When creating an assistant through the API, and running the assistant in the playground, I encountered several problems

API
, ,
1

I want to create an assistant through the API, but the API documentation seems a bit difficult, so I create and run it visually in the playground. Currently, I mainly do retrieval of my own data.

When creating an assistant online and testing it in the playground, I upload a “file” when creating the assistant, but I am also allowed to upload a “file” when running the Thread in the test. Is there any difference?
Each question to the assistant seems to consume a lot of tokens?
Every question has to go through queued, in_progress, and completed. Will it be time-consuming?
How to determine the thread’s life cycle? How to reuse?

图片
图片1825×844 32.2 KB

Why didn’t I get any results back?

图片
图片1846×897 58 KB

1 Like
2

Yes, as it may take multiple steps and use the information you provide.

I hate to handout “homework,” but the OpenAI Cookbook page for Assistants API is a great run-down on everything.

3

OK, thank you very much for your guidance. Hope this solves all my problems.
Can you help provide a Java version?

1 Like
4

Looks like there’s one Java library on the official OpenAI docs pages…

5

OK.Thanks. Since the API was first announced, I developed it myself. The community version is usually slow to upgrade with new features, so I tend to develop it myself.