Mem0

A hybrid memory layer combining vector, key-value, and graph stores — achieving 26% higher accuracy than full-context approaches with 91% lower latency. Supports hierarchical user/session/agent memory scopes for personal AI.

AI ToolFree Tier: Apache 2.0; 21K GitHub starsCompany: Mem0 AICategory: Personal AI SystemsOpen Source: YesQuick Start: Install with pip install mem0ai → Initialize with your preferred storage backend → Add and retrieve memories with automatic relevance scoring

Mem0

Hybrid memory layer — vector+KV+graph store; 26% higher accuracy on LoCoMo benchmark; 91% lower latency vs full-context

Visit Mem0 →

A hybrid memory layer combining vector, key-value, and graph stores — achieving 26% higher accuracy than full-context approaches with 91% lower latency. Supports hierarchical user/session/agent memory scopes for personal AI.

FREE TIER
Apache 2.0; 21K GitHub stars
COMPANY
Mem0 AI
CATEGORY
Personal AI Systems › Memory & Vector
OPEN SOURCE
Yes
PRIVACY
Self-hosted; hierarchical user/session/agent memory
TAGS
memoryhybridpersistentrag

QUICK START

Install with pip install mem0ai → Initialize with your preferred storage backend → Add and retrieve memories with automatic relevance scoring

BEST FOR

  • ▸Developers building private, high-performance personal AI systems.
  • ▸Engineers needing a hybrid memory layer for complex AI agent recall.
  • ▸Teams prioritizing self-hosted control, accuracy, and low latency.

NOT FOR

  • ✕Non-technical users seeking simple, plug-and-play AI memory solutions.
  • ✕Users preferring managed cloud services for their AI infrastructure.
Finding similar tools…
← Back to all tools