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

Agentic Task Completion on τ2-Bench

84Airline Success Rate

Claude-Opus-4.5

4.9625.484666.52Apr 7, 2026Apr 16, 2026Apr 25, 2026May 5, 2026May 14, 2026May 23, 2026Jun 2, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
8479.670.292.324.7
2026.05
8485.172.995.626.9
2026.05
80.567.555.561.612.4
7650.948.156.19.3
2026.04
4448.247--
2026.04
3840.439.6--
344340.3--
3240.437.8--
2026.04
3241.238.4--
2026.06
3034.229.323.7-
2026.04
2834.232.3--
2026.06
2842.128.615.8-
2026.06
26.835.12513.2-
2026.06
2616.721.822.8-
2026.04
2436.832.9--
2026.06
2428.925.824.6-
2026.06
2247.431.324.6-
2026.06
1612.31516.7-
2026.06
849.124.917.5-