- Category
- Developer Tools
- Rank
- No. 680Tools index
- Pricing
- Freemium
- Platform
- cli · web
- Type
- TOOL
- Builder
- @mem0ai
- GitHub
- 64.3k stars
- Latest release
- mem0-strands-v0.1.1
- Date
About
Mem0 is a memory layer that gives AI agents and apps persistent, cross-session context by extracting, compressing, and retrieving user/session/agent-level memories. Its latest algorithm combines semantic, keyword, and entity-linked retrieval with temporal reasoning, reporting large accuracy gains on memory benchmarks like LoCoMo and LongMemEval while cutting token usage.
What it can do
Extract memories from user/agent conversations or sessions
Conversation or session text → Extracted memory entries
Compress extracted memories to reduce token usage
Extracted memory entries → Compressed memory representation
Retrieve relevant memories using combined semantic, keyword, and entity-linked search
Query text → Ranked relevant memories
Apply temporal reasoning during memory retrieval
Query with time context → Time-aware memory results
Provide persistent cross-session context for AI agents/apps
User/session/agent identifiers → Stored contextual memory
Why it made the leaderboard
Gives agents persistent, updatable user context across sessions through a drop-in SDK, self-hosted server, or managed cloud, with entity linking and multi-signal retrieval rather than naive vector recall. Useful when your agent keeps forgetting who the user is and what already happened.
Tags
Tech Stack
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.