Getting Started with LangGraph OpenAI Serve¶

This tutorial will guide you through the process of setting up and running your first LangGraph OpenAI compatible server.

Prerequisites¶

Before you begin, make sure you have:

Python 3.11 or higher installed
Basic familiarity with LangGraph
Basic understanding of FastAPI (optional)

Installation¶

First, install the langgraph-openai-serve package:

# Using uv (recommended)
uv add langgraph-openai-serve

# Using pip
pip install langgraph-openai-serve

Basic Usage¶

Here's a simple example to help you get started. In this example, we'll create a basic server that exposes a simple LangGraph workflow through an OpenAI-compatible API.

1. Create a Python file for your server¶

Create a new file called server.py with the following content:

from langgraph_openai_serve import LangchainOpenaiApiServe, GraphRegistry, GraphConfig

# Import your LangGraph instances or use the default simple graph
# that comes with the package
from langgraph_openai_serve.graph.simple_graph import app as simple_graph

# Create a GraphRegistry
graph_registry = GraphRegistry(
    registry={
        "simple_graph": GraphConfig(graph=simple_graph, streamable_node_names=["generate"]),
    }
)

# Create a server instance with your graph(s)
graph_serve = LangchainOpenaiApiServe(
    graphs=graph_registry,
    configure_cors=True,  # Enable CORS for browser clients
)

# Bind the OpenAI-compatible endpoints
graph_serve.bind_openai_chat_completion(prefix="/v1")

# Run the app with uvicorn
if __name__ == "__main__":
    import uvicorn
    uvicorn.run(graph_serve.app, host="0.0.0.0", port=8000)

2. Run the server¶

Start the server by running:

python server.py

Your server should now be running at http://localhost:8000

3. Test the API¶

You can test the API using curl:

curl -X POST http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "simple_graph",
    "messages": [
      {"role": "user", "content": "Hello, how can you help me today?"}
    ]
  }'

Alternatively, you can use the OpenAI Python client:

from openai import OpenAI

# Create a client pointing to your API
client = OpenAI(
    base_url="http://localhost:8000/v1",
    api_key="any-value"  # API key is not verified
)

# Use the graph by specifying its name as the model
response = client.chat.completions.create(
    model="simple_graph",  # This maps to the graph name in your registry
    messages=[
        {"role": "user", "content": "Hello, how can you help me today?"}
    ]
)

print(response.choices[0].message.content)

Using with your own FastAPI application¶

If you already have a FastAPI application, you can integrate LangGraph OpenAI Serve with it:

from fastapi import FastAPI
from langgraph_openai_serve import LangchainOpenaiApiServe, GraphRegistry, GraphConfig
from langgraph_openai_serve.graph.simple_graph import app as simple_graph

# Create a FastAPI app
app = FastAPI(
    title="My API with LangGraph",
    version="1.0",
    description="API that includes LangGraph capabilities",
)

# Create a GraphRegistry
graph_registry = GraphRegistry(
    registry={
        "simple_graph": GraphConfig(graph=simple_graph, streamable_node_names=["generate"]),
    }
)

# Create the LangchainOpenaiApiServe instance
graph_serve = LangchainOpenaiApiServe(
    app=app,  # Pass in your existing FastAPI app
    graphs=graph_registry,
)

# Bind the OpenAI-compatible endpoints
graph_serve.bind_openai_chat_completion(prefix="/v1")

# Add your other routes as needed
@app.get("/")
async def root():
    return {"message": "Welcome to my API with LangGraph integration!"}

# Run the app with uvicorn
if __name__ == "__main__":
    import uvicorn
    uvicorn.run(app, host="0.0.0.0", port=8000)

Next Steps¶

Now that you have a basic understanding of how to use LangGraph OpenAI Serve, you might want to:

Create custom graphs to expose through your API
Learn more about connecting with OpenAI clients
Explore Docker deployment for production use