Python: Bug: Model o1 (o3-mini) is enabled only for api versions 2024-12-01-preview and later

[tamer-abdulghani]

Describe the bug
When making an await kernel.invoke(function=function, arguments=variables) call against the o1 & o3-mini models using the Azure OpenAI service, the following error is returned:

Model {modelName} is enabled only for api versions 2024-12-01-preview and later

We have other gpt models deployed in the AOAI service, all others are working well without any issue.

To Reproduce
Steps to reproduce the behavior:

1. Create a kernel chat completion service targeting o1 as the deployment model
2. Create an agent with instructions, temperature, and structured output
3. Call kernel.invoke()

Expected behavior
Response should be received without throwing this error.

Screenshots
N/A

Platform

• Language: Python
• Source: semantic-kernel==1.31.0
• AI model: o1:2024-12-17 & o3-mini:2025-01-31
• IDE: VS Code
• OS: Windows

[moonbox3]

Hi @tamer-abdulghani, as the error says, it requires you to pass in an Azure OpenAI api version (either by specifiying the kwarg directly or by configuring AZURE_OPENAI_API_VERSION) to at least 2024-12-01-preview. Azure OpenAI only started providing support for o-series models from the API version. If you don't specify an api version via .env/env vars/kwarg, it falls back to Azure OpenAI's latest GA API version, which is not 2024-12-01-preview.

[moonbox3]

@tamer-abdulghani, please have a look at the following table listing allowed API versions: https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/reasoning?tabs=python-secure%2Cpy#api--feature-support

I just tested a call against o3-mini with an api version of 2024-10-01-preview and I can see:

semantic_kernel.exceptions.service_exceptions.ServiceResponseException: 
("<class 'semantic_kernel.connectors.ai.open_ai.services.azure_chat_completion.AzureChatCompletion'> 
service failed to complete the prompt", 
BadRequestError("Error code: 400 - {'error': {'code': 'BadRequest', 'message': 'Model {modelName} is enabled only for api versions 2024-12-01-preview and later'}}"))

When I configure the api version to be >= 2024-12-01-preview, I am able to make the call.