Customer service and tool use
Berkeley Function Calling Leaderboard (BFCL)
Function selection, argument generation, multi-turn tool use and current agentic tracks.
About Berkeley Function Calling Leaderboard (BFCL)
What it measures
Category-specific function-call correctness.
Scope & limitations
Separate model-only calling quality from a complete product agent. Report BFCL version and categories, not an unspecified aggregate.
Evaluation environment
Tool-call model adapter or instrumented agent interface.