Back to Arena

Epsilla

by Epsilla Inc. (YC S23)

System Card

OrganizationEpsilla Inc. (YC S23)
Released2023-07
Architecturevector-rag / parallel graph traversal vector DB
DetailsEpsilla leverages parallel graph traversal algorithms pioneered by its founding team (ex-TigerGraph engineers) to achieve claimed 10x faster vector search over comparable HNSW implementations. The open-source core provides low-latency ANN search across billions of vectors; the cloud service adds an AI agent builder with knowledge-base connectors. Supports hybrid search with metadata filtering.
Parameters—
Domainrag-retrievalagent-memory
Open SourceYes
WebsiteVisit
parallel-graphfast-annagent-builderknowledge-basemulti-language

Capability Profile

Benchmark Scores

6 of 14 benchmarks
Long-Context Retrieval
1/5
RULER
no data
NIAH
no data
LooGLE
no data
LongBench
603p
∞Bench
no data
Multi-Turn Recall
1/2
LoCoMo
53.30p
MemoryBank
no data
Cross-Session Memory
1/1
LongMemEval
58.37p
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
61.69p
HotpotQA
65.431p
Agent Task Memory
1/1
AgentBench-Mem
6413p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:Epsilla vendor documentation; evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Epsilla vendor documentation; evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)Epsilla vendor documentation; evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)Epsilla vendor documentation; evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)Epsilla vendor documentation; evaluated on LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (Tsinghua KEG, 2308)Epsilla vendor documentation; evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)