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

Agentic Tool-use on ACEBench (agent-task)

85.8Average Success Rate

DFSDT + EvoSOP

0.88422.929544.97567.0205Jun 17, 2026Jun 20, 2026Jun 24, 2026Jun 27, 2026Jul 1, 2026Jul 4, 2026Jul 8, 2026
Updated 16d ago

Evaluation Results

MethodLinks
2026.07
85.8---
2026.07
85---
2026.07
84.2---
2026.07
84.2---
2026.07
84.2---
2026.07
83.3---
2026.07
82.5---
2026.07
80.8---
2026.07
78.3---
2026.07
75---
2026.07
74.2---
2026.07
60---
2026.07
53.3---
2026.07
45.8---
11.1514.38-
11157-
9.1513.35-
2026.06
8.8211.36.33-
8.65125.3-
4.153.35-
2026.01
-64.1710082.09
2026.01
-52.0910076.05
2026.01
-65.8394.3880.11
2026.01
-64.1788.7576.46
2026.01
-66.679580.84
2026.01
-97.54068.75
2026.01
-69.1792.580.84
2026.01
-72.587.580
2026.01
-57.56058.75
2026.01
-55.8363.7559.79
2026.01
-45.8357.551.67
2026.01
-54.1783.7568.96
2026.01
-6083.7571.88