Home  /  Agents  /  LangGraph
LangChain · Agent frameworks

LangGraph

LangGraph by LangChain: what it does, how it works, pricing and adoption — sourced and dated.

Agent frameworks Open source · MIT ≈42.5K GitHub stars (2026-09)

LangGraph is LangChain's low-level orchestration library for stateful, long-running agents, modelled as directed graphs with built-in checkpointing — a graph can fail, resume exactly where it left off, and replay earlier states. It reached 1.0 GA on October 22, 2025 and is the closest thing the category has to a production default (Klarna, Uber, LinkedIn and GitLab run agents on it). The trade-off is a steeper learning curve than role-based frameworks.

How it works

A framework, not a worker: you define the agent's graph — nodes for model calls and tools, edges for control flow, checkpointers for memory and resumability — and LangGraph runs it with durable execution and human-in-the-loop interrupts.

The facts

MakerLangChain
TypeAgent framework
CategoryAgent frameworks
StatusActive
LicenseMIT
ModelsModel-agnostic; works standalone or with the wider LangChain ecosystem.
GitHub stars42,526 (snapshot 2026-09) · langchain-ai/langgraph
Websitehttps://www.langchain.com/langgraph

Pricing

Framework is free and open source (MIT) — your real cost is model API usage. Commercial layers: LangGraph Platform for deployment and LangSmith observability (free tier, then $39/seat/month).

Pricing model: Free / open source (+ paid platform) · Source: published tiers · checked Oct 14, 2026

Frameworks run on models — compare them in the model database or check the price comparison before you build.

Also in Agent frameworks

Sources

Profile verified Oct 14, 2026. Spotted an error? Every figure above links to its source.