Everything you need to build

Install LangChain and wire up your first model provider.

Build a prompt, model and parser pipeline in under ten lines.

Runnables, messages, prompts and the composition model.

Ground model answers in your own documents with a vector store.

Let the model choose which function to call, and when.

Stream tokens, intermediate steps and tool events to the client.

Measure quality with datasets and graded runs before you ship.

Run LangChain applications in production with observability.