Benchmark Proof
MemQ benchmark, exactly as generated.
This page is sourced from the canonical benchmark snapshot published from the standalone MemQ performance repo and mirrored into this site during deploy. The current public proof is a retrieval benchmark comparing hosted MemQ against Mem0 OSS and a keyword baseline on the same published case set.
MemQ primary@1
100%
Current top-rank retrieval rate across the published case set.
Mem0 OSS primary@1
58%
Same benchmark harness, same cases, external comparator.
Delta vs Mem0
+42 pts
Published primary@1 uplift in the mirrored retrieval snapshot.
Reproducibility envelope
- Generated
- 2026-04-09T04:03:56.996Z
- Providers
- keyword_baseline, mem0_oss, memq_mcp
- Repetitions
- 1, 2, 3
- Benchmark type
- Retrieval benchmark
- Artifact count
- 108
- Case corpus
- 12 cases across 3 namespaces
- Latency
- MemQ avg 21 ms · p95 31 ms · Mem0 avg 3381 ms
What this proves today
- MemQ achieved 100% primary@1, 100% recall@k, and 100% leakage-free retrieval in the published mirrored snapshot.
- Mem0 OSS reached 58% primary@1 on the same case set, producing a published +42 point primary@1 delta for MemQ.
- MemQ averaged 21 ms versus 3381 ms for Mem0 OSS in the current published artifact set.
- The benchmark harness is real, versioned, repeated, and artifact-backed through the public performance repo.
- These numbers validate the hosted memory retrieval stack. They are not a blanket claim about every downstream agent task, model, or prompt format.
Published case families
The current corpus spans real retrieval families rather than a single toy fixture: freshness, runbooks, incidents, project memory, procedural recall, preference memory, and disambiguation across multiple namespaces.
Published benchmark assets
The public proof surface is made from the exact published snapshot and summary that ship with the MemQ performance repo. Download the machine-readable snapshot or the summary markdown directly from this site.
curl -L https://multinex.ai/downloads/memq/memq-benchmark-snapshot.json curl -L https://multinex.ai/downloads/memq/benchmark-summary.md
Navigate the proof system