Deep Search and Research Reasoning on xbench DeepSearch 2510 (Pass@1)
75Pass@1 AccuracyTongyi DeepResearch
Evaluation Results
| Method | Links | |
|---|---|---|
| Tongyi DeepResearchAgent Type=DeepResearch Agent, Configuration=30B-A3B2025.10 | 75 | |
| DeepSeek-V3.1Agent Type=LLM-based ReAct Agent2025.10 | 71 | |
| GLM 4.5Agent Type=LLM-based ReAct Agent2025.10 | 70 | |
| Kimi ResearcherAgent Type=DeepResearch Agent2025.10 | 69 | |
| OpenAI o3Agent Type=LLM-based ReAct Agent2025.10 | 67 | |
| Claude-4-SonnetAgent Type=LLM-based ReAct Agent2025.10 | 65 | |
| Kimi K2Agent Type=LLM-based ReAct Agent2025.10 | 50 |