Interactive Web Shopping on MemoryArena Bundled Web Shopping
34.67Success Rate (SR)MAGE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MAGEBackend Model=Gemma4-31B2026.06 | 34.67 | 75.89 | 1,550 | |
| MAGEBackend Model=Qwen3.6-35B-A3B2026.06 | 30.67 | 72.56 | 2,732 | |
| Long ContextBackend Model=Gemma4-31B2026.06 | 28 | 71.78 | 3,102 | |
| Long ContextBackend Model=Qwen3.6-35B-A3B2026.06 | 22.67 | 69.78 | 4,092 | |
| MemoRAGBackend Model=Qwen3.6-35B-A3B2026.06 | 17.33 | 66.22 | 3,888 | |
| MemoRAGBackend Model=Gemma4-31B2026.06 | 13.42 | 62.08 | 2,545 | |
| MemoryOSBackend Model=Qwen3.6-35B-A3B2026.06 | 12 | 61.44 | 3,835 | |
| MemoryOSBackend Model=Gemma4-31B2026.06 | 12 | 60.56 | 1,854 |