Is the o3-mini-high 0.8% hallucination rate claim real — and what it means for accuracy and reasoning vs GPT-5

https://technivorz.com/how-a-12m-healthtech-startup-nearly-triggered-a-hospital-incident-using-an-llm/

Which questions I’ll answer and why they matter for model choice and risk Organizations see a headline number like "0.8% hallucination" and want to buy into that promise. You should not