Long-context Question Answering on LOCOMO Multi-hop
56.72F1 ScoreMRAgent
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MRAgentModel (Backbone)=Claude2026.06 | 56.72 | 90.19 | |
| Mem0Model (Backbone)=Claude2026.06 | 48.66 | 75.88 | |
| Mem0Model (Backbone)=Gemini2026.06 | 45.17 | 68.79 | |
| LangMemModel (Backbone)=Claude2026.06 | 44.37 | 70.92 | |
| MRAgentModel (Backbone)=Gemini2026.06 | 43.69 | 75.17 | |
| A-MemModel (Backbone)=Claude2026.06 | 42.45 | 71.67 | |
| MemoryOSModel (Backbone)=Gemini2026.06 | 41.42 | 63.82 | |
| LangMemModel (Backbone)=Gemini2026.06 | 40.67 | 61.34 | |
| RAGModel (Backbone)=Gemini2026.06 | 34.89 | 58.16 | |
| RAGModel (Backbone)=Claude2026.06 | 34.53 | 57.45 | |
| A-MemModel (Backbone)=Gemini2026.06 | 33.81 | 53.54 | |
| MemoryOSModel (Backbone)=Claude2026.06 | 32.94 | 60.99 |