Long-horizon memory-based reasoning on LoCoMo
49.8Multi-hop R-1 ScoreFluxMem
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| FluxMem2026.02 | 49.8 | 19.34 | 90.39 | 56.96 | 23.95 | 92.81 | 37.42 | 9.29 | 89.47 | 63.83 | 38.98 | 92.58 | |
| LangMem2026.02 | 48.07 | 13.8 | 65.16 | 64.12 | 21.55 | 80.87 | 24.55 | 9.38 | 45.19 | 51.3 | 29.15 | 68.49 | |
| MemoryOS2026.02 | 44.27 | 17.19 | 89.91 | 44.87 | 19.84 | 91.17 | 28.2 | 9.68 | 88.67 | 50.77 | 30.29 | 91.07 | |
| O-mem2026.02 | 36.83 | 14.53 | 30.81 | 50.02 | 26.45 | 50.38 | 21.62 | 3.19 | 26.84 | 45.16 | 27.74 | 41.6 | |
| A-Mem2026.02 | 34.54 | 13.62 | 54.95 | 40.86 | 19.01 | 65.67 | 16.85 | 3.66 | 40.06 | 43.06 | 25.66 | 56.13 | |
| Mem02026.02 | 29.95 | 9.43 | 52.2 | 24.31 | 6.94 | 50.27 | 23.53 | 5.56 | 40.82 | 33.48 | 19.25 | 46.42 | |
| HippoRAG 22026.02 | 29.19 | 9.46 | 87.26 | 10.32 | 2.12 | 84.32 | 25.08 | 7.7 | 88.05 | 45.7 | 29.23 | 90.24 | |
| MEMOS2026.02 | 18.33 | 6.5 | 50.75 | 12.4 | 3.25 | 44.05 | 11.04 | 2.45 | 39.91 | 18.1 | 10.01 | 53.3 | |
| MemR32026.02 | 17.34 | 6.2 | — | 24.15 | 8 | — | 10.39 | 2.26 | — | 23.09 | 13.77 | — | |
| ZEP2026.02 | 8 | 2.11 | 32.38 | 6.56 | 1.06 | 19.05 | 6.1 | 0.92 | 34.09 | 6.94 | 2.26 | 26.29 |