How good are the top frontier LLMs at finding bugs? I ran a small experiment to find out — and yes, the results surprised me!