LangGraph

LangGraph is an AI agent framework for developers building applications on their own machines, including workflows that use local models through LangChain's ChatOllama integration. Ollama can run models such as gpt-oss locally. LangGraph also works with remote model providers, so where model requests go depends on the connection your application uses.

Its main distinction is control over the workflow. Developers can combine predictable, hand-coded steps with model-driven decisions in the same graph, rather than adopt a fixed agent architecture. The framework supports single agents, collaborating teams and hierarchical workflows.

Tasks can resume after interruptions or failures, preserving their progress instead of restarting the entire run. Short-term memory holds working context, while persistent memory carries conversation history across sessions. Human oversight lets people inspect or edit agent state and approve actions at chosen points.

Streaming exposes response tokens and agent actions as they arrive. Through ChatOllama, local workflows can also use tool calling, structured output and image input when the selected model supports those capabilities.

LangGraph is free and open source under the MIT license. It's a library for building an application, with an equivalent JavaScript and TypeScript library called LangGraph.js. It can work without LangChain, although ChatOllama is a LangChain integration. LangSmith is a separate platform for tracing, evaluation and deployment; enabling its tracing sends execution traces to that service.