Gemini felt dumb on March 13, 2026.

I tried asking Gemini models a simple current‑date riddle and they completely missed the point. Their answers were off‑base and confused, turning a straightforward query into a frustrating back‑and‑forth. The tool’s behavior felt baffling and wasted my time.

I tried writing a story with Gemini in AI Studio, but the model kept blocking any +18 scenes, even though I was the one reading it. The censorship was over‑zealous, making the flow choppy and forcing me to backtrack. I compared it to ChatGPT and wondered if Gemini will stay this limited or ever improve, because right now it feels more stupid than helpful.

I tried out the suggested prompt and was pleasantly surprised by how well the model answered. The responses were spot‑on, capturing the nuance I was aiming for without needing endless tweaks. I felt a mix of relief and excitement as the output matched my intent straight away, turning a tentative experiment into a smooth, productive workflow.

I tried to use Vertex AI Studio and the Model Garden Demo, but every attempt ends with a 500 internal error. The constant “Internal error encountered” message and failing request IDs make the tool completely unusable, leaving me frustrated and stuck until they fix it.