I have an assistant, and I have a button where users can press to add a file to speak to the assistant, then the Assistant will have the current Knowledgebase
AS well as the user’s file
how can I built it, I saw that there is the files API
You would need to incorporate the upload process using your own code by adding the file to the message.
-
Upload the file and retrieve the returned ID
https://platform.openai.com/docs/api-reference/files/create -
Add the file via ID to the message
https://platform.openai.com/docs/api-reference/messages/createMessage
Fair warning. There is currently no documented way to view all your current threads and you will be charged $0.20/GB/Day for every file that’s currently being used in a thread. This can leave you very susceptible to script-kiddies blowing up your account.
So you’ll need to be very careful in how you accept these files and basically create your own conversation & file management (Repeating yourself, having 2 sources of truth & rendering a chunk of the Assistants framework useless)
wow thanks for the reply
I am trying to have the users upload their file from another app and then use
Make.com to upload the file to GPT using a files API
then after 1 hour, I make a request to delete the file the user uploaded * this is how I plan on going about the costs
1 Like