OpenAI Integration (Experimental)

This folder contains an experimental integration for OpenAI models in ADK.

Usage in Code

To use the OpenAI integration in your Python code, instantiate OpenAILlm and assign it to your agent's model field:

from google.adk.agents.llm_agent import LlmAgent
from google.adk.labs.openai import OpenAILlm

# Create the OpenAI model instance
openai_model = OpenAILlm(model="gpt-4o")

# Create an agent and assign the model
agent = LlmAgent(
    name="my_openai_agent",
    model=openai_model,
    instruction="You are a helpful assistant.",
)

Requires the openai Python package and OPENAI_API_KEY environment variable.

OpenAI-Compatible Endpoints

To reach a host that speaks the OpenAI API, or to configure anything else the client supports, build an AsyncOpenAI yourself and pass it as client. Each model instance keeps its own client, so one process can talk to several hosts:

from openai import AsyncOpenAI
from google.adk.labs.openai import OpenAILlm

openai_model = OpenAILlm(
    model="my-model",
    client=AsyncOpenAI(base_url="https://my-host.example/v1", api_key="..."),
)

OpenAIResponsesLlm takes the same client field.

Tip: The OpenAI Python client also honors OPENAI_BASE_URL for OpenAI-compatible multi-model gateways — for example DaoXE at https://api.daoxe.com/v1.