Azure OpenAI
Configure MyDeskBot to use Azure OpenAI Service.
Overview
Azure OpenAI Service provides access to powerful OpenAI models with enterprise-grade security and compliance.
Supported Models
Azure OpenAI supports the same models as OpenAI, hosted on Azure infrastructure:
- GPT-4o - Most capable model, multimodal
- GPT-4 Turbo - High performance, cost-effective
- GPT-4 - Advanced reasoning
- GPT-3.5 Turbo - Fast, cost-effective
Getting Started
1. Create Azure OpenAI Resource
- Sign in to Azure Portal
- Search for "Azure OpenAI"
- Click "Create"
- Fill in the required fields:
- Resource group
- Region (select one with available capacity)
- Name
- Pricing tier
- Click "Create" and wait for deployment
2. Get API Keys and Endpoint
- Navigate to your Azure OpenAI resource
- Go to "Keys and Endpoint"
- Copy your API key (you have two, either works)
- Copy the endpoint URL
3. Deploy a Model
- Go to "Deployments" in your resource
- Click "Create deployment"
- Select a model (e.g., gpt-4o)
- Choose a deployment name
- Click "Create"
Configuration Options
Model Selection
The model you can use depends on your Azure deployment:
| Deployment | Best For | Cost | Speed |
|---|---|---|---|
| gpt-4o | Complex tasks, multimodal | High | Medium |
| gpt-4-turbo | Most tasks | Medium | Fast |
| gpt-4 | Advanced reasoning | High | Slow |
| gpt-35-turbo | Simple tasks, high volume | Low | Fast |
Parameters
Same as OpenAI models:
- Temperature (0-2) - Controls randomness
- Max Tokens - Maximum response length
- Top P - Alternative to temperature
- Frequency Penalty - Reduce repetition
- Presence Penalty - Encourage new topics
Pricing
Azure OpenAI pricing differs from OpenAI:
- Pay-as-you-go - Pay per token
- PTUs (Provisioned Throughput Units) - Reserved capacity
- Free tier - Limited credits for new accounts
Check Azure OpenAI Pricing for current rates.
Rate Limits
Azure OpenAI uses tokens per minute (TPM) and requests per minute (RPM) limits:
- Standard tiers - Shared quota
- PTUs - Reserved capacity, higher limits
Monitor usage in Azure Portal.
Why Choose Azure OpenAI?
Enterprise Features
- Data Privacy - Enterprise data privacy commitments
- Compliance - SOC, HIPAA, GDPR compliant
- Security - Azure security features
- Responsible AI - Azure AI safety standards
Integration
- Azure Services - Integrate with Azure Functions, Logic Apps
- Private Endpoints - Private network access
- Virtual Network - Network isolation
- Managed Identity - Secure authentication
Deployment Models
- Standard - Shared capacity
- Provisioned - Reserved capacity, predictable performance
- Global - Deploy to multiple regions
Troubleshooting
Invalid API Key
- Verify your API key is correct
- Check the key hasn't expired
- Ensure you're using the right Azure resource
Deployment Not Found
- Verify the deployment name
- Check the model is deployed
- Ensure the deployment is in the same region
Region Issues
- Some models are only available in specific regions
- Check Azure OpenAI availability in your region
- Consider deploying to a different region
Quota Exceeded
- Check your quota limits in Azure Portal
- Request quota increase if needed
- Consider PTUs for consistent performance