Back to Arena

LangGraph

by LangChain

System Card

OrganizationLangChain
Released2024-01
Architectureagentic-workflow / Graph state + short/long-term memory checkpointers
DetailsLow-level orchestration for stateful graph agents. Short-term working memory is persisted through checkpointers (SQLite, Postgres, Redis), long-term memory via store protocol (usually paired with LangMem).
Parameters—
Domainagent-memoryepisodic-session
Open SourceYes
WebsiteVisit
CodeRepository
checkpointergraph-statelangchaindurable

Capability Profile

Benchmark Scores

6 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
7658p
MemoryBank
74.238p
Cross-Session Memory
1/1
LongMemEval
76.957p
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
74.575p
HotpotQA
71.651p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:LangGraph (langchain-ai/langgraph); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)LangGraph (langchain-ai/langgraph); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)LangGraph (langchain-ai/langgraph); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)LangGraph (langchain-ai/langgraph); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)LangGraph (langchain-ai/langgraph); evaluated on MemoryBank: Enhancing LLMs with Long-Term Memory (Sun Yat-sen University, 2305)LangGraph (langchain-ai/langgraph); evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)