Skip to main content
Prerequisite: Complete the Quickstart to get an access key. All requests go to https://gateway.ngrok.ai.
LangChain is a framework for building applications with LLMs. The ngrok AI Gateway works with LangChain’s OpenAI-compatible integrations, adding automatic failover and key rotation.

​
Installation

pip install langchain langchain-openai

​
Basic usage

Configure LangChain to use your AI Gateway. Pass your access key (ng-xxxxx-g1-xxxxx) as the API key—the AI Gateway injects upstream provider credentials from your access key configuration. 
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx",  # Your access key
    model="gpt-4o"
)

response = llm.invoke("What is the capital of France?")
print(response.content)

​
Streaming

Stream responses for real-time output: 
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx",  # Your access key
    model="gpt-4o"
)

for chunk in llm.stream("Write a poem about AI"):
    print(chunk.content, end="", flush=True)

​
Using different providers

Route to different providers with model prefixes: 
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx",  # Your access key
    model="anthropic:claude-3-5-sonnet-latest"  # Use Anthropic through gateway
)

response = llm.invoke("Hello!")

​
Chains

Build chains that route through the gateway: 
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate

llm = ChatOpenAI(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx",  # Your access key
    model="gpt-4o"
)

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a helpful assistant that translates {input_language} to {output_language}."),
    ("human", "{input}")
])

chain = prompt | llm

response = chain.invoke({
    "input_language": "English",
    "output_language": "French",
    "input": "Hello, how are you?"
})

print(response.content)

​
Embeddings

Generate embeddings through the gateway: 
from langchain_openai import OpenAIEmbeddings

embeddings = OpenAIEmbeddings(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx",  # Your access key
    model="text-embedding-3-small"
)

vector = embeddings.embed_query("Hello world")

​
RAG applications

Build RAG applications with automatic failover: 
from langchain_openai import ChatOpenAI, OpenAIEmbeddings
from langchain_community.vectorstores import FAISS
from langchain_core.prompts import ChatPromptTemplate

llm = ChatOpenAI(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx",  # Your access key
    model="gpt-4o"
)

embeddings = OpenAIEmbeddings(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx"  # Your access key
)

# Create vector store
texts = ["Document 1 content", "Document 2 content"]
vectorstore = FAISS.from_texts(texts, embeddings)

# Query with RAG
retriever = vectorstore.as_retriever()
docs = retriever.invoke("query")

prompt = ChatPromptTemplate.from_template(
    "Answer based on context: {context}\n\nQuestion: {question}"
)

chain = prompt | llm
response = chain.invoke({"context": docs, "question": "Your question"})

​
Agents

Build agents that use the gateway: 
from langchain_openai import ChatOpenAI
from langchain.agents import create_react_agent, AgentExecutor
from langchain_core.tools import tool

llm = ChatOpenAI(
    base_url="https://gateway.ngrok.ai/v1",
    api_key="ng-xxxxx-g1-xxxxx",  # Your access key
    model="gpt-4o"
)

@tool
def search(query: str) -> str:
    """Search for information."""
    return f"Results for: {query}"

# Create and run agent
# ...

​
Next steps