Gemini felt dumb on January 26, 2026.

I’ve been trying to use Flow for the past two days, but it’s basically dead—constant “high demand” messages, jobs getting stuck or canceled, and endless errors. It feels like I’m paying for a service that can’t even handle normal traffic, and the whole experience has been incredibly frustrating and unproductive.

I noticed that the AI Studio kept hitting capacity limits, which was pretty annoying. Every time I tried to start a session it stalled or failed, breaking my workflow. The tool's behavior was frustrating, though it wasn’t catastrophic—just a consistent inconvenience that slowed me down.

I tested Gemini 3.0 Pro’s context window in the web app and was disappointed. The model dropped everything from the start of my story, only recalling the last ten responses, even though the narrative was short. It felt like Google was falsely touting a massive context size while tightening rate limits, leaving me frustrated and skeptical.

I recorded a competitor’s site scrolling through product listings, then fed the video to Gemini 3 Pro’s Vision Data Engineer prompt. The model parsed each frame, de‑duplicated entries, ignored blurry shots, and spit out a clean CSV of product names and prices in seconds. Watching it work felt like turning a screen capture into instant, ban‑proof data—mind‑blowing efficiency.

I asked Gemini 3 Flash about its own version, and it blurted out that the model name was “incorrect or non‑existent,” insisting only 1.5 and 2.0 existed. It even warned me not to “invent” model names or I’d get API errors. The whole thing felt like an identity crisis—frustrating and oddly hilarious.

I tried to share a conversation in AI Studio using the built‑in share button, but it just kept loading and then threw an error. When I switched to the workaround of making the Drive file shareable, I could open the chat from another account, yet I couldn’t send any messages at all. The whole process was irritating and made the tool feel unreliable.