Agent Performance on VitaBench OTA
9.75Avg@4NoisyAgent
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NoisyAgentbackbone=Qwen3-32B2026.05 | 9.75 | 15 | |
| DAPObackbone=Qwen3-32B2026.05 | 9.25 | 15 | |
| GSPObackbone=Qwen3-32B2026.05 | 9 | 14 | |
| GRPObackbone=Qwen3-32B2026.05 | 8.75 | 14 | |
| Qwen3-32Bbackbone=Qwen3-32B2026.05 | 7 | 12 | |
| NoisyAgentbackbone=Qwen3-8B2026.05 | 5 | 9 | |
| GSPObackbone=Qwen3-8B2026.05 | 4.5 | 8 | |
| GRPObackbone=Qwen3-8B2026.05 | 4.25 | 7 | |
| DAPObackbone=Qwen3-8B2026.05 | 4 | 7 | |
| Qwen3-8Bbackbone=Qwen3-8B2026.05 | 1.75 | 4 |