I'll bite. The benchmark is actually pretty good. It shows in an extremely comprehensible way how far LLMs have come. Someone not in the know has a hard time understanding what 65.4% means on "Terminal-Bench 2.0". Comparing some crappy pelicans on bicycles is a lot easier.