Azure Custom Provider for AI SDK

The Quail-AI/azure-ai-provider enables unofficial integration with Azure-hosted language models that use Azure's native APIs instead of the standard OpenAI API format.

Language Models

This provider works with any model in the Azure AI Foundry that is compatible with the Azure-Rest AI-inference API.

Models Tested:

DeepSeek-R1
LLama 3.3-70B Instruct
Cohere-command-r-08-2024

Setup

Installation

Install the provider via npm:

bash

npm i @quail-ai/azure-ai-provider

Provider Instance

Create an Azure AI resource and set up your endpoint URL and API key. Add the following to your .env file:

bash

AZURE_API_ENDPOINT=https://<your-resource>.services.ai.azure.com/models
AZURE_API_KEY=<your-api-key>

Import createAzure from the package to create your provider instance:

import { createAzure } from '@quail-ai/azure-ai-provider';

const azure = createAzure({
  endpoint: process.env.AZURE_API_ENDPOINT,
  apiKey: process.env.AZURE_API_KEY,
});

Basic Usage

Generate text using the Azure custom provider:

import { generateText } from 'ai';

const { text } = await generateText({
  model: azure('your-deployment-name'),
  prompt: 'Write a story about a robot.',
});

Status

✅ Chat Completions: Working with both streaming and non-streaming responses
⚠️ Tool Calling: Functionality highly dependent on model choice
⚠️ Embeddings: Implementation present but untested