High-Performance Agent Memory Engine
Queries across L1 Atomic Facts, L2 Causal Graph, and L3 Dynamic Persona with cryptographic proof verification.
| Model / Engine | PersonaMem | LongMemEval | LoCoMo | Recall p95 | Token Saving | Memory Cost |
|---|---|---|---|---|---|---|
| ● InfinityDB (Ours) | 89.7% | 88.4% | 91.2% | < 5.0ms | 81.4% | $0 (Zero LLM) |
| Zep (Graphiti) | 84.2% | 82.1% | 83.5% | 142.0ms | 45.0% | $149/mo |
| TencentDB Agent Memory | 76.0% | 74.3% | 75.1% | 38.2ms | 61.3% | $30/mo |
| Mem0 | 68.5% | 66.9% | 69.2% | 35.4ms | 42.1% | $99/mo |
| Letta / MemGPT | 71.2% | 70.5% | 72.0% | 68.5ms | 38.5% | $45/mo |