Agentic Search on BrowseComp (Score)
77.3ScoreSeed-2.0-Pro
Evaluation results
| Method | Links | |
|---|---|---|
| Seed-2.0-ProParam=-, Training=-2026.05 | 77.3 | |
| MiroThinker-1.7-miniParam=30B, Training=CPT + SFT + RL2026.05 | 67.9 | |
| Claude-Opus-4.5Param=-, Training=-2026.05 | 67.8 | |
| DeepSeek-V3.2Param=671B, Training=-2026.05 | 67.6 | |
| GLM-4.7Param=358B, Training=-2026.05 | 67.5 | |
| LongSeekerParam=30B, Training=SFT2026.05 | 61.5 | |
| Gemini-3.0-ProParam=-, Training=-2026.05 | 59.2 | |
| RedSearcherParam=30B, Training=CPT + SFT + RL2026.05 | 57.4 | |
| MiroThinker-1.5-miniParam=30B, Training=CPT + SFT + RL2026.05 | 56.1 | |
| GPT-5Param=-, Training=-2026.05 | 54.9 | |
| Tongyi-DeepResearchParam=30B, Training=CPT + SFT + RL, Context Management=None2026.05 | 43.4 | |
| IterResearchParam=30B, Training=SFT + RL2026.05 | 37.3 | |
| AgentFoldParam=30B, Training=SFT2026.05 | 36.2 | |
| OpenSeekerParam=30B, Training=SFT, Context Management=None2026.05 | 29.5 | |
| ReAct + GraphPOBase Model=Qwen2.5-7B, Agent Framework=ReAct2026.06 | 3.7 | |
| ReAct + Tree-GRPOBase Model=Qwen2.5-7B, Agent Framework=ReAct2026.06 | 2.5 | |
| ReAct + DAPOBase Model=Qwen2.5-7B, Agent Framework=ReAct2026.06 | 2.4 | |
| ReAct + GRPOBase Model=Qwen2.5-7B, Agent Framework=ReAct2026.06 | 2.3 | |
| ReActBase Model=Qwen2.5-7B, Agent Framework=ReAct2026.06 | 1.4 |