Python MIT orchestrationgraph
LangGraph
Build stateful, multi-actor LLM applications as graphs
LangGraph is the LangChain team's framework for building agent workflows as explicit state graphs. Where most frameworks hide the control flow, LangGraph forces you to draw it, which pays off as soon as your agents need branching, retries, or human review steps.
This is the placeholder body for LangGraph. Editorial review pending.
Key features
- Graph-based agent orchestration
- Stateful workflows with persistent memory
- Human-in-the-loop checkpoints
- Streaming and async support
- Studio UI for visualizing graphs
Frequently Asked Questions
What is LangGraph?
LangGraph is an open-source Python and TypeScript framework for building agent workflows as state graphs. Instead of writing chains, you define nodes (steps) and edges (transitions), which makes branching and retries explicit.
Is LangGraph free?
Yes, LangGraph itself is open source and MIT-licensed. The optional LangSmith observability platform from the same team is a separate paid product.
When should I use LangGraph instead of CrewAI?
Use LangGraph when your workflow has serious branching, retries, or human approval steps that you want to model explicitly. Use CrewAI when you want a quick role-based prototype where agents hand work off to each other.