Back to Arena

MemPalace

by Ben Sigman / Milla Jovovich (independent open-source)

System Card

OrganizationBen Sigman / Milla Jovovich (independent open-source)
Released2026-04
Architecturehierarchical-summary / Memory Palace Hierarchical Spatial Index
DetailsMemPalace organizes all conversation data into a four-level hierarchy — Wings (projects/people), Halls (memory types), Rooms (specific ideas), and Drawers (verbatim files) — with 30x-compressed AAAK-format Closet summaries alongside verbatim originals in ChromaDB + SQLite. No summarization or extraction is performed on stored content; only the retrieval layer (170-token structural context injection) uses the hierarchy to boost accuracy 34% over flat search. Entirely offline, zero API calls. Benchmarked at 96.6% accuracy; benchmark methodology disputed by community.
Parameters—
Domainepisodic-sessionlifelong-learningrag-retrieval
Open SourceYes
WebsiteVisit
open-sourceofflinememory-palacelocal-llmhierarchicalmit-license

Capability Profile

Benchmark Scores

3 of 14 benchmarks
Long-Context Retrieval
0/5
RULER
no data
NIAH
no data
LooGLE
no data
LongBench
no data
∞Bench
no data
Multi-Turn Recall
2/2
LoCoMo
88.998p
MemoryBank
80.378p
Cross-Session Memory
1/1
LongMemEval
96.699p
Multi-Hop QA
0/3
BABILong
no data
MultiHop-RAG
no data
HotpotQA
no data
Agent Task Memory
0/1
AgentBench-Mem
no data
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:github.com/milla-jovovich/mempalace/blob/main/benchmarks/BENCHMARKS.md — R@5 retrieval recall (NOT answer accuracy); 483/500 verbatim ChromaDB raw mode. Third-party reviews note methodology caveats.github.com/milla-jovovich/mempalace/blob/main/benchmarks/BENCHMARKS.md — R@10 retrieval recall (Hybrid v5, no rerank, top-10)mempalace.tech/benchmarks — MemPalace label is 'MemBench' — may differ from canonical MemoryBank paper eval; treat with caution