Azure OpenAI

Integrate OpenAI models through Microsoft Azure with enterprise-grade security and compliance

Overview

Azure OpenAI Service provides REST API access to OpenAI’s models including GPT-4, GPT-3.5-Turbo, and embeddings through Microsoft Azure’s enterprise-grade infrastructure with enhanced security, compliance, and regional availability.

Base URL: https://{resourceName}.openai.azure.com/openai

Supported Features

• ✅ Chat Completions (including streaming)
• ✅ Completions (legacy)
• ✅ Embeddings
• ✅ Image Generation (DALL-E)
• ✅ Image Editing
• ✅ Text-to-Speech (TTS)
• ✅ Speech-to-Text (Whisper)
• ✅ Audio Translation
• ✅ Function Calling & Tools
• ✅ Vision (GPT-4 Vision)
• ✅ Batch API
• ✅ Fine-tuning
• ✅ Multiple Authentication Methods

Quick Start

Basic Configuration

Python

from portkey_ai import Portkey

client = Portkey(
    provider="azure-openai",
    api_key="***",                    # Azure API key
    resource_name="my-resource",      # Your Azure resource name
    deployment_id="gpt-4-deployment", # Your deployment name
    api_version="2024-02-15-preview"  # API version
)

response = client.chat.completions.create(
    model="gpt-4",  # Not used in Azure, deployment_id is used instead
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is Azure OpenAI?"}
    ]
)

print(response.choices[0].message.content)

Configuration Options

Required Parameters

Parameter	Description	Example
resource_name	Azure OpenAI resource name	my-openai-resource
deployment_id	Deployment name in Azure	gpt-4-deployment
api_version	Azure API version	2024-02-15-preview
api_key	Azure API key	***

Authentication Methods

Azure OpenAI supports multiple authentication methods:

1. API Key (Default)

client = Portkey(
    provider="azure-openai",
    api_key="***",
    resource_name="my-resource",
    deployment_id="gpt-4",
    api_version="2024-02-15-preview"
)

2. Azure AD Token

client = Portkey(
    provider="azure-openai",
    azure_ad_token="Bearer eyJ0eXAiOiJKV1QiLCJhbGc...",
    resource_name="my-resource",
    deployment_id="gpt-4",
    api_version="2024-02-15-preview"
)

3. Entra ID (Service Principal)

client = Portkey(
    provider="azure-openai",
    azure_auth_mode="entra",
    azure_entra_tenant_id="***",
    azure_entra_client_id="***",
    azure_entra_client_secret="***",
    azure_entra_scope="https://cognitiveservices.azure.com/.default",
    resource_name="my-resource",
    deployment_id="gpt-4",
    api_version="2024-02-15-preview"
)

4. Managed Identity

client = Portkey(
    provider="azure-openai",
    azure_auth_mode="managed",
    azure_managed_client_id="***",  # Optional
    azure_entra_scope="https://cognitiveservices.azure.com/",
    resource_name="my-resource",
    deployment_id="gpt-4",
    api_version="2024-02-15-preview"
)

5. Workload Identity (Kubernetes)

client = Portkey(
    provider="azure-openai",
    azure_auth_mode="workload",
    azure_workload_client_id="***",
    azure_entra_scope="https://cognitiveservices.azure.com/.default",
    resource_name="my-resource",
    deployment_id="gpt-4",
    api_version="2024-02-15-preview"
)

# Requires environment variables:
# AZURE_AUTHORITY_HOST
# AZURE_TENANT_ID
# AZURE_FEDERATED_TOKEN_FILE

API Versions

Azure OpenAI uses API versions for versioning. Common versions:

API Version	Features	Status
2024-02-15-preview	Latest features, GPT-4 Turbo	Preview
2023-12-01-preview	GPT-4 Vision, DALL-E 3	Preview
2023-05-15	Stable release	GA
v1	OpenAI-compatible	Special

Use api_version="v1" for OpenAI-compatible endpoints without deployment IDs.

Available Deployments

You must create deployments in Azure before using them:

Model	Recommended Deployment Name	Capabilities
GPT-4	gpt-4	Advanced reasoning
GPT-4 Turbo	gpt-4-turbo	128K context
GPT-3.5 Turbo	gpt-35-turbo	Fast, cost-effective
text-embedding-ada-002	text-embedding-ada-002	Embeddings
DALL-E 3	dall-e-3	Image generation
Whisper	whisper	Speech-to-text

Advanced Features

Streaming

stream = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Count to 5"}],
    stream=True
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

Function Calling

tools = [
    {
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get weather information",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {"type": "string"}
                },
                "required": ["location"]
            }
        }
    }
]

response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "What's the weather in Seattle?"}],
    tools=tools
)

Vision

response = client.chat.completions.create(
    model="gpt-4-vision",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What's in this image?"},
            {
                "type": "image_url",
                "image_url": {"url": "https://example.com/image.jpg"}
            }
        ]
    }]
)

Embeddings

response = client.embeddings.create(
    model="text-embedding-ada-002",
    input="Azure OpenAI provides enterprise-grade AI"
)

embedding = response.data[0].embedding

Image Generation

response = client.images.generate(
    model="dall-e-3",
    prompt="A futuristic datacenter in the clouds",
    size="1024x1024",
    quality="hd"
)

image_url = response.data[0].url

Multi-Region Configuration

Load balance across multiple Azure regions:

config = {
    "strategy": {"mode": "loadbalance"},
    "targets": [
        {
            "provider": "azure-openai",
            "api_key": "***",
            "resource_name": "eastus-resource",
            "deployment_id": "gpt-4",
            "api_version": "2024-02-15-preview",
            "weight": 0.5
        },
        {
            "provider": "azure-openai",
            "api_key": "***",
            "resource_name": "westus-resource",
            "deployment_id": "gpt-4",
            "api_version": "2024-02-15-preview",
            "weight": 0.5
        }
    ]
}

client = Portkey().with_options(config=config)

Fallback to OpenAI

Fallback to standard OpenAI if Azure is unavailable:

config = {
    "strategy": {"mode": "fallback"},
    "targets": [
        {
            "provider": "azure-openai",
            "api_key": "***",
            "resource_name": "my-resource",
            "deployment_id": "gpt-4",
            "api_version": "2024-02-15-preview"
        },
        {
            "provider": "openai",
            "api_key": "sk-***",
            "override_params": {"model": "gpt-4"}
        }
    ]
}

client = Portkey().with_options(config=config)

Error Handling

from portkey_ai.exceptions import (
    RateLimitError,
    APIError,
    AuthenticationError
)

try:
    response = client.chat.completions.create(
        model="gpt-4",
        messages=[{"role": "user", "content": "Hello"}]
    )
except RateLimitError as e:
    print(f"Rate limit: {e}")
except AuthenticationError as e:
    print(f"Auth error: {e}")
except APIError as e:
    print(f"API error: {e}")

Key Differences from OpenAI

Aspect	OpenAI	Azure OpenAI
Authentication	API key	API key, AD token, Managed Identity
Endpoint	Fixed	Custom resource name
Model Specification	model parameter	deployment_id
API Versioning	Not required	Required api_version
Regional	Global	Multi-region support
Compliance	Standard	Enterprise (HIPAA, SOC 2, etc.)
Data Location	US	Choose your region

Request URL Structure

https://{resource_name}.openai.azure.com/openai/deployments/{deployment_id}/{endpoint}?api-version={api_version}

Example:

https://my-resource.openai.azure.com/openai/deployments/gpt-4/chat/completions?api-version=2024-02-15-preview

Best Practices

1. Use Managed Identity - Most secure for Azure-hosted applications
2. Deploy to multiple regions - Better availability and latency
3. Set appropriate api_version - Use stable versions for production
4. Monitor quota limits - Azure has per-deployment quotas
5. Use private endpoints - Enhanced security for enterprise
6. Implement retry logic - Handle transient failures
7. Cache responses - Reduce costs and latency

Enterprise Features

• Private Endpoints: Connect via Azure Private Link
• Customer Managed Keys: Bring your own encryption keys
• Virtual Networks: Restrict access to your VNet
• Managed Identity: Eliminate credential management
• Azure Monitor: Full observability integration
• Compliance: HIPAA, SOC 2, ISO 27001, GDPR
• Data Residency: Keep data in your region

Pricing

Azure OpenAI pricing is similar to OpenAI but billed through Azure:

Azure OpenAI Pricing

View Azure OpenAI Service pricing

Related Resources

OpenAI

Standard OpenAI integration

Load Balancing

Multi-region load balancing

Fallbacks

Fallback configurations

Enterprise Deployment

Enterprise deployment guide