Benchmark
Berkeley Function Calling Leaderboard V4
Evaluate whether models use tools correctly across calls, conversations, search, and memory.
Published leaderboard
Explore scores and evaluation settings on the publisher’s leaderboard.
Open leaderboardCoverage
Single-turn and multi-turn function calling, relevance handling, web search, memory, and separate format-sensitivity analysis.
Methodology
Check function structure or environment state as defined by each task. V4 combines category scores using specified weights; format sensitivity is reported separately from the main score.