Multi-Agent Reasoning on AIME 24
30CallsCoTAgent
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CoTAgent2026.01 | 30 | 0.05 | 0.03 | 0.38 | |
| MAS-Orchestra2026.01 | 51 | 0.03 | 0.01 | 0.5 | |
| ReflexionAgent2026.01 | 112 | 0.14 | 0.06 | 0.67 | |
| SCAgent2026.01 | 156 | 0.25 | 0.13 | 1.88 | |
| MAS-GPT2026.01 | 228 | 0.74 | 0.37 | 5.47 | |
| DebateAgent2026.01 | 332 | 0.37 | 0.13 | 2.29 | |
| MaAS2026.01 | 501 | 1.28 | 0.62 | 6.96 | |
| AFlow2026.01 | 1,288 | 3.47 | 1.46 | 19.28 | |
| ToolOrchestra2026.01 | 1,319 | 3.77 | 1.76 | 31.35 |