langchain-dependencies

Install langchain>=1.0, langchain-core>=1.0, and langsmith>=0.3.0 as the base, then add langgraph or deepagents for orchestration plus your model provider package such as langchain-openai or langchain-anthropic.

Use LangGraph when you need fine-grained control over custom graphs, loops, and branching. Use Deep Agents when you want batteries-included planning, memory, and file context; it installs LangGraph as a transitive dependency.

LangChain 1.0 requires Python 3.10 or higher; Python 3.9 and below are not supported. For TypeScript, Node.js 20 or higher is required.

langchain-community does not follow semantic versioning, so minor releases can contain breaking changes. Pin it to an exact minor series like >=0.4.0,<0.5.0, or switch to a dedicated integration package such as langchain-chroma.

Prefer dedicated integration packages like langchain-tavily, langchain-chroma, or langchain-pinecone when they exist. They are independently versioned, better tested, and more stable than the community package equivalents.

@langchain/core is a peer dependency and is not always hoisted automatically in yarn workspaces or monorepos. Always list it explicitly in your package.json dependencies alongside langchain and @langchain/langgraph.