Skip to main content

​
Overview

Azure OpenAI Service provides enterprise-grade access to OpenAI models through Microsoft Azure. It offers the same GPT-4 and GPT-3.5 models as OpenAI’s direct API, but with Azure’s security, compliance, and regional data residency guarantees. This makes Azure OpenAI the preferred choice for organizations that require enterprise-level data governance. Key differences from direct OpenAI:
  • Deployment-based — Models are deployed to named endpoints rather than selected by model ID
  • Regional hosting — Choose the Azure region where your data is processed
  • Enterprise security — Azure AD authentication, private endpoints, and managed identity support
  • Compliance — SOC 2, HIPAA, GDPR, and other certifications through Azure

​
Setup

1

Create an Azure OpenAI Resource

In the Azure Portal, create an Azure OpenAI resource. Select your subscription, resource group, and region.
Azure OpenAI requires approval. If you do not have access, apply at aka.ms/oai/access.
2

Deploy Models

In Azure OpenAI Studio, go to Deployments and create deployments for the models you need. Each deployment gets a unique name that you will use in Nadoo.Common deployment setup:
Deployment NameModelUse Case
gpt-4o-deploygpt-4oGeneral chat and reasoning
gpt-35-deploygpt-3.5-turboCost-efficient tasks
embedding-deploytext-embedding-ada-002Vector embeddings for RAG
3

Get Credentials

From your Azure OpenAI resource in the Azure Portal, navigate to Keys and Endpoint. Copy:
  • Endpoint URL (e.g., https://your-resource.openai.azure.com/)
  • API Key (Key 1 or Key 2)
4

Configure in Nadoo

Go to Admin > Model Providers > Azure OpenAI and enter:
FieldRequiredDescription
Endpoint URLYesYour Azure OpenAI endpoint (e.g., https://your-resource.openai.azure.com/)
API KeyYesOne of your Azure OpenAI API keys
API VersionYesThe API version to use (e.g., 2024-02-15-preview)
Deployment NamesYesComma-separated list of your model deployment names
5

Test Connection

Click Test to verify the endpoint and key are valid. Enable the deployments you want to use in your workspace.

​
Available Models

Azure OpenAI supports the same models as direct OpenAI, deployed under your custom deployment names.

​
Chat / LLM

ModelContext WindowBest For
gpt-4o128K tokensGeneral-purpose, multimodal, fastest GPT-4 class
gpt-4o-mini128K tokensCost-efficient tasks, high-volume workloads
gpt-4-turbo128K tokensComplex reasoning with large context
gpt-3.5-turbo16K tokensSimple tasks, lowest cost

​
Embedding

ModelDimensionsBest For
text-embedding-ada-0021536Standard embeddings for RAG
text-embedding-3-small1536Cost-efficient embeddings
text-embedding-3-large3072Highest quality embeddings
Model availability varies by Azure region. Check Azure OpenAI model availability for the latest regional support.

​
Capabilities

Chat Completion

Conversational AI with streaming, function calling, and JSON mode — identical capabilities to direct OpenAI.

Embeddings

Generate vector representations of text for semantic search and RAG knowledge bases.

Enterprise Security

Azure AD, managed identity, private endpoints, and virtual network integration.

Regional Compliance

Data stays in your chosen Azure region for regulatory compliance.

​
Deployment Configuration

Unlike direct OpenAI where you select a model by name, Azure OpenAI uses deployment names. When configuring an application in Nadoo, you select your Azure deployment instead of the base model name. 
Direct OpenAI:  model = "gpt-4o"
Azure OpenAI:   deployment = "my-gpt4o-deployment"

​
Multiple Deployments

You can create multiple deployments of the same model for different purposes:
  • Production deployment with higher token-per-minute (TPM) quota
  • Development deployment with lower quota for testing
  • Embedding deployment dedicated to knowledge base indexing

​
Environment Variables

When self-hosting, configure Azure OpenAI via environment variables: 
AZURE_OPENAI_API_KEY=your-api-key-here
AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/
AZURE_OPENAI_API_VERSION=2024-02-15-preview
If both environment variables and the admin UI configuration are set, the admin UI values take precedence.
Use CaseRecommended DeploymentReason
General chatbotgpt-4oBest balance of speed, quality, and cost
High-volume supportgpt-4o-mini or gpt-3.5-turboLow cost with strong performance
Document analysisgpt-4oLong context window with vision support
Knowledge base searchtext-embedding-3-smallHigh quality at low cost
Premium RAG pipelinetext-embedding-3-largeMaximum embedding quality

​
Rate Limits and Quotas

Azure OpenAI uses Tokens Per Minute (TPM) quotas that are configured per deployment. You can adjust quotas in the Azure Portal under your deployment’s settings.
Azure OpenAI quotas are shared across all applications using the same deployment. Monitor usage in the Azure Portal and increase quotas before scaling production workloads.

​
Troubleshooting

The endpoint URL or deployment name is incorrect. Verify both values in the Azure Portal under your OpenAI resource’s Keys and Endpoint and Deployments sections.
Your API key is invalid or has been regenerated. Copy a fresh key from the Azure Portal.
You have exceeded your deployment’s TPM quota. Increase the quota in the Azure Portal or distribute traffic across multiple deployments.
Not all models are available in every Azure region. Check model availability and consider creating a resource in a supported region.