Implementation:Openai Openai node AzureOpenAI

Knowledge Sources	Openai_Openai_node
Domains	SDK, Azure
Last Updated	2026-02-15 12:00 GMT

Overview

The AzureOpenAI class extends the base OpenAI client to provide seamless integration with Azure OpenAI Service endpoints.

Description

AzureOpenAI is a specialized client class that adapts the standard OpenAI SDK for use with Microsoft Azure's hosted OpenAI services. It extends the base OpenAI class and overrides authentication and request building logic to conform to Azure's API conventions, including support for Azure-specific endpoint formatting, API versioning via the api-version query parameter, and deployment-based URL routing.

The constructor accepts Azure-specific configuration through the AzureClientOptions interface, which adds apiVersion, endpoint, deployment, and azureADTokenProvider on top of the standard ClientOptions. It enforces mutual exclusivity between apiKey and azureADTokenProvider, and between baseURL and endpoint. Environment variables (AZURE_OPENAI_API_KEY, AZURE_OPENAI_ENDPOINT, OPENAI_API_VERSION) serve as fallback defaults.

The class overrides buildRequest to automatically inject /deployments/{model} into API paths for supported endpoints (completions, chat completions, embeddings, audio, images, and batches). It also overrides authHeaders to use the api-key header instead of the standard Authorization: Bearer header when a string API key is provided.

Usage

Use AzureOpenAI whenever you need to call OpenAI models hosted on Azure rather than the OpenAI platform directly. This is the primary entry point for Azure-hosted deployments, supporting both API key and Microsoft Entra (Azure AD) token-based authentication.

Code Reference

Source Location

• Repository: openai-node
• File: src/azure.ts

Signature

export interface AzureClientOptions extends ClientOptions {
  apiVersion?: string | undefined;
  endpoint?: string | undefined;
  deployment?: string | undefined;
  apiKey?: string | undefined;
  azureADTokenProvider?: (() => Promise<string>) | undefined;
}

export class AzureOpenAI extends OpenAI {
  deploymentName: string | undefined;
  apiVersion: string;

  constructor(opts?: AzureClientOptions);
  override async buildRequest(
    options: FinalRequestOptions,
    props?: { retryCount?: number },
  ): Promise<{ req: RequestInit & { headers: Headers }; url: string; timeout: number }>;
  protected override async authHeaders(opts: FinalRequestOptions): Promise<NullableHeaders | undefined>;
}

Import

import { AzureOpenAI } from 'openai';

I/O Contract

Inputs

Name	Type	Required	Description
apiVersion	string	Yes	Azure API version string. Defaults to process.env['OPENAI_API_VERSION'].
endpoint	string	Conditional	Azure resource endpoint URL, e.g. https://example-resource.azure.openai.com/. Required if baseURL is not set. Defaults to process.env['AZURE_OPENAI_ENDPOINT'].
deployment	string	No	Model deployment name. When set, routes requests through /deployments/{deployment}.
apiKey	string	Conditional	Azure API key. Mutually exclusive with azureADTokenProvider. Defaults to process.env['AZURE_OPENAI_API_KEY'].
azureADTokenProvider	() => Promise<string>	Conditional	Async function returning a Microsoft Entra access token. Mutually exclusive with apiKey.
baseURL	string	No	Custom base URL. Mutually exclusive with endpoint. Defaults to process.env['OPENAI_BASE_URL'].

Outputs

Name	Type	Description
(instance)	AzureOpenAI	An initialized Azure OpenAI client that exposes the same resource methods as the base OpenAI client (chat, completions, embeddings, etc.).
deploymentName	undefined	The configured deployment name, if any.
apiVersion	string	The resolved Azure API version string.

Usage Examples

import { AzureOpenAI } from 'openai';

// Using API key authentication
const client = new AzureOpenAI({
  apiKey: 'my-azure-api-key',
  endpoint: 'https://my-resource.openai.azure.com/',
  apiVersion: '2024-06-01',
  deployment: 'gpt-4o',
});

const completion = await client.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello!' }],
});

// Using Azure AD token provider
const adClient = new AzureOpenAI({
  azureADTokenProvider: async () => getAccessToken(),
  endpoint: 'https://my-resource.openai.azure.com/',
  apiVersion: '2024-06-01',
});

Related Pages

• Environment:Openai_Openai_node_Node_20_Runtime