Deep Research on HotpotQA (HOT) (test)
92.8Mean Answer RateWebSailor
Evaluation results
| Method | Links | |
|---|---|---|
| WebSailorType=Live web2025.10 | 92.8 | |
| Our best agentType=Live web2025.10 | 92 | |
| ASearcherType=Live web2025.10 | 84.8 | |
| Our base agentType=Live web2025.10 | 80.3 | |
| DeepResearcherType=Live web2025.10 | 79.8 | |
| Search-R1Type=Static & simulated web2025.10 | 72.8 | |
| R1-SearcherType=Static & simulated web2025.10 | 62.6 | |
| ZeroSearchType=Static & simulated web2025.10 | 13.8 |