LangMem (LangChain)
https://langchain-ai.github.io/langmem/
First-class memory module inside LangChain — explicit cognitive-science-style typing of memories.
At a glance
- Type
- Episodic / semantic / procedural
- Tier
- T2
- Section
- Framework-embedded memory
- Created
- 2025-01
- Latest release
- no releases
- License
- MIT
- GitHub
- 1.4k★ +30/mo Python
- Pricing
- OSS library free; LangSmith (observability/platform) Free 5k traces/mo; Plus $39/mo; Enterprise custom. LangGraph P…
- Funding
- $160M total $1.2B val Series B · 2025-10
Taxonomy
- storage
- vector
- retrieval
- similarity
- persistence
- long-term
- update
- extraction
- unit
- episode
- governance
- inspectable
- conflict
- llm-arbitrate
When to use
Optimised for: cognitive-typed memory (episodic/semantic/procedural) inside LangGraph
Anti-fit: not for non-LangChain stacks (Python-first)
Pros & cons
Pros
First-party LangChain integration so memory composes with retrievers, chains, and graphs without glue code.
Cons
LangChain-shaped — maximum value if you've already adopted LangChain; less useful otherwise.
Claims & capabilities
First-class cognitive-typed memory module (episodic/semantic/procedural) inside LangChain. Reports 58.10% accuracy on LOCOMO with p95 search latency of 59.82s — much slower than Mem0 (0.200s, 67.13%). LangChain raised $160M Series B at $1.2B valuation (October 2025). MIT-licensed; ~1.4k★ GitHub
Technical surface
- API surface
- SDK: Python (create_manage_memory_tool, create_search_memory_tool, background memory manager)
- Backend storage
- pluggable
- Deployment
- Both (LangSmith cloud managed + self-hosted Kubernetes; LangGraph Platform cloud or self-hosted)
- Embedding model
- multiple supported
- Multi-tenancy
- not applicable — library; namespacing is hierarchical (tuple-keyed) within whatever store the embedding app provides
- MCP
- via official adapter — LangChain MCP adapters
- A2A
- supported (LangChain A2A integration)
- OpenTelemetry
- first-class — LangSmith + OTel exporter
Similar systems
Other framework-embedded memory in the catalog, ranked by inbound references.
- LangGraph Persistence T2
Distinct from LangMem. Built-in checkpointer saves graph state per super-step (short-term, thread-scoped). Store System adds long-term hierarchical key-value memory across threads with optional vector search + TTL. Postgres / Mongo / Redis stores for production.
- AutoGen Memory T2
ListMemory chronological context + teachable agents that vectorise corrections. Integrates with Mem0/Zep rather than building deep memory natively.
- CrewAI Memory T2
Memory subsystem inside the CrewAI orchestration framework; integrates with Mem0 for the long-term tier.
- AGiXT Adaptive Memory T2
Open-source AI automation platform. Routes between short-term and long-term memory adaptively across any LLM provider; plugin system for storage backends. Memory managed at the instruction-management layer — task context, instruction state, conversation history as unified agent state.
- Agno (Phidata) Memory T2
Agno (formerly Phidata). AgentStorage persists sessions to a DB; AgentMemory auto-classifies/store user preferences and conversation summaries. Single-line integrations with LanceDB, Pinecone, Weaviate, Qdrant.
- Botpress LLMz T1
Per-plan vector-DB storage quota + LLMz autonomous engine (in-session working memory) + Knowledge Base (semantic search over uploaded docs). Long-term user memory persists across sessions.
Related systems
References (3)
- LangChain (framework) depends on at runtime — adjacent-infrastructure cell: requires LangChain / LangGraph; BYO LLM + store
- LangGraph (orchestration) depends on at runtime — adjacent-infrastructure cell: requires LangChain / LangGraph; BYO LLM + store
- LangGraph Persistence integrates with — First-party LangChain integration so memory composes with retrievers, chains, and graphs without glue code.