Microsoft Azure OpenAI

v1 API

For Azure OpenAI's current v1 API, use the standard OpenAI client with your Azure endpoint:

import OpenAI from 'openai';
import { getBearerTokenProvider, DefaultAzureCredential } from '@azure/identity';

const endpoint = process.env['AZURE_OPENAI_ENDPOINT'];
const deployment = process.env['AZURE_OPENAI_DEPLOYMENT'];
if (!endpoint || !deployment) throw new Error('Missing Azure OpenAI configuration');

const tokenProvider = getBearerTokenProvider(new DefaultAzureCredential(), 'https://ai.azure.com/.default');

const openai = new OpenAI({
  baseURL: `${endpoint.replace(/\/+$/, '')}/openai/v1/`,
  apiKey: tokenProvider,
});

const result = await openai.chat.completions.create({
  model: deployment,
  messages: [{ role: 'user', content: 'Say hello!' }],
});

console.log(result.choices[0]!.message?.content);

With the v1 API, the model parameter is your Azure deployment name. See the Azure examples for complete runnable programs.

Dated API versions

For dated Azure OpenAI API versions, use the AzureOpenAI class instead of the OpenAI class.

import { AzureOpenAI } from 'openai';
import { getBearerTokenProvider, DefaultAzureCredential } from '@azure/identity';

const credential = new DefaultAzureCredential();
const scope = 'https://cognitiveservices.azure.com/.default';
const azureADTokenProvider = getBearerTokenProvider(credential, scope);

const openai = new AzureOpenAI({
  azureADTokenProvider,
  apiVersion: '<The API version, e.g. 2024-10-01-preview>',
});

const result = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Say hello!' }],
});

console.log(result.choices[0]!.message?.content);

For more information on support for the Azure API, see azure.md.

Realtime API

This SDK provides real-time streaming capabilities for Azure OpenAI through the OpenAIRealtimeWS and OpenAIRealtimeWebSocket clients described previously.

To utilize the real-time features, begin by creating a fully configured AzureOpenAI client and passing it into either OpenAIRealtimeWS.azure or OpenAIRealtimeWebSocket.azure. For example:

const cred = new DefaultAzureCredential();
const scope = 'https://cognitiveservices.azure.com/.default';
const deploymentName = 'gpt-4o-realtime-preview-1001';
const azureADTokenProvider = getBearerTokenProvider(cred, scope);
const client = new AzureOpenAI({
  azureADTokenProvider,
  apiVersion: '2024-10-01-preview',
  deployment: deploymentName,
});
const rt = await OpenAIRealtimeWS.azure(client);

Once the instance has been created, you can then begin sending requests and receiving streaming responses in real time.