Implementation:Openai Openai python Azure Client

Knowledge Sources	Openai_Openai_python
Domains	SDK_Infrastructure, Python
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete tool for Azure OpenAI API client integration provided by the openai-python SDK.

Description

The Azure Client module provides AzureOpenAI (synchronous) and AsyncAzureOpenAI (asynchronous) client classes for interacting with Azure-hosted OpenAI services. Both classes extend the standard OpenAI / AsyncOpenAI clients and the BaseAzureClient mixin, which handles Azure-specific URL construction and deployment-based routing. The clients automatically inject api-version as a default query parameter and support three authentication methods: API key (via api_key or AZURE_OPENAI_API_KEY), Azure AD token (via azure_ad_token or AZURE_OPENAI_AD_TOKEN), and Azure AD token provider (a callable invoked per request). The api_key, azure_ad_token, and azure_ad_token_provider arguments are mutually exclusive. The BaseAzureClient overrides _build_request() to automatically route deployment-aware endpoints (completions, chat, embeddings, audio, images) via /deployments/{model} URL paths, and overrides _prepare_url() to handle non-deployment endpoints correctly when both azure_endpoint and azure_deployment are set.

Usage

Use the Azure Client when you need to call OpenAI models hosted on Microsoft Azure rather than the OpenAI platform directly. This is required for enterprise deployments that use Azure-specific endpoints, API versioning, and Azure Active Directory authentication. The clients provide the same resource interface (chat, completions, embeddings, etc.) as the standard OpenAI clients.

Code Reference

Source Location

Repository: openai-python
File: src/openai/lib/azure.py
Lines: 1-647

Signature

AzureOpenAI

class AzureOpenAI(BaseAzureClient[httpx.Client, Stream[Any]], OpenAI):
    def __init__(
        self,
        *,
        api_version: str | None = None,
        azure_endpoint: str | None = None,
        azure_deployment: str | None = None,
        api_key: str | Callable[[], str] | None = None,
        azure_ad_token: str | None = None,
        azure_ad_token_provider: AzureADTokenProvider | None = None,
        organization: str | None = None,
        project: str | None = None,
        webhook_secret: str | None = None,
        websocket_base_url: str | httpx.URL | None = None,
        base_url: str | None = None,
        timeout: float | Timeout | None | NotGiven = NOT_GIVEN,
        max_retries: int = DEFAULT_MAX_RETRIES,
        default_headers: Mapping[str, str] | None = None,
        default_query: Mapping[str, object] | None = None,
        http_client: httpx.Client | None = None,
        _strict_response_validation: bool = False,
    ) -> None: ...

AsyncAzureOpenAI

class AsyncAzureOpenAI(BaseAzureClient[httpx.AsyncClient, AsyncStream[Any]], AsyncOpenAI):
    def __init__(
        self,
        *,
        azure_endpoint: str | None = None,
        azure_deployment: str | None = None,
        api_version: str | None = None,
        api_key: str | Callable[[], Awaitable[str]] | None = None,
        azure_ad_token: str | None = None,
        azure_ad_token_provider: AsyncAzureADTokenProvider | None = None,
        organization: str | None = None,
        project: str | None = None,
        webhook_secret: str | None = None,
        base_url: str | None = None,
        websocket_base_url: str | httpx.URL | None = None,
        timeout: float | Timeout | None | NotGiven = NOT_GIVEN,
        max_retries: int = DEFAULT_MAX_RETRIES,
        default_headers: Mapping[str, str] | None = None,
        default_query: Mapping[str, object] | None = None,
        http_client: httpx.AsyncClient | None = None,
        _strict_response_validation: bool = False,
    ) -> None: ...

BaseAzureClient

class BaseAzureClient(BaseClient[_HttpxClientT, _DefaultStreamT]):
    _azure_endpoint: httpx.URL | None
    _azure_deployment: str | None

    def _build_request(
        self,
        options: FinalRequestOptions,
        *,
        retries_taken: int = 0,
    ) -> httpx.Request: ...

    def _prepare_url(self, url: str) -> httpx.URL: ...

Import

from openai.lib.azure import AzureOpenAI, AsyncAzureOpenAI

I/O Contract

Inputs (AzureOpenAI.init)

Name	Type	Required	Description
azure_endpoint	str or None	Conditional	The Azure resource endpoint URL (e.g., `https://example-resource.azure.openai.com/`). Required if `base_url` is not provided. Falls back to `AZURE_OPENAI_ENDPOINT` env var.
azure_deployment	str or None	No	The Azure model deployment name. When set with `azure_endpoint`, the base URL includes `/deployments/{azure_deployment}`.
api_version	str or None	Yes	The Azure API version string (e.g., "2024-02-01"). Falls back to `OPENAI_API_VERSION` env var.
api_key	str or Callable or None	Conditional	Azure OpenAI API key. Falls back to `AZURE_OPENAI_API_KEY`. Mutually exclusive with `azure_ad_token` and `azure_ad_token_provider`.
azure_ad_token	str or None	Conditional	Azure Active Directory token. Falls back to `AZURE_OPENAI_AD_TOKEN`. Mutually exclusive with `api_key` and `azure_ad_token_provider`.
azure_ad_token_provider	Callable or None	Conditional	A callable that returns an Azure AD token string, invoked on every request. Mutually exclusive with `api_key` and `azure_ad_token`.
base_url	str or None	Conditional	Explicit base URL. Mutually exclusive with `azure_endpoint`.
timeout	float or Timeout or None	No	Client-level request timeout.
max_retries	int	No	Maximum number of retries for failed requests. Defaults to 2.
organization	str or None	No	OpenAI organization ID. Falls back to `OPENAI_ORG_ID`.

Outputs

Name	Type	Description
instance	AzureOpenAI or AsyncAzureOpenAI	A fully configured client providing the same resource interface as the standard OpenAI client (chat, completions, embeddings, etc.).

Usage Examples

Basic Usage with API Key

from openai import AzureOpenAI

client = AzureOpenAI(
    azure_endpoint="https://my-resource.openai.azure.com/",
    api_key="my-azure-api-key",
    api_version="2024-02-01",
)

response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}],
)

Async Usage with Azure AD Token Provider

from openai import AsyncAzureOpenAI
from azure.identity.aio import DefaultAzureCredential

credential = DefaultAzureCredential()

async def get_token() -> str:
    token = await credential.get_token("https://cognitiveservices.azure.com/.default")
    return token.token

client = AsyncAzureOpenAI(
    azure_endpoint="https://my-resource.openai.azure.com/",
    azure_ad_token_provider=get_token,
    api_version="2024-02-01",
)

response = await client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}],
)

Using Environment Variables

import os
os.environ["AZURE_OPENAI_API_KEY"] = "my-key"
os.environ["AZURE_OPENAI_ENDPOINT"] = "https://my-resource.openai.azure.com/"
os.environ["OPENAI_API_VERSION"] = "2024-02-01"

from openai import AzureOpenAI
client = AzureOpenAI()  # all config inferred from environment

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment