Deep Search on BrowseComp-ZH
66.6AccuracyGLM-4.7
Evaluation Results
| Method | Links | |
|---|---|---|
| GLM-4.7Context Budget=-2026.05 | 66.6 | |
| TaSType=MA, Backbone=GPT-5 Medium-Think2026.02 | 63.7 | |
| GPT-5 High-ThinkType=N/A2026.02 | 63 | |
| GPT-5 Medium-ThinkType=MA2026.02 | 62.9 | |
| OpenAI-o3Context Budget=-2026.05 | 58.1 | |
| GPT-5 Medium-ThinkType=SA2026.02 | 56.5 | |
| MiroThinker-v1.0-72BType=SA2026.02 | 55.6 | |
| MiroThinker-v1.0-30BType=SA2026.02 | 47.8 | |
| Tongyi DeepResearch (30B)Type=SA2026.02 | 46.7 | |
| Qwen3-30B-A3B-thinking-SFT + SAPOContext Budget=128K2026.05 | 45.7 | |
| GLM-4.6Context Budget=-2026.05 | 45.1 | |
| Tongyi DeepResearchContext Budget=128K2026.05 | 43.2 | |
| OpenAI Deep ResearchType=N/A2026.02 | 42.9 | |
| MiroThinker-v1.0-8BType=SA2026.02 | 40.2 | |
| TaSType=MA, Backbone=Qwen3-Max2026.02 | 35.3 | |
| TaSType=MA, Backbone=Gemini-2.5-Flash2026.02 | 34.9 | |
| Qwen3-MaxType=MA2026.02 | 34.3 | |
| Qwen3-30B-A3B-thinking-SFT + GRPOContext Budget=128K2026.05 | 33.2 | |
| Claude-4-Sonnet (Thinking)Type=SA2026.02 | 29.1 | |
| Kimi K2Context Budget=-2026.05 | 28.8 | |
| Gemini-2.5-FlashType=MA2026.02 | 28.4 | |
| Gemini-2.5-ProType=SA2026.02 | 27.8 | |
| Qwen3-30B-A3B-thinking-SFTContext Budget=128K2026.05 | 27.3 | |
| Gemini-2.5-FlashType=SA2026.02 | 26.6 | |
| Web-30B-E-GRPOContext Budget=-2026.05 | 26.4 | |
| Qwen3-MaxType=SA2026.02 | 23.5 | |
| Qwen3-8B-SFT + SAPOContext Budget=64K2026.05 | 22.2 | |
| Qwen3-8B-SFT + GRPOContext Budget=64K2026.05 | 19.7 | |
| WebSailor-32BContext Budget=32K2026.05 | 18 | |
| Qwen3-8B-SFTContext Budget=64K2026.05 | 17.9 | |
| MiroThinker-v1.0-8BContext Budget=64K2026.05 | 16.3 | |
| Qwen3-8B-SFT + ARPOContext Budget=64K2026.05 | 15.9 | |
| WebSailor-7BContext Budget=32K2026.05 | 9.8 | |
| WebExplorer-8BContext Budget=64K2026.05 | 6.2 | |
| MiroThinker-v1.5-30BContext Budget=128K2026.05 | 5.7 |