I made a general AI benchmark (opens in new tab)

A structured benchmark for evaluating AI models on arithmetic, logic, geography, science, history, and world knowledge. Open-source questions, leaderboard included.