Berkeley Function Calling Leaderboard
A function-calling and tool-use benchmark covering single-turn, multi-turn, live, and agentic scenarios.
What this benchmark measures
A function-calling and tool-use benchmark covering single-turn, multi-turn, live, and agentic scenarios.
Rows on this page are sourced from public benchmark artifacts, leaderboard exports, or source-linked model reports. Each row keeps benchmark version, source model name, and available run details attached to the score.
The metric shown here is accuracy. It should be interpreted within Berkeley Function Calling Leaderboard, not compared as part of a site-wide ranking.
What to be careful about
BFCL includes source-provided within-benchmark aggregates; label them as BFCL metrics, never evals.report composites.
Frequently asked
What is Berkeley Function Calling Leaderboard?
A function-calling and tool-use benchmark covering single-turn, multi-turn, live, and agentic scenarios. It is a tool use benchmark measured by accuracy.
What does accuracy mean on Berkeley Function Calling Leaderboard?
Berkeley Function Calling Leaderboard reports accuracy (%); higher is better. Scores are shown only within Berkeley Function Calling Leaderboard and are never averaged with other benchmarks.
What is the top reported Berkeley Function Calling Leaderboard score?
Claude Opus 4.5 has the top reported score on Berkeley Function Calling Leaderboard: 77.47% (accuracy).
Why do Berkeley Function Calling Leaderboard scores differ across runs?
Harness, scaffold, reasoning effort, and prompt setup change results, so two runs of the same model can differ. evals.report keeps each score with its run context so the differences stay visible.
Does evals.report rank models across benchmarks?
No. Berkeley Function Calling Leaderboard scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".