Interesting. I tried the same question (each a new conversation, GPT4) and 10V and above it catches, but 8V seems to be in its blind spot. Maybe it hasn't learned anywhere to take absolute maximum ratings seriously?
It's clearly missing experience with magic smoke.