Azure OpenAI¶

Uses the openai SDK's AzureOpenAI client. No extra install needed (openai is a core dependency).

Configure¶

Azure needs three values:

export AZURE_OPENAI_API_KEY=<from Azure portal -> Keys & Endpoint>
export AZURE_OPENAI_ENDPOINT=https://<resource>.openai.azure.com/
export AZURE_OPENAI_API_VERSION=2024-08-01-preview

Use¶

Prefix: azure/. The model ID after the prefix is your deployment name (the name you gave when deploying the model in Azure AI Studio), not the underlying base model name.

curl http://localhost:11434/api/chat -d '{
  "model": "azure/my-gpt4o-deployment",
  "messages": [{"role": "user", "content": "Hello!"}]
}'

Multiple resources / API versions¶

For deployments spread across multiple Azure resources or API versions, use the AZURE_DEPLOYMENTS dict in config.py:

AZURE_DEPLOYMENTS = {
    "gpt4o-prod": {
        "api_key": "...",
        "endpoint": "https://prod.openai.azure.com/",
        "api_version": "2024-08-01-preview",
    },
    "gpt35-dev": {
        "api_key": "...",
        "endpoint": "https://dev.openai.azure.com/",
        "api_version": "2024-02-01",
    },
}

Deployments not listed here fall back to the default AZURE_CONFIG.

Embeddings¶

Azure embeddings are supported via /api/embed and /api/embeddings — just route through an embedding deployment:

azure/my-text-embedding-3-small-deployment