List your company

ResearchFormerly WizWand

Tool Use

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

Unified evaluation benchmark D Single-Turn (test)

93SP

9

Apr 14, 2026

Unified evaluation benchmark D Multi-Hop (test)

Claude 4.6 Sonnet

83.3SP

9

Apr 14, 2026

Unified evaluation benchmark D Single-Hop (test)

Qwen3-8B (Upper Bound)

96.1SP Score

9

Apr 14, 2026

Claude 3.5 Sonnet

90.2Accuracy

9

Feb 26, 2026

58.7Accuracy

9

Feb 26, 2026

Claude-Opus-4.5

55.3Overall Score

8

Jul 2, 2026

0.792Accuracy

8

Jun 30, 2026

Qwen2.5-Instruct-14B with TOUCAN-SFT + COVERT-RL

61.8Accuracy

8

Apr 14, 2026

BFCL Agentic v4 (out-of-distribution)

39Web-base Score

8

Feb 26, 2026

ACEBench-en (out-of-distribution)

77.9Normal Score

8

Feb 26, 2026

τ²-Bench (out-of-distribution)

54.9Retail Score

8

Feb 26, 2026

τ²-Bench telecom

98.7Score

7

Jul 2, 2026

69.9Live Success Rate

7

Mar 18, 2026

MAGNET-14B-mDPO

73.3Success Rate

7

Feb 26, 2026

Terminal Bench Hard

36Accuracy

6

Jul 7, 2026

81.3Score

6

May 28, 2026

ACEBench Normal

0.651Score

6

May 28, 2026

Robotic Simulation Tool Use

78Success Rate

6

Apr 30, 2026

TAU-2 & BFCL-V4 Multi-turn

Qwen3-235B-A22B-Instruct

52Average Score

6

Apr 24, 2026

89.4Instruction Accuracy

6

Apr 15, 2026

BEHEMOTH ToolBench (out-of-distribution)

26.82Success Rate

6

Apr 14, 2026

ToolBench standard evaluation

51.3Pass Rate

6

Mar 4, 2026

StableToolBench Overall Average

70.3SL (Success Rate)

6

Feb 26, 2026

StableToolBench G3 Instruction

66.3SL Score

6

Feb 26, 2026

StableToolBench G2 Instruction

68.8SL Score

6

Feb 26, 2026