BFCL

The 'BFCL' dataset/benchmark contains 400 function-calling tasks used to evaluate operational metrics such as determinism, reliability, security, and cost in the context of LLM-based code generation for enterprise workflows.

Papers using BFCL (3)

ParaTool: Shifting Tool Representations from Context to Parameters2026The Cold-Start Safety Gap in LLM Agents2026Compiled AI: Deterministic Code Generation for LLM-Based Workflow Automation2026