Long-term memory retrieval on LongMemEval
96.15Knowledge UpdateChronos Low
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Chronos LowLLM-as-Judge=GPT-4o2026.05 | 96.15 | 91.73 | 100 | 80 | 94.29 | 90.23 | 92.6 | |
| MemCog (full)Configuration=full, LLM-as-Judge=GPT-4o2026.05 | 91.03 | 92.48 | 98.21 | 96.67 | 100 | 98.5 | 95.8 | |
| MemCog w/o ProactiveConfiguration=without Proactive component, LLM-as-Judge=GPT-4o2026.05 | 89.74 | 92.48 | 98.21 | 96.67 | 100 | 96.24 | 95 | |
| MastraLLM-as-Judge=GPT-4o2026.05 | 84.8 | 85.9 | 79.7 | 82.14 | 73.33 | 98.57 | 85.71 | |
| MemCog w/o Graph OverlayConfiguration=without Graph Overlay component, LLM-as-Judge=GPT-4o2026.05 | 84.42 | 91.73 | 98.21 | 90 | 100 | 95.45 | 93.37 | |
| HindsightLLM-as-Judge=GPT-4o2026.05 | 83.6 | 84.6 | 79.7 | 94.6 | 66.7 | 95.7 | 79.7 | |
| ZEPLLM-as-Judge=GPT-4o2026.05 | 83.3 | 57.9 | 80.4 | 56.7 | 92.9 | 62.4 | 71.2 | |
| SupermemoryLLM-as-Judge=GPT-4o2026.05 | 81.6 | 88.5 | 71.4 | 96.4 | 70 | 97.1 | 76.7 | |
| mem0LLM-as-Judge=GPT-4o2026.05 | 79.5 | 70.7 | 46.4 | 76.7 | 94.3 | 51.1 | 67.8 | |
| Full ContextLLM-as-Judge=GPT-4o2026.05 | 78.2 | 44.3 | 94.6 | 20 | 81.4 | 45.1 | 60.2 |