OpenAI continues to capitalize on the success of ChatGPT. App developers can now integrate the underlying language model into their own platforms for a fee.

Given the popularity of the tool, it seemed only a matter of time before OpenAI would sooner or later build a lucrative revenue model around ChatGPT. A first step was the sale of “ChatGPT Plus” subscriptions, and starting today developers can also get started with a paid API to integrate the advanced AI functionalities into their platform.

The company reportedly always planned to market ChatGPT as an API, CEO Greg Brockman told TechCrunch. However, OpenAI wanted to make sure the quality was there before rolling out the tool.

ChatGPT Turbo

The API provides concrete access to “GPT 3.5 Turbo”, an improved version of the language model on which ChatGPT is built. This version is even faster and more responsive than the Plus version. The AI ​​model lends itself to applications other than chatting with users; For example, e-commerce companies can offer more personalized shopping experiences with GPT 3.5 turbo.

A major improvement in the Turbo version is that it should be less prone to “hallucinations”. From the first experiments with ChatGPT and Bing Chat, the AI ​​sometimes overstepped the mark and gave answers that ranged from completely off-topic to offensive. It turns out that fooling a language model with your text prompts is relatively easy. means fast engineering that should be corrected.

The cost of the API is $0.002 per 1000 “tokens,” which is 750 words, according to TechCrunch. The revenue should offset the high cloud computing costs involved in running a complex model like ChatGPT. OpenAI may be a profit-sharing company, but it can’t rely on investments to stay healthy.