Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

TauBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Multi-turn Agentic Tool-useTauBench-Airline out-of-distribution
Success Rate (SR)35.5
10
Multi-turn Agentic Tool-useTauBench Retail
Success Rate (SR)50.3
10
Agentic ReasoningTauBench V2
Airline Score76
9
Customer ServiceTauBench-Telephony (TBTel)
Accuracy90.5
8
Customer ServiceTauBench (TauB)
Accuracy91.8
8
Domain-specific agentic reasoningTauBench V3
Airline Accuracy85.8
7
AgenticTauBench V2
Airline Score56.9
5
Agentic ReasoningTauBench Telecom V2
Accuracy42.7
2
Showing 8 of 8 rows