File Search Cost Optimization

Looking for some advice here.

I am uploading files using assistants api and file search tool to find relevant info in json documents. Each call is a new unique document so there is no need to maintain context for this use case. I noticed while testing with relatively small json files(think 200kb) using gpt4o model that the cost to scale this out to thousands of calls would be pretty high.

Is it necessary to use assistants api in this case? Does anyone have experience optimizing for cost with file search api calls? Any advice is appreciated.

1 Like