Error "Invalid URL (POST /v1/chat/completions)" when using gpt-4o-mini with azure-ai-inference library

[nischays]

• Package Name: azure-ai-inference
• Package Version: 1.0.0b9
• Python Version: 3.11

Describe the bug
When attempting to use the gpt-4o-mini model (Deployment type: Standard) with ChatCompletionsClient, an error occurs:

Error: (None) Invalid URL (POST /v1/chat/completions)
Code: None
Message: Invalid URL (POST /v1/chat/completions)

To Reproduce
Steps to reproduce the behavior:

1. Set the environment variables AZURE_INFERENCE_ENDPOINT and AZURE_INFERENCE_KEY correctly.
2. Run the following code:

import os
from azure.ai.inference import ChatCompletionsClient
from azure.core.credentials import AzureKeyCredential
from azure.ai.inference.models import SystemMessage, UserMessage

try:
    client = ChatCompletionsClient(
        endpoint=os.getenv("AZURE_INFERENCE_ENDPOINT"),
        credential=AzureKeyCredential(os.getenv("AZURE_INFERENCE_KEY")),
    )

    response = client.complete(
        messages=[
            SystemMessage(content="You are a helpful assistant."),
            UserMessage(content="What is the capital of France?"),
        ],
        model="gpt-4o-mini",
    )

    print("azure-ai-inference response:", response.choices[0].message.content)

except Exception as e:
    print(f"Error: {e}")

Expected behavior
I expected the API call to succeed and return a chat completion response from the gpt-4o-mini model, similar to how it works with other models like gpt-4o.

Screenshots
N/A (Error is text-based and printed to console)

Additional context

• I have verified that the endpoint and key are correct.
• The same code structure works with other models.
• The same model works when using the AzureOpenAI library

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @dargilco @jhakulin @trangevi.

[JeffreyCA]

Running into this for OpenAI models as well, via azure-ai-projects:

from azure.ai.projects import AIProjectClient

project = AIProjectClient.from_connection_string(
    conn_str=project_connection_string, credential=DefaultAzureCredential()
)

chat = project.inference.get_chat_completions_client()
response = chat.complete(
    model="gpt-4o-mini",
    messages=messages,
)

Non-OpenAI models seem to work fine though.

If using AIProjectClient, as a workaround you can use get_azure_openai_client instead:

project = AIProjectClient.from_connection_string(conn_str=project_connection_string, credential=credential)
openai_client = project.inference.get_azure_openai_client(api_version="2024-06-01")
response = openai_client.chat.completions.create(
    model='gpt-4o-mini',
    messages=messages,
)
print(response.choices[0].message.content)

Similar issue reported on Learn: Running into problems trying out a Microsoft Learn Sample on AI Foundry