Deep Search on XBench-DS v1.0 (test)
68Success RateOpenAI o3*
Evaluation Results
| Method | Links | |
|---|---|---|
| OpenAI o3*2026.06 | 68 | |
| OpenAI DeepResearch*2026.06 | 66.7 | |
| Claude-4.5-Sonnet*2026.06 | 66 | |
| TreeSeekerBackbone=gpt-5.22026.06 | 56.3 | |
| Gemini-2.5-Pro*2026.06 | 56 | |
| Kimi K2*2026.06 | 54 | |
| WebSailor-32B*2026.06 | 53.3 | |
| Gemini-DR*2026.06 | 53 | |
| Claude-4-Sonnet-Thinking*2026.06 | 53 | |
| Flash-SearcherBackbone=gpt-5.22026.06 | 50.7 | |
| Grok3 DeepResearch*2026.06 | 50 | |
| Doubao DeepResearch*2026.06 | 50 | |
| Qwen3-235B-A22B-Instruct-2507*2026.06 | 45.5 | |
| Tongyi-DeepSearch-30B-A3B2026.06 | 45 | |
| IterResearchBackbone=gpt-5.22026.06 | 44 | |
| WebDancer-32B*2026.06 | 39 | |
| DeepSeek-R1*2026.06 | 32.7 | |
| LATSBackbone=gpt-5.22026.06 | 31 | |
| OpenAI GPT-5*2026.06 | 30 | |
| Search-o1-32B*2026.06 | 25 | |
| TreeSeekerBackbone=gpt-4.12026.06 | 23 | |
| Flash-SearcherBackbone=gpt-4.12026.06 | 21.3 |