Could it be that ChatGPT is just good at plagiarizing existing answers? (because when I ask it technical questions that I understand well than the answers are mediocre at best: convincing at the surface level, and often nonsense if you dig deeper).
It's quite knowledge in general that you need to know leet code and that there is a known entry barrier and tbh there are plenty of people I know who can't do that.
If language model that couldn't do the work passes your test then it's a bad test. Especially if it rejects some people who could possibly do the work.