Models Selector

Foundry Models Selector

Configure AZURE_LOCATION and AI_PROJECT_DEPLOYMENTS azd environment variables by selecting the right models and regions for the user's needs.

Preferred Models by Task & Modality

Use this section to recommend models based on what the user needs. Always suggest the best-fit model first, then offer alternatives.

Chat & General Purpose (Text In / Text Out)

Advanced Reasoning & Complex Problem Solving

Coding & Development

Multimodal (Text + Image Input)

Embeddings

Image Generation

Document Processing

Cost-Optimized Routing

How to Set Environment Variables

Set Azure Location

azd env set AZURE_LOCATION "<region>"

Example:

azd env set AZURE_LOCATION "eastus2"

Set Model Deployments

azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'<deployment-name>', 'model': {'name': '<model-name>','format': '<format>', 'version':'<version>'},'sku': {'name': '<sku-name>','capacity':<capacity>}}]"

Single model example:

azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'gpt-5-mini', 'model': {'name': 'gpt-5-mini','format': 'OpenAI', 'version':'2025-08-07'},'sku': {'name': 'GlobalStandard','capacity':100}}]"

Multiple models example:

azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'gpt-4.1-mini', 'model': {'name': 'gpt-4.1-mini','format': 'OpenAI', 'version':'2025-04-14'},'sku': {'name': 'GlobalStandard','capacity':100}}, {'name':'DeepSeek-V3.2', 'model': {'name': 'DeepSeek-V3.2','format': 'DeepSeek', 'version':'1'},'sku': {'name': 'GlobalStandard','capacity':100}}]"

Deployment Entry Format

Each deployment entry in the JSON array:

{
  'name': '<deployment-name>',        # Deployment name (typically same as model name)
  'model': {
    'name': '<model-name>',           # Exact model name from tables below
    'format': '<format>',             # Provider format (see Model Catalog)
    'version': '<version>'            # Model version string
  },
  'sku': {
    'name': '<sku-name>',             # Usually 'GlobalStandard'
    'capacity': <number>              # TPM in thousands (e.g., 100 = 100K TPM)
  }
}

Model Catalog

Azure OpenAI Models (format: OpenAI)

Azure OpenAI OSS Models (format: OpenAI-OSS)

DeepSeek Models (format: DeepSeek)

Meta Models (format: Meta)

Microsoft Models (format: Microsoft)

Mistral Models (format: MistralAI)

Cohere Models (format: Cohere)

Moonshot AI Models (format: Moonshot)

xAI Models (format: xAI)

Black Forest Labs Models (format: BlackForestLabs)

Region Availability

Regions with Maximum Azure OpenAI Model Coverage

These regions have the broadest model availability. Prefer these when needing the widest selection:

Regions with Good Azure OpenAI Coverage

These regions support all mainstream models (GPT-5-mini/nano, GPT-4.1 series, o-series, GPT-4o, embeddings):

Regions with Standard Azure OpenAI Coverage

Support GPT-4.1 series, o-series, GPT-4o, embeddings, but fewer GPT-5 variants:

Non-Azure-OpenAI Models (DeepSeek, Meta, Cohere, xAI, Mistral, Moonshot, Black Forest Labs)

Most non-OpenAI models sold directly by Azure are available in all Global Standard regions listed above. Notable exceptions:

• Mistral-Large-3: West US 3 only
• model-router: East US 2 and Sweden Central only
• codex-mini, o3-pro: East US 2 and Sweden Central only

Registration-Required Models

These models require application approval before use:

• gpt-5, gpt-5-pro, gpt-5-codex, gpt-5.1, gpt-5.1-codex, gpt-5.1-codex-max, gpt-5.2, gpt-5.2-codex
• computer-use-preview
• gpt-image-1, gpt-image-1-mini, gpt-image-1.5
• grok-code-fast-1, grok-4

Models not requiring registration: gpt-5-mini, gpt-5-nano, gpt-5-chat, gpt-4.1 series, o4-mini, o3, all DeepSeek, Meta, Cohere, Moonshot, and most xAI models.

Workflow

1. Ask the user what task/modality they need (chat, reasoning, coding, embeddings, image gen, etc.)
2. Recommend models from "Preferred Models by Task" section above
3. Confirm which region works for them (suggest eastus2 or swedencentral for maximum coverage)
4. Confirm capacity needs (default to 100 for chat models, 30 for embedding models)
5. Run the azd env set commands:

azd env set AZURE_LOCATION "<selected-region>"
azd env set AI_PROJECT_DEPLOYMENTS "[<deployment-entries>]"

Quick-Start Examples

Basic Chat App

azd env set AZURE_LOCATION "eastus2"
azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'gpt-5-mini', 'model': {'name': 'gpt-5-mini','format': 'OpenAI', 'version':'2025-08-07'},'sku': {'name': 'GlobalStandard','capacity':100}}]"

RAG Application (Chat + Embeddings)

azd env set AZURE_LOCATION "eastus2"
azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'gpt-5-mini', 'model': {'name': 'gpt-5-mini','format': 'OpenAI', 'version':'2025-08-07'},'sku': {'name': 'GlobalStandard','capacity':100}}, {'name':'text-embedding-3-large', 'model': {'name': 'text-embedding-3-large','format': 'OpenAI', 'version':'1'},'sku': {'name': 'GlobalStandard','capacity':30}}]"

Coding Agent

azd env set AZURE_LOCATION "eastus2"
azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'gpt-4.1', 'model': {'name': 'gpt-4.1','format': 'OpenAI', 'version':'2025-04-14'},'sku': {'name': 'GlobalStandard','capacity':100}}, {'name':'o4-mini', 'model': {'name': 'o4-mini','format': 'OpenAI', 'version':'2025-04-16'},'sku': {'name': 'GlobalStandard','capacity':50}}]"

Multi-Provider Setup

azd env set AZURE_LOCATION "eastus2"
azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'gpt-4.1-mini', 'model': {'name': 'gpt-4.1-mini','format': 'OpenAI', 'version':'2025-04-14'},'sku': {'name': 'GlobalStandard','capacity':100}}, {'name':'DeepSeek-R1-0528', 'model': {'name': 'DeepSeek-R1-0528','format': 'DeepSeek', 'version':'1'},'sku': {'name': 'GlobalStandard','capacity':100}}]"

Cost-Optimized with Model Router

azd env set AZURE_LOCATION "eastus2"
azd env set AI_PROJECT_DEPLOYMENTS "[{'name':'model-router', 'model': {'name': 'model-router','format': 'Microsoft', 'version':'2025-11-18'},'sku': {'name': 'GlobalStandard','capacity':100}}]"

Reference Links

• Azure OpenAI models
• Other model collections (DeepSeek, Meta, xAI, etc.)
• Deployment types
• Azure OpenAI pricing