Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
GUI Navigation on OSWorld
Loading...
28.2
Accuracy
BEAP-Agent
4.072
10.336
16.6
22.864
Jan 29, 2026
Accuracy
Backtracking Task Rate
Backtrack Success Rate
Avg Backtrack Steps
Updated 4d ago
Evaluation Results
Method
Method
Links
Accuracy
Backtracking Task Rate
Backtrack Success Rate
Avg Backtrack Steps
BEAP-Agent
Model=GPT-4o + UI-TARS...
2026.01
28.2
35.8
65.5
2.72
Agent S2
Model=GPT-4o + UI-TARS...
2026.01
26.6
-
-
-
BEAP-Agent - w/o Backtrack
Model=GPT-4o + UI-TARS...
2026.01
26.3
-
-
-
JEDI
Model=GPT-4o + JEDI-7B...
2026.01
25
-
-
-
UI-TARS
Model=UI-TARS-1.5-7B,...
2026.01
24
-
-
-
BEAP-Agent - w/o Tracker
Model=GPT-4o + UI-TARS...
2026.01
23.6
-
-
-
AGUVIS
Model=GPT-4o + AGUVIS-...
2026.01
17
-
-
-
Qwen2.5
Model=Qwen2.5-vl-72B,...
2026.01
8.8
-
-
-
OpenAI
Model=GPT-4o, Steps=-
2026.01
5
-
-
-
Feedback
Search any
task
Search any
task