ChatGPT API: Possibilities, Features, Industries

Recently, OpenAI announced that they opened their #API #ChatGPT and Whisper (text-to-speech) to the public. Developers now can integrate artificial intelligence into their apps. The first cases involved Snapchat, Quizlet, Instacart, and Shopify and the language-learning app Speak. Early users of ChatGPT and Whisper API mainly used the model as a chatbot and assistant. However, are there other use cases for app integration??

Examples of ChatGPT API application

Chatbots and Q&As will for sure be the most popular use cases in apps. Yet, OpenAI shows the categories where API could be used. In fact, they have a whole page dedicated to examples of possible applications.?

The most prominent use cases of API are:

• Text completion
• Code completion
• Chat completion?
• Image generation - generating or manipulating images with DALL·E models
• Fine-tuning - customizing GPT-3 model for your app
• Speech to text - turning audio into text

Based on experimental features offered by some #apps already, we can assume others. In Edtech, chat-like features could be used for testing in a more friendly manner or explaining more difficult concepts. In Fintech apps, we could use ChatGPT as an assistant or financial advisor (if it’s customized for the app). Also, it could be of great help to developers: to look for several solutions, look for bugs, or come up with simple code.?

How much does it cost?

The ChatGPT API costs $0.002 per 1000 tokens (that's about 750 words), making it ten times cheaper than the GPT-3.5 model subscriptions. Also, it was the initial aim of OpenAI team: to monetize the chat better. Now, many ChatGPT Plus users could go for API instead.?

Integrating ChatGPT with Programming languages

Two languages that are now frequently used to integrate with ChatGPT API are Python and Node.js. To be honest, setting up the APIs is not that different from any usual API integration. You get a unique API key, and copy the code snippet from the official documentation that will call ChatGPT API.?

To make a call to API for chat completion, you’ll need to use using OpenAI Python v0.27.0 for the code below to work. Messages must be an array of message objects, where each object has a role (either “system”, “user”, or “assistant”) and content (the content of the message). The “System” role goes before the actual messaging starts. That’s for you to define how the ChatGPT should respond. The “user” is basically human and the “assistant” is the ChatGPT.?

An example API call looks like:

And the response example looks like this:

The only drawback I found is that there’s a token limit and the messages won’t be as long. This parameter needs adjusting.?

Integration into Python or Node.js applications isn’t truly complicated. Besides, there are already good articles and guides on the setup with code snippets and everything:?

1. A Beginner's Guide to to Integrating ChatGPT with Node.js?
2. A Simple Guide to The (New) ChatGPT API with Python

If you follow the steps, you shouldn’t face any issues.?

Short Afterword

ChatGPT and Whisper API could become a more cost-effective way of using AI. Also, almost any industry could use the GPT-s model to add a new feature to their apps. The features like chatbots, assistants, translators, image creations, etc. I would say that technology is a bit raw now but with the right knowledge of machine learning and code, you could get it to work for you.