Scientific Foundation Model Collaboration on EywaBench Overall
67.61Utility ScoreEywaMAS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EywaMASSetting=Multi-Agent2026.04 | 67.61 | 72.11 | 11,214 | |
| EywaOrchestraSetting=Dynamic Orchestration2026.04 | 67.46 | 48.16 | 8,335 | |
| EywaAgentSetting=Single-Agent2026.04 | 65.58 | 22.78 | 3,137 | |
| Debate MASSetting=Multi-Agent2026.04 | 64.6 | 78.22 | 13,216 | |
| Refine MASSetting=Multi-Agent2026.04 | 62.94 | 60.59 | 8,673 | |
| MoASetting=Multi-Agent2026.04 | 62.73 | 57.75 | 15,317 | |
| X-MASSetting=Multi-Agent2026.04 | 61.88 | 77.42 | 16,537 | |
| Single-LLM-AgentSetting=Single-Agent2026.04 | 61.54 | 25.22 | 4,469 |