Information Retrieval and Deep Research on xbench DeepSearch 2505
83.7ScoreStep-3.5-Flash
Evaluation Results
| Method | Links | |
|---|---|---|
| Step-3.5-FlashSize=196B-A11B, Context management=false2026.06 | 83.7 | |
| SearchSwarmSize=30B-A3B, Context management=true2026.06 | 80.8 | |
| DeepSeek V3.2Size=671B-A37B, Context management=false2026.06 | 78 | |
| LongSeekerSize=30B-A3B, Context management=true2026.06 | 78 | |
| GPT-5Size=-, Context management=false2026.06 | 77.8 | |
| Tongyi DeepResearchSize=30B-A3B, Context management=false2026.06 | 75 | |
| Tongyi DR SwarmSize=30B-A3B, Context management=false2026.06 | 75 | |
| MiroThinker-1.5-miniSize=30B-A3B, Context management=true2026.06 | 73.1 | |
| GLM-4.7Size=355B-A32B, Context management=false2026.06 | 72 | |
| MiniMax-M2Size=230B-A10B, Context management=false2026.06 | 72 | |
| Claude-4.5-SonnetSize=-, Context management=false2026.06 | 66.5 |