With the web GPT-4 version I can upload document and ask the same question (the same as by API with gpt-4-1106-preview), the document analyze takes up to 1-2 seconds, whole response takes up to 1 min (with my actions like open and upload file and copy question to thread). What should I do to reach the same speed with api? Why the api process takes up to 3-7 minutes?
The steps to reproduce the issue with api are:
- Upload file
- Create assistant with uploaded file id,
- Create thread
- Add Message and ask to analyze the docx file
- Create run
- Check run status
- Get answer
I’ve tried to make shorter instruction for API and attache file to message instead of assistant, but got the same result.
The help center respond me with the common answer on my question without any arguments why so big difference. (looks like bot answer)
We use the service to analyze customers’ documents, it doesn’t fit us with this speed. Can you fix the issue?