LMArena Leaderboard – A crowdsourced benchmark for Large Language Models (LLMs) using a blind, head-to-head “battle” format. It ranks models via human preference and an Elo rating system to reflect real-world performance.