What prompt would you use to truly test an AI model?

We built DualMind Arena so humans can judge AI models blindly instead of trusting benchmark leaderboards.

But we need your help.

What is the single hardest or most interesting prompt you would throw at two AI models to really expose which one is better? Could be a coding challenge, a creative brief, a logic puzzle, a debate topic anything.

Drop it in the comments and we will run the best ones as featured battles on dualmindlab.tech

Also would love to know what models you want to see added next. Currently thinking about expanding beyond the usual suspects. What would make this useful for your actual workflow?