Multi-task agent evaluation on Skill-MAS Evaluation Suite Aggregate
41.05Average Performance (%)Skill-MAS-optimized
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Skill-MAS-optimizedMeta-agent=DeepSeek-V4-Flash2026.06 | 41.05 | 2.12 | |
| Skill-MAS-optimizedMeta-agent=Qwen3.5-Plus2026.06 | 38.41 | 2.43 | |
| AFlowMeta-agent=DeepSeek-V4-Flash2026.06 | 35.7 | 1.45 | |
| AOrchestraMeta-agent=DeepSeek-V4-Flash2026.06 | 34.36 | 2.69 | |
| EvoAgentMeta-agent=DeepSeek-V4-Flash2026.06 | 34.3 | 3.33 | |
| Skill-MAS-initMeta-agent=DeepSeek-V4-Flash2026.06 | 33.72 | 1.14 | |
| Skill-MAS-initMeta-agent=Qwen3.5-Plus2026.06 | 32.61 | 1.83 | |
| AFlowMeta-agent=Qwen3.5-Plus2026.06 | 32.23 | 1.56 | |
| EvoAgentMeta-agent=Qwen3.5-Plus2026.06 | 31.1 | 7.91 | |
| MAS-OrchestraMeta-agent=DeepSeek-V4-Flash2026.06 | 31.05 | 1.98 | |
| Skill-MAS-optimizedMeta-agent=Gemini-3.1-Flash2026.06 | 29.49 | 2.82 | |
| AOrchestraMeta-agent=Qwen3.5-Plus2026.06 | 29.02 | 1.91 | |
| Skill-MAS-optimizedMeta-agent=GPT-5.4-Nano2026.06 | 27.55 | 5.22 | |
| MAS-OrchestraMeta-agent=Qwen3.5-Plus2026.06 | 26.92 | 1.35 | |
| EvoAgentMeta-agent=GPT-5.4-Nano2026.06 | 24.83 | 6.26 | |
| MAS2Meta-agent=Qwen3.5-Plus2026.06 | 22.89 | 4.04 | |
| Skill-MAS-initMeta-agent=Gemini-3.1-Flash2026.06 | 21.68 | 2.63 | |
| AFlowMeta-agent=Gemini-3.1-Flash2026.06 | 21.29 | 0.44 | |
| AOrchestraMeta-agent=GPT-5.4-Nano2026.06 | 20.25 | 2.45 | |
| MAS2Meta-agent=DeepSeek-V4-Flash2026.06 | 20.06 | 0.63 | |
| Skill-MAS-initMeta-agent=GPT-5.4-Nano2026.06 | 19.64 | 4.24 | |
| EvoAgentMeta-agent=Gemini-3.1-Flash2026.06 | 19.12 | 8.2 | |
| AFlowMeta-agent=GPT-5.4-Nano2026.06 | 19.11 | 0.69 | |
| MAS-OrchestraMeta-agent=Gemini-3.1-Flash2026.06 | 19.02 | 2.48 | |
| MAS-OrchestraMeta-agent=GPT-5.4-Nano2026.06 | 18.99 | 1.84 | |
| AOrchestraMeta-agent=Gemini-3.1-Flash2026.06 | 18.95 | 2.31 | |
| MAS2Meta-agent=GPT-5.4-Nano2026.06 | 16.68 | 1.27 | |
| MAS2Meta-agent=Gemini-3.1-Flash2026.06 | 15.02 | 0.91 |