Azure OpenAI Provider¶

Integrates Azure OpenAI Service into Amplifier.

Module ID¶

provider-azure-openai

Installation¶

providers:
  - module: provider-azure-openai
    source: git+https://github.com/microsoft/amplifier-module-provider-azure-openai@main
    config:
      azure_endpoint: https://your-resource.openai.azure.com
      deployment_name: your-deployment-name

Authentication¶

The provider supports three authentication methods:

API Key (Default)¶

config:
  azure_endpoint: https://your-resource.openai.azure.com
  api_key: ${AZURE_OPENAI_API_KEY}

Or via environment variable:

export AZURE_OPENAI_API_KEY="your-api-key"
export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com"

Managed Identity¶

For Azure-hosted environments (VMs, App Service, AKS, etc.), use managed identity for passwordless authentication.

Prerequisite

Requires the azure-identity package: pip install azure-identity

System-Assigned IdentityUser-Assigned Identity

config:
  azure_endpoint: https://your-resource.openai.azure.com
  use_managed_identity: true

Or via environment:

export AZURE_USE_MANAGED_IDENTITY=true

config:
  azure_endpoint: https://your-resource.openai.azure.com
  use_managed_identity: true
  managed_identity_client_id: your-client-id

Or via environment:

export AZURE_USE_MANAGED_IDENTITY=true
export AZURE_MANAGED_IDENTITY_CLIENT_ID="your-client-id"

Default Azure Credential¶

Uses the Azure SDK's DefaultAzureCredential chain (CLI, environment, managed identity fallback):

config:
  azure_endpoint: https://your-resource.openai.azure.com
  use_default_credential: true

Or via environment:

export AZURE_USE_DEFAULT_CREDENTIAL=true

Configuration¶

Option	Type	Default	Description
`azure_endpoint`	string	`$AZURE_OPENAI_ENDPOINT`	Azure OpenAI resource endpoint
`api_key`	string	`$AZURE_OPENAI_API_KEY`	API key (if using key auth)
`api_version`	string	`2024-10-01-preview`	Azure API version
`deployment_name`	string	null	Azure deployment name (maps to model)
`default_model`	string	`gpt-5.4`	Default model name
`max_tokens`	integer	`4096`	Maximum output tokens
`temperature`	float	null	Sampling temperature
`reasoning`	string	null	Reasoning effort: `minimal\\|low\\|medium\\|high`
`reasoning_summary`	string	`detailed`	Reasoning verbosity: `auto\\|concise\\|detailed`
`truncation`	string	`auto`	Automatic context management
`enable_state`	boolean	`false`	Enable stateful conversations
`use_managed_identity`	boolean	`false`	Use Azure Managed Identity
`use_default_credential`	boolean	`false`	Use DefaultAzureCredential
`managed_identity_client_id`	string	null	Client ID for user-assigned identity
`debug`	boolean	`false`	Enable standard debug events
`raw_debug`	boolean	`false`	Enable ultra-verbose raw API I/O logging
`timeout`	float	600.0	API timeout in seconds
`max_retries`	int	5	Retry attempts before failing

Deployment Mapping¶

Azure uses deployment names instead of model names. Map them:

providers:
  - module: provider-azure-openai
    config:
      azure_endpoint: https://your-resource.openai.azure.com
      deployment_name: my-gpt5-deployment
      default_model: gpt-5.4

Or use environment variables:

export AZURE_OPENAI_DEPLOYMENT_NAME="my-gpt5-deployment"

Reasoning Configuration¶

Control reasoning effort and verbosity:

providers:
  - module: provider-azure-openai
    config:
      reasoning: "medium"
      reasoning_summary: "detailed"

Tool Calling¶

Supports OpenAI Responses API tool calling:

providers:
  - module: provider-azure-openai
    config:
      azure_endpoint: https://your-resource.openai.azure.com
      deployment_name: my-deployment

Tools are automatically available when declared in your configuration.

Environment Variables¶

# Required
export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com"

# Authentication (choose one)
export AZURE_OPENAI_API_KEY="your-api-key"
# OR
export AZURE_USE_MANAGED_IDENTITY=true
# OR
export AZURE_USE_DEFAULT_CREDENTIAL=true

# Optional
export AZURE_OPENAI_API_VERSION="2024-10-01-preview"
export AZURE_OPENAI_DEPLOYMENT_NAME="my-deployment"
export AZURE_OPENAI_DEFAULT_MODEL="gpt-5.1-codex"

Example Configuration¶

API Key Authentication¶

providers:
  - module: provider-azure-openai
    source: git+https://github.com/microsoft/amplifier-module-provider-azure-openai@main
    config:
      azure_endpoint: https://myresource.openai.azure.com
      api_key: ${AZURE_OPENAI_API_KEY}
      deployment_name: my-gpt5-deployment
      default_model: gpt-5.4
      max_tokens: 4096

Managed Identity Authentication¶

providers:
  - module: provider-azure-openai
    source: git+https://github.com/microsoft/amplifier-module-provider-azure-openai@main
    config:
      azure_endpoint: https://myresource.openai.azure.com
      use_managed_identity: true
      deployment_name: my-gpt5-deployment
      default_model: gpt-5.4

DefaultAzureCredential (Development)¶

providers:
  - module: provider-azure-openai
    source: git+https://github.com/microsoft/amplifier-module-provider-azure-openai@main
    config:
      azure_endpoint: https://myresource.openai.azure.com
      use_default_credential: true
      deployment_name: my-gpt5-deployment
      default_model: gpt-5.4