GPT-3 Pricing and integration in Java

Hi everyone,

I came here to fetch a piece of advice about the OpenAI API, more specifically about the GPT-3 model. I got this university project where I need to integrate a GPT-3 model into an application build on Java.

My first problem is the pricing of the model. The given prices on the website are prices/1000 tokens and since I never used it, I would like to know how much it will cost for a frequent usage during the entirety of the project (two months).

This brings me to my second question, what are effective differences between the 4 presented models (Ada, Babbage, Curie and Davinci) in terms of performances and complexity.

I’m also wondering myself if the OpenAI java’s API can work with Java 8.

Hope you can help and thank you all.

1 Like

Welcome to the forum!

  1. I would try to guess how many tokens you would use over the course of a project. You could test with Curie or even Ada for lower costs during development.

  2. Davinci is the best with the rest not as large but faster. Some simple tasks you might find that Curie or Curie fine-tuned works better than Davinci. What are you trying to do, exactly?

  3. No idea about API and Java 8, but if it can cURL to the API, you should be fine.

Hope this helps!

1 Like

Hi Paul,
Thank you for your response!

To give a context to my question, my project is to integrate a conversational model into a Java application. My university suggested using the GPT-3 model. For this, I need to give them an approximate price of how much the development and the use will cost.

Thanks again!