LangChain

LangChain is an open-source framework for developers building AI agents and LLM applications, including apps that run models locally through Ollama. Its Python ChatOllama integration connects to open-weight models such as gpt-oss. You write the application in code, while Ollama handles local inference.

The framework provides shared interfaces for models, embeddings, vector stores and document retrieval. That lets developers compare model providers or change parts of an application without rewriting every connection. Tool and data integrations let agents consult internal information or external systems as part of their work.

Through ChatOllama, applications can use tool calling, structured output and streamed responses. Image input is available with a compatible model such as Gemma 3. These capabilities depend on the selected model, so choosing Ollama as the backend doesn't give every model the same features.

LangChain offers agent templates for a starting point and lower-level components for custom behavior. LangGraph is a separate framework for finer control over agent workflows; Deep Agents builds on LangChain with planning, subagents and filesystem access.

The Python framework uses the MIT license. LangChain.js provides an equivalent JavaScript and TypeScript library. Local inference through Ollama is separate from remote model-provider integrations. Optional LangSmith tracing connects model calls to a companion platform for monitoring, evaluation and debugging.