Anvia

Azure OpenAI ​

@anvia/azure connects Anvia to Azure OpenAI and Azure AI Foundry endpoints. One AzureOpenAIClient creates provider-neutral models for completions, embeddings, image generation, speech generation, and transcription from your Azure deployments.

ts
import { AzureOpenAIClient } from '@anvia/azure'

export const azure = new AzureOpenAIClient({
  endpoint: process.env.AZURE_OPENAI_ENDPOINT!,
  apiKey: process.env.AZURE_OPENAI_API_KEY!,
})

Azure owns endpoint, authentication, and deployment configuration here. Request and response mapping is shared with @anvia/openai, so agents, extractors, and pipelines receive the same Anvia model contracts.

Choose a model ​

Use completionModel({ modelId, api }) for agents, direct completions, tools, and structured output. modelId is your Azure deployment name, and api is 'responses' or 'chat' (the default).

Use embeddingModel(), imageGenerationModel(), speechGenerationModel(), and transcriptionModel() for the other capabilities, and listModels() for the endpoint's model list. Each needs a matching deployment, and availability depends on your Azure endpoint.

What the provider owns ​

The package configures the endpoint and credentials, labels models with the azure-openai provider identity, adds explicit message types to Responses requests for Foundry project endpoints, and restores function names that Azure omits from terminal tool-argument stream events. The application still owns instructions, tool permissions, memory, deployment selection, credential storage, fallback policy, retries, and observability.

Start here ​

  1. Install and configure the client, with an API key or Microsoft Entra.
  2. Create deployment models for completions and other capabilities.
  3. Review the production checklist, including migration from OpenAIClient.

For package details, see the @anvia/azure reference.

Built for Anvia.