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

Agentic Task Performance on τ2-Bench Airline 1.0 (test)

96.4CAP

NOD

Updated 21d ago

Evaluation Results

Method	Links
NOD 2026.05		96.4	47.3
REVISE-only 2026.05		96	48.7
NOD 2026.05		96	52
Self-aware Abstention 2026.05		95.3	38
Weak Director 2026.05		95.3	43.3
Self-Reflection 2026.05		94.7	46
Self-aware Abstention 2026.05		93.7	39.3
Weak Director 2026.05		92.6	37.3
Vanilla LLM 2026.05		91.9	43.3
Self-aware Abstention 2026.05		91.7	43.3
REVISE-only 2026.05		91.4	50.7
Vanilla LLM 2026.05		91.2	42.7
NOD 2026.05		90.7	43.3
REVISE-only 2026.05		90.3	42
NOD 2026.05		90.3	42.7
Weak Director 2026.05		90	40
Weak Director 2026.05		89.7	48
NOD 2026.05		89.3	40.7
AutoGen 2026.05		89.1	40.7
NOD 2026.05		89	36.7
AutoGen 2026.05		88.8	44.7
REVISE-only 2026.05		88.8	42
Multi-Agent Debate 2026.05		88.7	35.3
Multi-Agent Debate 2026.05		88.3	40.7
REVISE-only 2026.05		87.2	42.7
Multi-Agent Debate 2026.05		87	48
Weak Director 2026.05		87	35.3
AutoGen 2026.05		86.9	42.7
Weak Director 2026.05		86	36
REVISE-only 2026.05		85.9	44
AutoGen 2026.05		85.4	37.3
Self-Reflection 2026.05		84.7	36
Self-aware Abstention 2026.05		84.7	30.7
Self-Reflection 2026.05		84.3	46
Self-Reflection 2026.05		84.3	41.3
Multi-Agent Debate 2026.05		84.3	39.3
Vanilla LLM 2026.05		83.8	36.7
Self-Reflection 2026.05		83.6	34
Self-aware Abstention 2026.05		83.4	37.3
AutoGen 2026.05		82.3	31.3
Self-aware Abstention 2026.05		78.3	32.7
Multi-Agent Debate 2026.05		78.3	34
AutoGen 2026.05		77.2	29.3
Vanilla LLM 2026.05		76.7	30
Vanilla LLM 2026.05		76.2	31.3
Self-Reflection 2026.05		64.2	22
Multi-Agent Debate 2026.05		63.9	18.7
Vanilla LLM 2026.05		63.5	21.3