Gemini 3 – amazing for the first 3 days, now making mistakes

The new release was amazing for the first few days, then it reverted to worse than 2.5 Pro.

I have been using Gemini chat primarily for Python / Jupyter coding questions – I switched back to Kilo Code w/ Sonnet 4.5 and GPT5.1 Codex bc I just could not get Gemini to provide correct answers to basic questions (not using it to analyze code base).

Is anyone else running into this problem?

Leave a Reply