We introduce the Berkeley Function Leaderboard (BFCL), the first comprehensive and executable function call evaluation dedicated to assessing Large Language Models' (LLMs) ability to invoke functions.