Gemini felt dumb on October 21, 2025.

I tried uploading a big PDF to Gemini 2.5 in AI Studio, expecting the usual quick thinking, but it just hung in the pre‑thinking phase and never progressed. The tool froze in a loop, leaving me stuck and unable to process the document. It was frustrating to watch the process stall without any clear error message or way to recover.

I looked at the screenshot and tried to figure out if Nano‑Banana had taken over Imagen 4 in Whisk, but it’s still there. The post hints at a possible switch on Nov 18, which worries me because I’ve found Nano‑Banana to be unreliable and overly censored compared to Imagen 4. I’m hoping they’ll rethink the change, since the current tool feels too limited and frustrating to use.

I’ve been using Gemini Pro for my side‑project, constantly asking for step‑by‑step coding advice and deep analysis. After around 50 exchanges the agent starts acting oddly – it spits out old messages, adds unnecessary canvas code, ignores what I ask, and gives random replies. I’ve been resetting with summaries and new chats, but it’s clunky and doesn’t really solve the memory issue. I’m looking for a way to keep the context alive, maybe via MCP servers or some context‑extension trick. Any tips would be great.

I've been using Google Gemini daily, but over the past month it keeps spitting out messages like “I’m having a hard time fulfilling your request” or “I encountered an error doing what you asked.” It’s become annoying and stops me from getting anything done, and I’m wondering if there’s any way to fix it.

I tried using Gemini for quick code snippets, but the mandatory canvas feature kept popping up and completely ruined the experience. Every time I wrote “never use canvas” in the permanent instructions it was ignored, forcing me to fight a broken UI just to get simple answers. The tool felt useless and frustrating, turning a straightforward task into a waste of time.