Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Web Agent Navigation on Mind2Web OOD (test)
Loading...
51.56
Step Accuracy
SkillFlow
17.4272
26.2886
35.15
44.0114
May 13, 2026
Step Accuracy
Action F1 Score
Updated 2mo ago
Evaluation Results
Method
Method
Links
Step Accuracy
Action F1 Score
SkillFlow
Method Type=Ours, Back...
2026.05
51.56
84.89
FlowSteer
Method Type=Agent+RL,...
2026.05
41.41
75.22
AgentFlow
Method Type=Agent+RL,...
2026.05
39.84
71.77
SkillRL
Method Type=Agent+RL,...
2026.05
35.16
69.54
AFlow
Method Type=Search-bas...
2026.05
30.47
62.29
v4-flash
Method Type=Baseline,...
2026.05
27.57
66.21
Qwen3.5
Method Type=Baseline,...
2026.05
26.64
63.85
Qwen3.5
Method Type=GRPO, Back...
2026.05
23.51
66.98
Qwen3.5
Method Type=SFT, Backb...
2026.05
18.74
50.49
Feedback
Search any
task
Search any
task
