Azure OpenAI Assistants API, file upload with "vision" purpose and other issues

API
, , ,
1

Hey, I’m wondering has anyone successfully uploaded an image file using client.files.create and setting the purpose to “vision”?

I get a purpose is not supported error: ‘Invalid value for purpose.’.

I have also tried to find relevant documentation but there is nothing that specific in the Azure openai documentation.

They made some changes to the file handling 5 days ago but Im not sure are they related to this.

I also get an type error when I tried to upload an image to the gpt4o assistant in the file content. Only type : text is supported.

I analyze the python client sourcecode and it should support both of these functionalities. Also openai official documentation supports this but not azure openai.

1 Like
2

Hi there -

you might have more luck getting an answer via the Microsoft Community:

3

The Vision purpose is currently not supported for the Assistant API. no response or timeline from azure on when this will be supported as far as i know.

Google ; “Azure OpenAI Assistants - Can’t upload file with vision purpose” and youll find a response from azure regarding the same (i couldnt post links)

4

Hi.
Did you get an answer to this? I am also getting the Only type : text is supported error.
Thanks.

5

I just used the purpose “assistants”. As it seems the purpose “vision” is not indeed supported yet in the Azure Openai assistants API. Make a tool for the assistants that actually analyzes the image with gpt4o chat completion api.

6

are you using the library or the rest api?