Gemini felt dumb on October 7, 2025.

I was excited to use Gemini 2.5 for scripting my YouTube videos, and the first full‑script hit 132k views. But after the recent nerf, the same content now only pulls about 1k views per day. The drop feels disappointing and makes the tool seem far less useful than it once was, leaving me frustrated with the sudden loss of performance.

I’ve been playing with the new apps in ChatGPT and was instantly impressed by how smooth and polished everything feels. The integration is seamless, turning simple ideas into functional tools without a hitch. I found myself thinking the experience was so good I wish Google could copy it right away—this feels like a genuine leap forward in usability.

I tried to discuss a tricky subject with Gemini, and every time I nudged it toward the right answer it dug its heels in. It would concede only the obvious errors, but on harder points it kept insisting on its own mistaken view and wouldn't explore further. The stubbornness was frustrating and made the conversation feel like a dead end.

I tested SeeDream 4 by asking it to edit an image with a very specific street‑scene prompt—Caracas, tropical outfit, sunglasses, flip‑flops, a non‑chalant pose, and a menacing cane corso. The result blew my mind; it was absurdly detailed and way beyond what I expected, leaving me joking that Google needs a “Giga Banana” upgrade already. The output was so striking it felt like a glimpse of SeeDream 5 in the making.

I tried using Gemini and it was a nightmare – the responses were nonsensical, totally missed the point, and even produced bizarre or harmful output. Watching the video made me cringe; the tool’s behavior was absurdly wrong and the whole experience felt like a comedy of errors, leaving me frustrated and wary of relying on it again.

I keep hitting the “Something went wrong (9)” message after just a handful of queries in Flash, and it’s really slowing me down. I’m not sure if this is a bug that’s popping up temporarily or if the system is silently throttling me. The repeated crashes feel like an unnecessary roadblock, and I’m left wondering whether I’ve hit an intentional usage limit or just a flaky glitch.

I rely on attaching Google Docs to AI Studio for context, but when I edit those docs the changes never show up in the chat—even after refreshing the page. It’s frustrating because my knowledge base lives in those documents and needs to stay current. I can edit my own messages retroactively, so a simple “refresh” option for attached docs would solve this and make the tool actually usable.

I asked Gemini who the current FBI director was, hoping for a quick answer, but it insisted Kash Patel held the role—clearly wrong. When I tried to correct it, the model just stopped mentioning the topic altogether, leaving me frustrated and doubting its reliability for basic factual queries.