What's your go to question to check if newly released model is just codemaxxxed slop for the leaderboards or not?

I generally just ask it "describe the main cast of (insert somewhat known cartoon show from the 2010s)", could either be Totally Spies, Randy Cunningham, Slugterra etc, most models in the 30b range completely fumble, looking at you Qwen, but the ones that manage to answer that are gems that can actu