Adding file_search + vector_store_ids to assistants endpoint

Question

I am working with the assistant preview endpoint on the Azure OpenAI Service. I have 2 questions to this endpoint.

1. I have a 4o model deployment. The assistants endpoint is available when using version 2024-05-13 the assistants endpoint is available, but when running version 2024-08-06 the run generates this error:

{

  "code": "invalid_engine_error",

  "message": "Unsupported engine: gpt-4o-2024-08-06 for deployment GPT-4o. Please make sure you are using model version 0613 or newer (0613, 1106, 0125) or gpt-4 turbo or gpt-35 turbo. Version 0301 does NOT support assistants."

}

2. According to the documentation found here it should be possible to attach a vector store to both the assistant and the thread. When I attach a vector store to my assistant, i get a response that looks correct (answer shortened for brievety):

  "tools": [
    {
      "type": "file_search"
    }
  ],
  "top_p": 1.0,
  "temperature": 1.0,
  "tool_resources": {
    "file_search": {
      "vector_store_ids": "vs_3ZLO7p2c5nC5TDyCPqfDZdR3"
    }
  },

However, when I retrieve the agent form the endpoint again, file_search has been replaced by retrieval:

  "tools": [
    {
      "type": "retrieval"
    }
  ],
  "top_p": 1.0,
  "temperature": 1.0,
  "file_ids": [],

If I then prompt the assistant in the thread/run flow, it complains that no specific files or documents have been provided.

Is the documentation correct or can vector stores only be added to threads?

Answer 1

Hello Axel,

Thanks for reaching out to us, Azure OpenAI Assistants file search tool (Preview) is only available in 2024-05-01-preview version.

reference - https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/file-search?tabs=rest#api-version

Assistant is a preview feature and not available in 0806 version, please check on below table for more details -

https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models#assistants-preview

I hope this helps.

Regards,

Yutong

-Please kindly accept the answer if you feel helpful to support the community, thanks a lot.