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

Deep Search on xbench-DS

75Accuracy

Qwen3-30B-A3B-thinking-SFT + SAPO

-0.9218.7938.558.21May 28, 2026
Updated 5d ago

Evaluation Results

MethodLinks
2026.05
75
2026.05
69
2026.05
67
2026.05
66.7
2026.05
53
2026.05
50
2026.05
48.5
2026.05
22
2026.05
20
2026.05
18
2026.05
16
2026.05
13.3
2026.05
11
2026.05
9.3
2026.05
5
2026.05
2