Agentic Tool-use

Benchmarks

Dataset Name	SOTA Method	Metric
Tau2-Bench		Retail Score90.4	59	1mo ago
tau2-bench Airline	CODEDELEGATOR	Pass@163.5	30	2mo ago
tau2-bench Retail	CODEDELEGATOR	Pass@182	30	2mo ago
Agentic Macro-aggregate	Uno-Orchestra	Pass@170.3	22	2mo ago
MCP-Bench 3-server 1.0		Overall Score3.95	20	1mo ago
MCP-Bench Single 1.0		Overall Score3.54	20	1mo ago
AppWorld (Challenge)	RCL (all primitives)	TGC83.7	20	3mo ago
AppWorld Normal	RCL (all primitives)	Task Goal Completion (TGC)89.3	20	3mo ago
ACEBench (agent-task)	DFSDT + EvoSOP	Average Success Rate85.8	20	16d ago
τ²-Bench Telecom		Accuracy100	18	2mo ago
τ2-Bench (Tau-bench) Retail and Telecom		Overall Success Rate85.79	17	4mo ago
ACEBench	GEAR	ACE-E Score37.5	16	1mo ago
BFCL	GEAR	BFCL Average Score94.1	16	2mo ago
ToolSandbox Multi-Tool	GEAR	TS-M Score53.7	16	2mo ago
Tau-Bench	Qwen3-235B-Instruct-2507	Retail Score71.3	13	4mo ago
τ²-Bench Retail	Seed2.0 Pro	Accuracy90.4	10	23d ago
tau^2 Bench official evaluation setting GPT-4.1 simulator	REACT(GPT-5)	Retail Score0.775	9	4mo ago
Tau2-Telecom	LongCat-Flash-Lite	Avg@872.8	8	3mo ago
Tau2 Retail	LongCat-Next	Avg@873.68	8	3mo ago
Tau2-Airline	LongCat-Flash-Lite	Avg@858	8	3mo ago
τ²-bench telecom v1 (test)		Pass@185.9	6	25d ago
τ²-bench airline v1 (test)		Pass@153.5	6	25d ago
Multi-step tool-calling agent tasks Aggregate over 4 models: Nova 2 Lite, Nova 2 Pro, Claude 3.5 Haiku, Claude Sonnet 4.6 (test)	Gold CMTF	Success Rate99	5	1mo ago
τ²-Bench Airline		Accuracy67.5	5	3mo ago
SpreadSheetBench		Success Rate55.36	5	4mo ago

Showing 25 of 29 rows