That moment changed everything about Google Gemini 2.0 Flash accuracy vs GPT-4o: why I stopped trusting AI summaries blindly
https://64v80.stick.ws/
I used to accept model summaries at face value. One tidy paragraph and I moved on. Then I saw a supposedly 0.7% hallucination claim for Gemini-2.0-Flash-001 and realized the number meant very little without context