Berkeley Function Calling Leaderboard (BFCL)

The Berkeley Function-Calling Leaderboard (BFCL) is a benchmark dataset that contains problem instances used to evaluate the performance of large language model agents in complex, multi-turn tool-use tasks.

Papers using Berkeley Function Calling Leaderboard (BFCL) (10)

ToolACE: Winning the Points of LLM Function Calling2024 · 3 citesWorld Modelling Improves Language Model Agents2025 · 2 citesReasoning through Exploration: A Reinforcement Learning Framework for Robust Function Calling2025 · 1 citesOn the Robustness of Agentic Function Calling2025 · 1 citesBrief Is Better: Non-Monotonic Chain-of-Thought Budget Effects in Function-Calling Language Agents2026Awakening the Sleeping Agent: Lean-Specific Agentic Data Reactivates General Tool Use in Goedel Prover2026Don't Just Fine-tune the Agent, Tune the Environment2025Tinyllm: Evaluation And Optimization Of Small Language Models For Agentic Tasks On Edge Devices2025xLAM: A Family of Large Action Models to Empower AI Agent Systems2024 · 5 citesAsynchronous LLM Function Calling2024 · 1 cites