Long-Horizon Search Intelligence on BrowseComp-zh
33.3Pass@1WebSailor-30B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WebSailor-30BRL=ReSum-GRPO, Paradigm=ReSum2025.09 | 33.3 | 48.8 | |
| WebExplorer-8BRL=GRPO, Paradigm=ReAct2025.09 | 32 | — | |
| DeepDive-32BRL=GRPO, Paradigm=ReAct2025.09 | 29.7 | — | |
| WebSailor-30BRL=GRPO, Paradigm=ReSum2025.09 | 29.3 | 42.6 | |
| WebSailor-30BRL=MEM1-GRPO, Paradigm=MEM12025.09 | 29.1 | 45 | |
| WebSailor-7BRL=ReSum-GRPO, Paradigm=ReSum2025.09 | 27.1 | 39.5 | |
| WebSailor-7BRL=GRPO, Paradigm=ReSum2025.09 | 25.4 | 40.8 | |
| WebSailor-30BRL=-, Paradigm=ReAct2025.09 | 23.9 | 38.4 | |
| WebSailor-30BRL=GRPO, Paradigm=ReAct2025.09 | 23.3 | 36.7 | |
| WebSailor-3BRL=ReSum-GRPO, Paradigm=ReSum2025.09 | 20.5 | 34.3 | |
| WebSailor-7BRL=GRPO, Paradigm=ReAct2025.09 | 18.7 | 31.8 | |
| WebSailor-3BRL=GRPO, Paradigm=ReSum2025.09 | 17.3 | 29.1 | |
| MiroThinker-32Bv0.1RL=DPO, Paradigm=ReAct2025.09 | 17 | — | |
| ASearcher-QWQ-32BRL=GRPO, Paradigm=ReAct2025.09 | 15.6 | — | |
| MiroThinker-8Bv0.1RL=DPO, Paradigm=ReAct2025.09 | 13.6 | — | |
| WebSailor-7BRL=-, Paradigm=ReAct2025.09 | 13.2 | 25.6 | |
| WebSailor-3BRL=GRPO, Paradigm=ReAct2025.09 | 11.8 | 22.5 | |
| WebSailor-3BRL=-, Paradigm=ReAct2025.09 | 8.2 | 17 |