Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Function Calling research benchmarks – Page 5 · SOTA2 Research

Function Calling

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Granite-4.1-8B + PROVE

33.5Score

4

Jun 4, 2026

BFCL unseen (test)

48.5Accuracy

4

Feb 26, 2026

BFCL rand (test)

56.6Accuracy

4

Feb 26, 2026

BFCL non-live (test)

82AST Accuracy (Simple Python)

4

Feb 26, 2026

Tool-call benchmark (n=200)

TA-Fast-Semantic

100Accuracy (12-char)

3

Apr 14, 2026

91.2Structural Accuracy

3

Feb 26, 2026

AISA-AR-FunctionCall Maghrebi 1.0 (test)

AISA-AR-FunctionCall-FT

61.58Function Name Accuracy

2

Mar 19, 2026

AISA-AR-FunctionCall Levantine 1.0 (test)

AISA-AR-FunctionCall-FT

69.48Function Name Accuracy

2

Mar 19, 2026

AISA-AR-FunctionCall Egyptian 1.0 (test)

AISA-AR-FunctionCall-FT

68.34Function Name Accuracy

2

Mar 19, 2026

AISA-AR-FunctionCall Gulf 1.0 (test)

AISA-AR-FunctionCall-FT

69.72Function Name Accuracy

2

Mar 19, 2026

AISA-AR-FunctionCall MSA 1.0 (test)

AISA-AR-FunctionCall-FT

76.13Function Name Accuracy

2

Mar 19, 2026

Function Calling in-domain

vocabulary trimming

364.54Latency (s)

2

Mar 6, 2026

84.4Accuracy (Single)

2

Feb 26, 2026

89.6Score (Single)

2

Feb 26, 2026

93.1Tool Match (TM)

2

Feb 26, 2026

—

—Score

0

Feb 26, 2026

Page 5 of 5

Popular datasets

BFCL Berkeley Function Calling Leaderboard AISA-AR-FunctionCall ToolBench ACEBench

Follow for updates

Join SOTA2 Contact