Custom Azure OpenAI client

This is only for using a Custom Azure OpenAI client

This note only meant for those who are trying to use a custom Azure client, and is NOT TYPICAL for most users. For typical usage of Azure-deployed models with Langroid, see the docs, the test_azure_openai.py and example/basic/chat.py

Example showing how to use Langroid with Azure OpenAI and Entra ID authentication by providing a custom client.

By default, Langroid manages the configuration and creation of the Azure OpenAI client (see the Setup guide for details). In most cases, the available configuration options are sufficient, but if you need to manage any options that are not exposed, you instead have the option of providing a custom client, in Langroid v0.29.0 and later.

In order to use a custom client, you must provide a function that returns the configured client. Depending on whether you need to make synchronous or asynchronous calls, you need to provide the appropriate client. A sketch of how this is done (supporting both sync and async calls) is given below:

def get_azure_openai_client():
    return AzureOpenAI(...)

def get_azure_openai_async_client():
    return AsyncAzureOpenAI(...)

lm_config = lm.AzureConfig(
    azure_openai_client_provider=get_azure_openai_client,
    azure_openai_async_client_provider=get_azure_openai_async_client,
)

Microsoft Entra ID Authentication

A key use case for a custom client is Microsoft Entra ID authentication. Here you need to provide an azure_ad_token_provider to the client. For examples on this, see examples/basic/chat-azure-client.py and examples/basic/chat-azure-async-client.py.