LangGraph
An open source Python framework for AI agents with persistent memory, durable execution, and control over complex workflows.
LangGraph is a free, MIT-licensed Python framework for developers building AI agents that need to keep state and handle complex tasks. It gives teams control over how an agent moves between steps, where people can intervene, and what happens when a long-running task is interrupted.
Its workflows can include a single agent, multiple agents, or agents arranged in a hierarchy. Developers can set checks that let a person inspect, change, or approve an agent's actions. Durable execution preserves progress through failures so an agent can resume its work. Memory keeps conversation context across sessions, while token-by-token streaming lets an application show responses as they arrive.
The framework works with any model provider, so it doesn't require a particular model backend. It's a library for building an agent application, rather than a ready-made chat interface or local LLM runner. A separate JavaScript and TypeScript library, LangGraph.js, serves developers working in those languages. LangSmith is a separate platform for tracing agent decisions, evaluating changes, and deploying agents.
