Comparison

Tenure vs. Supermemory

Supermemory is the best-performing competitor in PrecisionMemBench. Memory graph with dynamic dreaming. Mean precision: 0.43. Still returns 2-3 irrelevant beliefs per correct one.

TL;DR

  • Best competitor precision at 0.43 (Tenure: 1.00). The only other system with any active retrieval passes.
  • 17/43 active passes vs. Tenure's 43/43. Still fails on more than half of precision-assertion cases.
  • Retrieval latency: 819.48ms p50 vs. Tenure's 9.77ms (84x slower).
  • Session performance: 2/12 turns passed, drift score 0.17 (better than most, still not isolated).
  • Cloud-first SaaS. Requires account and API key. Data leaves your machine.
Tenure precision
1.00
43/43 active passes
Supermemory precision
0.43
17/43 active passes
Tenure latency
9.77ms
p50 retrieval
Supermemory latency
819.48ms
p50 retrieval

Credit where it's due

Supermemory is architecturally more sophisticated than Mem0, Zep, or Hindsight. Its memory graph and dynamic dreaming mechanism provide better signal than raw cosine similarity. 17 active retrieval passes is meaningful; no other competitor scores above zero.

But 0.43 precision still means that for every correct belief returned, there are 1-2 irrelevant ones alongside it. On 26 of 43 precision-assertion cases, the system fails. The memory graph helps, but it does not resolve the fundamental problem: semantic proximity within a domain-specific corpus cannot be discriminated by any similarity-based mechanism.

0.43 precision means your model still has to sort signal from noise on most queries. Every irrelevant belief in context is tokens you are paying for and latency you are waiting on. Tenure's structural approach (alias-weighted BM25 + hard scope isolation) does not degrade with corpus growth. It gets more precise as the alias set expands.

Latency and deployment

Supermemory's p50 retrieval latency of 819.48ms is 84x slower than Tenure's 9.77ms. Session-turn latency averages 867.83ms. This is the cost of running a memory graph traversal and dynamic dreaming pipeline on every query.

Supermemory is a cloud-first SaaS product. Your data leaves your machine. Tenure runs entirely on localhost. Nothing ever leaves your machine. No account required.

Full comparison

Property Tenure Supermemory
Mean retrieval precision 1.00 0.43
Active retrieval passes 43/43 17/43
Total passes (77 non-session) 77/77 44/77
Session turns passed 12/12 2/12
Retrieval latency (p50) 9.77ms 819.48ms
Session latency (p50) 47.79ms 867.83ms
Drift score 0.00 0.17
Runs locally Always Cloud SaaS
Account required No Yes
Data leaves machine Never Always (cloud)
Scope isolation Hard filter Partial (graph-based)
Per-turn injection audit Yes No
Pricing Free (MIT) $19-$399+/mo
License MIT Proprietary

Full methodology: arXiv:2605.11325. Dataset: HuggingFace. Live leaderboard: HuggingFace Spaces.

2.3x more precise. 84x faster. Free.

No cloud. No account. No monthly bill. Thirty seconds to install. Runs on your machine forever.