API access to Microsoft's Azure OpenAI models

Install this plugin in the same environment as LLM.

llm install llm-azure-openai

Export the environment variables:

export AZURE_TENANT_ID='...'
export AZURE_CLIENT_ID='...'
export AZURE_CLIENT_SECRET='...'

Configure the models you want to use from your deployments

On Mac: ~/Library/Application Support/io.datasette.llm/azure-openai-models.yaml

On Linux: ~/.config/io.datasette.llm/azure-openai-models.yaml

- model_id: o3-mini
  model_name: o3-mini
  azure_endpoint: https://example.openai.azure.com
  api_version: '2024-12-01-preview'
  aliases: ['azure-o3-mini']
  use_azure_ad: true
- model_id: gpt-35-turbo-blue
  model_name: gpt-35-turbo-blue
  azure_endpoint: https://example.openai.azure.com
  api_version: '2024-02-01'
  aliases: ['azure-gpt-35']
  use_azure_ad: true

Now run the model using -m your-model, for example:

llm -m azure-o3-mini "..."

To set up this plugin locally, first checkout the code. Then create a new virtual environment:

cd llm-azure-openai
python3 -m venv venv
source venv/bin/activate

Now install the dependencies and test dependencies:

llm install -e '.[test]'

To run the tests:

pytest

This package is signed using sigstore to provide supply chain security. When you install this package from PyPI, you can verify its authenticity by checking the digital signatures.