Azure

Runtime provider id: azure. Uses the Vercel AI SDK @ai-sdk/azure provider to reach Azure OpenAI.

Provider options

Field	Type	Notes
resourceName	string	Optional. Your Azure OpenAI resource name.
baseURL	string	Optional. Override the resource-derived endpoint.
apiKey	string	Optional (or set via environment).
headers	Record<string, string>	Optional. Extra HTTP headers.
apiVersion	string	Optional. Azure OpenAI API version.
useDeploymentBasedUrls	boolean	Optional. Use /deployments/<name> style URLs.
reasoning	object	Optional. See Reasoning.

Important: set the LLM model field to your Azure deployment name, not the raw OpenAI model id.

Reasoning

Reasoning is forwarded to the Azure OpenAI Responses API; only reasoning-capable deployments honor effort/summary. display is not applicable to Azure.

Example configs

{
  "name": "Azure OpenAI",
  "provider": "azure",
  "options": {
    "resourceName": "my-resource",
    "apiKey": "...",
    "apiVersion": "2024-10-21"
  }
}

{
  "name": "GPT-4o (Azure)",
  "provider": "Azure OpenAI",
  "options": {
    "model": "my-gpt-4o-deployment"
  },
  "default": true
}

With reasoning summaries on a reasoning-capable deployment:

{
  "name": "Azure Reasoning",
  "provider": "azure",
  "options": {
    "resourceName": "my-resource",
    "apiKey": "...",
    "reasoning": { "effort": "medium", "summary": "detailed" }
  }
}

{
  "name": "o4-mini (Azure)",
  "provider": "Azure Reasoning",
  "options": {
    "model": "my-o4-mini-deployment"
  }
}