Edit

How to switch between OpenAI and Azure OpenAI endpoints with Python

  • How-to

While OpenAI and Azure OpenAI Service rely on a common Python client library, there are small changes you need to make to your code in order to swap back and forth between endpoints. This article walks you through the common changes and differences you'll experience when working across OpenAI and Azure OpenAI.

This article only shows examples with the new OpenAI Python 1.x API library. For information on migrating from 0.28.1 to 1.x refer to our migration guide.

Prerequisites

None

Authentication

We recommend using environment variables. If you haven't done this before, our Python quickstarts walks you through this configuration.

API key

OpenAI Azure OpenAI 
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.getenv("OPENAI_API_KEY")
)




import os
from openai import AzureOpenAI
    
client = AzureOpenAI(
    api_key=os.getenv("AZURE_OPENAI_API_KEY"),  
    api_version="2023-12-01-preview",
    azure_endpoint=os.getenv("AZURE_OPENAI_ENDPOINT")
)

Microsoft Entra ID authentication

OpenAI Azure OpenAI 
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.getenv("OPENAI_API_KEY")
)









from azure.identity import DefaultAzureCredential, get_bearer_token_provider
from openai import AzureOpenAI

token_provider = get_bearer_token_provider(
    DefaultAzureCredential(), "https://cognitiveservices.azure.com/.default"
)

api_version = "2023-12-01-preview"
endpoint = "https://my-resource.openai.azure.com"

client = AzureOpenAI(
    api_version=api_version,
    azure_endpoint=endpoint,
    azure_ad_token_provider=token_provider,
)

Keyword argument for model

OpenAI uses the model keyword argument to specify what model to use. Azure OpenAI has the concept of unique model deployments. When you use Azure OpenAI, model should refer to the underlying deployment name you chose when you deployed the model.

Important

When you access the model via the API in Azure OpenAI, you need to refer to the deployment name rather than the underlying model name in API calls, which is one of the key differences between OpenAI and Azure OpenAI. OpenAI only requires the model name. Azure OpenAI always requires deployment name, even when using the model parameter. In our docs, we often have examples where deployment names are represented as identical to model names to help indicate which model works with a particular API endpoint. Ultimately your deployment names can follow whatever naming convention is best for your use case.

OpenAI Azure OpenAI 
completion = client.completions.create(
    model="gpt-3.5-turbo-instruct",
    prompt="<prompt>"
)

chat_completion = client.chat.completions.create(
    model="gpt-4",
    messages="<messages>"
)

embedding = client.embeddings.create(
    model="text-embedding-ada-002",
    input="<input>"
)

completion = client.completions.create(
    model="gpt-35-turbo-instruct", # This must match the custom deployment name you chose for your model.
    prompt="<prompt>"
)

chat_completion = client.chat.completions.create(
    model="gpt-35-turbo", # model = "deployment_name".
    messages="<messages>"
)

embedding = client.embeddings.create(
    model="text-embedding-ada-002", # model = "deployment_name".
    input="<input>"
)

Azure OpenAI embeddings multiple input support

OpenAI and Azure OpenAI currently support input arrays up to 2,048 input items for text-embedding-ada-002. Both require the max input token limit per API request to remain under 8,191 for this model.

OpenAI Azure OpenAI 
inputs = ["A", "B", "C"] 

embedding = client.embeddings.create(
    input=inputs,
    model="text-embedding-ada-002"
)



inputs = ["A", "B", "C"] #max array size=2048

embedding = client.embeddings.create(
    input=inputs,
    model="text-embedding-ada-002" # This must match the custom deployment name you chose for your model.
    # engine="text-embedding-ada-002"
)

Related content