Software Engineering on SWE-Bench Verified (Pass %, Out tok, Energy)
72Pass RateHyperAgent + Librarian
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HyperAgent + LibrarianBackbone Model=Qwen3.6-27B, Agent Architecture=HyperAgent, Agent Configuration=+ Librarian2026.05 | 72 | 26.3 | 91.45 | |
| BOADBackbone Model=Qwen3.6-27B, Agent Architecture=BOAD, Agent Configuration=Base2026.05 | 71.8 | 24.3 | 88.96 | |
| BOAD + LibrarianBackbone Model=Qwen3.6-27B, Agent Architecture=BOAD, Agent Configuration=+ Librarian2026.05 | 71.8 | 19.7 | 71.76 | |
| BOAD + LibrarianBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=BOAD, Agent Configuration=+ Librarian2026.05 | 71.6 | 24.6 | 14.66 | |
| BOAD + Caveman + LibrarianBackbone Model=Qwen3.6-27B, Agent Architecture=BOAD, Agent Configuration=+ Caveman + Librarian2026.05 | 71.4 | 13.5 | 51.21 | |
| BOAD + Caveman + LibrarianBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=BOAD, Agent Configuration=+ Caveman + Librarian2026.05 | 70.4 | 20.6 | 12.37 | |
| BOAD + LastNObservationBackbone Model=Qwen3.6-27B, Agent Architecture=BOAD, Agent Configuration=+ LastNObservation2026.05 | 70.2 | 26.2 | 122.91 | |
| BOAD + CavemanBackbone Model=Qwen3.6-27B, Agent Architecture=BOAD, Agent Configuration=+ Caveman2026.05 | 70.2 | 17.5 | 66.2 | |
| BOADBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=BOAD, Agent Configuration=Base2026.05 | 69.8 | 28.5 | 17.45 | |
| HyperAgentBackbone Model=Qwen3.6-27B, Agent Architecture=HyperAgent, Agent Configuration=Base2026.05 | 69.8 | 35.4 | 121.84 | |
| BOAD + CavemanBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=BOAD, Agent Configuration=+ Caveman2026.05 | 68.4 | 24.7 | 14.79 | |
| HyperAgent + CavemanBackbone Model=Qwen3.6-27B, Agent Architecture=HyperAgent, Agent Configuration=+ Caveman2026.05 | 68 | 21.9 | 79.4 | |
| HyperAgent + Caveman + LibrarianBackbone Model=Qwen3.6-27B, Agent Architecture=HyperAgent, Agent Configuration=+ Caveman + Librarian2026.05 | 67.6 | 18 | 66.05 | |
| HyperAgent + LibrarianBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=HyperAgent, Agent Configuration=+ Librarian2026.05 | 67.2 | 42.2 | 22.73 | |
| HyperAgent + CavemanBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=HyperAgent, Agent Configuration=+ Caveman2026.05 | 66.2 | 37.7 | 19.99 | |
| HyperAgent + Caveman + LibrarianBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=HyperAgent, Agent Configuration=+ Caveman + Librarian2026.05 | 66.2 | 34.7 | 18.63 | |
| HyperAgentBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=HyperAgent, Agent Configuration=Base2026.05 | 66 | 48.5 | 25.64 | |
| HyperAgent + LastNObservationBackbone Model=Qwen3.6-27B, Agent Architecture=HyperAgent, Agent Configuration=+ LastNObservation2026.05 | 64.8 | 34 | 135.84 | |
| BOAD + LastNObservationBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=BOAD, Agent Configuration=+ LastNObservation2026.05 | 63.2 | 33.5 | 26.87 | |
| HyperAgent + LastNObservationBackbone Model=Qwen3.6-35B-A3B, Agent Architecture=HyperAgent, Agent Configuration=+ LastNObservation2026.05 | 56.8 | 53.3 | 34.79 |