Gemini felt dumb on September 16, 2025.

I tried using nano-banana to edit a photo of my high school’s mural. The AI struggled to remove the glass reflection and fill in the missing parts, even hallucinating angels out of nowhere. The result looked off-putting and unhelpful, leaving me frustrated that the tool couldn’t handle such a basic image correction.

I finally had a bad day with Nano-banana. It used to flood me with perfect colorized images, but now quality drifts and I have to keep hitting retry. The throttling is brutal—just under ten shots a day, nothing good. Gemini’s Chrome interface freezes, no chat, no upload, while Opera stays fine. It feels like the tool’s been intentionally softened, and I’m left feeling frustrated and betrayed.

I’m really frustrated because Gemini’s visual block suggestions keep generating useless diagrams from Nano Banana and Imagen 4—despite hours tweaking prompts, the images are unusable. I had to abandon them and go back to Photoshop, which takes forever. I’d love to hear if anyone else’s AI diagram tools worked better or have any tricks to try.

I was so excited, hoping the AI could swap my awkward face, smile, and hair in a key photo, but every time I hit save, the output was just the original image I fed in. I tried different prompts, added more detail, even stuck in a loop. It felt like I was talking to a wall—frustrating, confusing, and honestly a bit disheartening.

I was excited to try Gemini Pro for image creation, so I started making a few videos with veo3. But the tool kept giving me explanations about veo and nano instead of generating images. Every time I asked, I got the same broken, "dogshit" response—no visuals, just nonsense. It was frustrating and felt like a major let‑off from the promised capability.

I tried using Gemini Code Assistant and felt it wasn’t up to par. I wanted a solid, ready‑to‑go tool like GPT‑5 Codex, but the experience was underwhelming. It’s frustrating that I’d need an extra subscription just to get the Code Assistant on the Pro plan, and the support feels lagging. I wish it were tighter and more integrated.

I tried asking Gemini about a coding question, and it stuck in a loop, repeating the same vague answer over and over. It felt like the model was stuck in a feedback loop, not understanding or advancing the conversation. The frustration was real—my time was wasted, and it wasn’t even close to helpful. I just hope they fix this soon.

I’ve been trying to use Gemini Pro to generate and edit images over the past few days, but every time I hit the button the tool never lets me download the finished picture. The logo keeps spinning, the “save as” workaround only works until I refresh, and the images either vanish or come back corrupted. I feel stuck, losing my work and frustrated that the AI keeps crashing on a basic task.

I asked Gemini about the Samsung Galaxy S25 Ultra and it told me that the phone doesn't exist, which threw me off because the latest model is the S24 Ultra. I was like, “What’s going on?” It didn’t even give me a correct answer or a helpful suggestion—just a wrong fact. I felt frustrated because I expected a quick, accurate update, but it just repeated misinformation.

I stumbled onto Nightjar and was blown away by how it auto‑generated product shots – even of a literal trash can – that looked stunning. I didn’t type any prompt, yet the tool produced polished, eye‑catching images instantly. The whole process felt effortless and surprisingly creative, turning mundane objects into showcase pieces.

I prompted Nano Banana to generate a specific character image, but every time the result was a random, unrelated picture—furniture, historical photos, a woman's face, a hat, and more. I felt like I was receiving someone else's images, not mine. It’s annoying and makes me doubt the tool’s reliability, but it's not hazardous—just clearly not delivering what I asked.

I keep trying to ask about public figures, and every time I mention even part of a name, the model just blocks me. It feels like the system suddenly decided to shut down all historical or contemporary references. I’m bumping my head against a wall, wondering if this is a bug or a deliberate policy change—frustrating and confusing.

I tried just tweaking a few files, but the Build tool keeps scrambling through endless error fixes, looping over code I didn’t even have in the app. It’s annoying how the AI keeps hunting bugs in the wrong place, making me waste time chasing it rather than completing my original task. Frustrating and slow.

I bought Gemini Deepthink hoping to dive straight into big projects, but the first thing I ran into was a complete failure. I kept seeing error screens, the app wouldn’t launch, and customer help was vague. My entire workflow stalled—hours lost, deadlines missed, and the frustration was real. I need a refund before I morphed into a full‑on tech‑horror story.

I felt frustrated when Gemini 2.5 Pro threw me a curveball. I first asked it to transcribe a screen recording, it seemed fine, but when I switched to the “Share screen with Live” feature it started thinking I was speaking and pulled up the wrong context. Even after I clarified, it stuck with the old session. I hadn’t asked to save anything, yet it behaved like a glitchy memory system. The abrupt shift and confusion made the whole experience feel unreliable and annoying.