We've rebranded our service for deploying agents at scale as LangGraph Platform (formerly known as LangGraph Cloud). LangGraph Platform provides With LangGraph Platform, we’ve expanded to offer multiple deployment options, which include:

• Self-Hosted Lite: A free (up to 1 million nodes executed), limited version of LangGraph Platform that can be run locally or in a self-hosted manner. Try it out here.

• Cloud SaaS: Fully managed and hosted as part of LangSmith. Allows teams to deploy quickly, with automatic updates and zero maintenance. Currently free while in beta for LangSmith Plus or Enterprise plan users. Get started now.

• Bring Your Own Cloud (BYOC): Run LangGraph Platform in your VPC with our managed service. You can keep data in your environment while we handle provisioning and maintenance. Contact us to learn more.

• Self-Hosted Enterprise: Deploy LangGraph applications entirely on your own infrastructure for maximum control. Contact us to learn more.

Key Features of LangGraph Platform

LangGraph Platform combines essential components for a robust, scalable agent infrastructure:

• LangGraph Server:

  This is our server that provides an opinionated way to deploy and interact with agents (e.g. endpoints for streaming, human-in-the-loop, etc) as well as manage agents (e.g. create assistants, version assistants, etc). This includes:

  • Horizontally scalable infrastructure and task queues to handle high volume or bursts of incoming requests

  • Support for long-running agents that can handle continuous, stateful tasks (unlike most web infrastructure, which is aimed at running short jobs)

  • Ability to persist data within and across conversation threads

  LangGraph Platform’s APIs allow users to create interactive, context-aware agent experiences. Humans can more easily steer their agent and interact with it with the following features:

  • Streaming runs for interactive UX and real-time outputs seen by users

  • Background runs for batch processing to support research-style or time-intensive tasks

  • Interactive state tracking for humans to interact with the persistence layers and update it, rollback, etc.

  • Concurrency control, including ways to deal with multiple incoming user messages before the agent can respond to the first

  • Cron jobs and webhooks to support multi-step workflows

• LangGraph Studio: Our visual studio for debugging and visualizing agent interactions.

• LangGraph CLI & SDK: Supports seamless integration with both Python and JavaScript, allowing developers to easily incorporate LangGraph into their tech stacks.

Read the blog for more on LangGraph Platform: https://blog.langchain.dev/langgraph-platform-announce/

Get started in our docs with the deployment option right for you.