AI Benchmarks evaluate the performance of artificial intelligence models through standardized tests, measuring capabilities in areas like reasoning, coding, and language understanding.