LM Arena

Neutral AI model evaluation platform using human preference data to benchmark and rank large language models across categories including math, coding, and instruction-following