"Unknown parameter: 'tool_resources.file_search.file_ids'."

[zhoumengbo]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• This is an issue with the Python library

Describe the bug

I want to use Assistants to upload a document and then ask the model some questions based on the document content. When I specify file_ids, I always get a parameter error. Could you please help me figure out where the problem is?

I referred to the following link: https://platform.openai.com/docs/assistants/migration/agents#accessing-v1-data-in-v2

   file = client.files.create(
        file=open(file_path, "rb"),
        purpose='assistants'
    )
    response = client.beta.assistants.create(
        model="gpt-4o-mini",
        name="QA Assistant",
        description="An assistant to answer questions based on uploaded documents",
        instructions="You are a helpful assistant who answers questions based on the information provided. Be concise and clear.",
        tools=[{"type": "file_search"}],
        tool_resources={"file_search": {"file_ids": [file.id]}},
        # file_ids=[file.id],
        temperature=0.0,
        top_p=1
    )

The above code produces the error：

Traceback (most recent call last):
File "D:\product\auto_test\openai_api\test-pdf-assistants.py", line 85, in
main()
File "D:\product\auto_test\openai_api\test-pdf-assistants.py", line 50, in main
response = client.beta.assistants.create(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\software\Anaconda\Anaconda\envs\textgrad\Lib\site-packages\openai\resources\beta\assistants.py", line 146, in create
return self._post(
^^^^^^^^^^^
File "D:\software\Anaconda\Anaconda\envs\textgrad\Lib\site-packages\openai_base_client.py", line 1280, in post
return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\software\Anaconda\Anaconda\envs\textgrad\Lib\site-packages\openai_base_client.py", line 957, in request
return self._request(
^^^^^^^^^^^^^^
File "D:\software\Anaconda\Anaconda\envs\textgrad\Lib\site-packages\openai_base_client.py", line 1061, in _request
raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "Unknown parameter: 'tool_resources.file_search.file_ids'.", 'type': 'invalid_request_error', 'param': 'tool_resources.file_search.file_ids', 'code': 'unknown_parameter'}}

Process finished with exit code 1

To Reproduce

1.  file = client.files.create(
        file=open(file_path, "rb"),
        purpose='assistants'
    )
    
 2. response = client.beta.assistants.create(
      model="gpt-4o-mini",
      name="QA Assistant",
      description="An assistant to answer questions based on uploaded documents",
      instructions="You are a helpful assistant who answers questions based on the information provided. Be concise and clear.",
      tools=[{"type": "file_search"}],
      tool_resources={"file_search": {"file_ids": [file.id]}},
      # file_ids=[file.id],
      temperature=0.0,
      top_p=1
  )

Code snippets

No response

OS

Windows

Python version

python v3.11.11

Library version

openai v1.58.1

[ltanase77]

I think this piece in the documentation says that you need to pass a vectorstore id.
"Assistants have tools and tool_resources instead of file_ids. The retrieval tool is now the file_search tool. The tool_resource for the file_search tool is a vector_store."

[zhoumengbo]

I think this piece in the documentation says that you need to pass a vectorstore id. "Assistants have tools and tool_resources instead of file_ids. The retrieval tool is now the file_search tool. The tool_resource for the file_search tool is a vector_store."

Thanks for your reply, I found some examples in https://platform.openai.com/docs/assistants/tools/file-search?context=streaming#ensure-readiness-before-creating-runs, which may be useful.