ResearchBenchmarksFunction Calling on BFCL v2Follow1Spearman Rho (FullPair Agreement)FullPair0.69840.77670.8550.9333May 28, 2026Evaluation ResultsMethodMethodLinksSpearman Rho (FullPair Agreement)Top-1 AgreementResolution Gain over OriginalFullPairProtocol=FullPairProtocol=FullPair2026.051—0.64SEALProtocol=SEALProtocol=SEAL2026.050.98—0.8ListwiseProtocol=ListwiseProtocol=Listwise2026.050.95—1.83PointwiseProtocol=PointwiseProtocol=Pointwise2026.050.93—0.7FixedRubProtocol=FixedRubProtocol=FixedRub2026.050.88—0.79FlatBrkProtocol=FlatBrkProtocol=FlatBrk2026.050.83—0.54OriginalProtocol=OriginalProtocol=Original2026.050.71—1