Web Agent Task Completion on XBench DeepSearch
79Accuracy
SlimSearcher (SFT+RL)
Updated 1mo ago
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
2026.06 | 79 | 5.92 | 4,888 | |
2026.06 | 78 | 14.06 | 4,466.5 | |
2026.06 | 77 | 10.87 | 5,092 | |
2026.06 | 77 | 10.22 | 5,155 | |
2026.06 | 74 | 13.44 | 5,515 | |
2026.06 | 71.3 | 14.26 | 6,918 | |
2026.06 | 67.6 | 12.5 | 6,321 | |
2026.06 | 67 | - | - | |
2026.06 | 64.6 | - | - | |
2026.06 | 55 | 15.04 | 2,395 | |
2026.06 | 54 | 5.98 | 1,316 | |
2026.06 | 53 | 2.23 | 422 | |
2026.06 | 51.7 | 9.05 | 2,246 | |
2026.06 | 49 | 8.84 | 938 |