never asked about that. The randomness was surprising and a bit irritating, making the interaction feel unreliable and frustrating.

I’ve been a huge fan of Anti‑Gravity IDE, but over the last two days it’s turned into a nightmare. The UI drags and often freezes, and the AI agents that edit files now corrupt the code almost every time—even though the generated snippets look fine. It feels like the tool itself has broken, making it practically unusable, and I’m desperate for a fix.

I tried to get the model to produce a yelling scene, even simplifying the prompt to just “yelling,” because I wanted the left character to shout a line like “No one crosses our path!” but the AI kept refusing to generate the image I was after. The tool’s stubborn refusal was frustrating and left me feeling stuck, as I couldn’t get the simple output I needed.

I just upgraded to the Pro version, hoping to unlock the Nano Banan Pro picture‑generation feature, but nothing changed. The app still hides the option to create images, leaving me stuck and frustrated. It feels like the upgrade was useless, and I’m left wondering why the promised functionality is blocked despite paying for it.

I tried using Gemini 3 Pro’s “thinking” feature and ran into a bug that stopped it from working properly. The tool just froze/errored and I couldn’t get any output, which was really annoying because I needed it for my project. I felt frustrated and stuck, hoping the developers would fix the issue soon.

I tried using Gemini’s nano‑banana tool to strip the watermark from a photo we got at a theme park, hoping to avoid the $50 fee. Instead, the AI kept spitting out a weird excuse that it can’t edit photos of Britney Spears, even though my wife only vaguely resembles her. The nonsensical refusal was irritating and left me feeling the tool was useless for this simple task.

I dug into the old Gemini 2.0 image generator and lined it up against the Nano Banana previews. I posted a series of screenshots – the base character from Nano Banana, then Gemini 2.0’s take, followed by the original Nano Banana selfie‑angle mishap, the corrected angle in Nano Banana GA, and finally the Nano Banana Pro preview. The side‑by‑side view let me see the strengths and quirks of each model, especially how the angle issue was handled.

I’ve been testing Gemini 3 Pro for real‑time research, travel tips, and advice, and while the new model looks slick, it feels sluggish and only marginally better than GPT‑5.1. The slow responses make it hard to use on the fly, and the boost in quality doesn’t outweigh the lag. I’d love a faster “thinking‑only” version, especially since I’m on Android and love the seamless integration, but for now GPT‑5.1 still wins for practical use.

I was absolutely blown away when Gemini’s Nano Banana Pro churned out a banner that perfectly matched the vibe I needed for my Suno background. Every genre’s character blended into a seamless collage, even shaping the letters of my username. It felt like the tool had leapt into a whole new league—something no image generator could have pulled off just a week ago. This awe‑inspiring result left me speechless.