This repository contains packages with NVIDIA integrations with LangChain:
- langchain-nvidia-ai-endpoints integrates NVIDIA AI Foundation Models and Endpoints, including Nemotron, NVIDIA's open model family built for agentic AI.
- langchain-nvidia-trt implements integrations of NVIDIA TensorRT models.
- langchain-nvidia-langgraph provides NVIDIA-optimized LangGraph execution with parallel and speculative execution strategies.
Tip
For developing, debugging, and deploying AI agents and LLM applications, see LangSmith.
See the notebooks in the cookbook directory for examples of using ChatNVIDIA and NVIDIAEmbeddings with LangGraph for agentic RAG and tool-calling agents.
See the studio directory to test the agentic RAG workflow in LangGraph Studio.
Simply load the studio directory in LangGraph Studio and click the "Run" button with an input question.
This will run agentic RAG where it first reflects on the question to decide whether to use web search or vectorstore retrieval. It also grades retrieved documents as well as generated answers.
