ChatGPT, the chatbot that was opened to everyone in November 2022, has started to be used in many sectors, even in professional jobs, with what it has been able to do so far. Likewise, independent and professional developers have integrated the chatbot into their services and continue to do so.
But until now, the developers were not able to access the highest capability of ChatGPT. OpenAI only offered access to the previous generation language model “text-davinci-003”. But as of March, this has changed.
Anyone can access today’s ChatGPT API
In the announcement it shared, OpenAI announced that they have made the API of the GPT 3.5 language model, which is the current version of ChatGPT, available. Developers will now be able to access a more advanced language model and thus more advanced capabilities using the “gpt-3.5-turbo” model.
Not only ChatGPT can be used, but also Whisper, which converts audio to text.
Alongside ChatGPT, OpenAI has also released the API for Whisper, which can translate English audio to text with high accuracy. With the “whisper-1” model, the developers had the opportunity to integrate Whisper into the systems they developed.
So what will this do for us? Here is the answer to this question so far:
Although the use of the GPT 3.5 model and Whisper has only just been made public, it has been available to big names recently. For example, Snapchat started to host the chatbot called ‘My AI’ with the GPT-3.5 API.
Whisper, on the other hand, was used by Speak and appeared in an artificial intelligence-based application that facilitates learning English in South Korea.
The combined use of both Whisper and GPT 3.5 has great potential. Users will now be able to get answers to their requests from the GPT 3.5 model just by speaking out loud. Soon we will see examples where many developers use this duo.
Making use of APIs is of course not free:
Those who want to take advantage of OpenAI’s API will pay a fee for each word they send to the AI. But this fee is actually much lower than in the early days of ChatGPT.
Every transaction to be made with the gpt-3.5-turbo model will be priced on OpenAI’s token system. The cost of the transaction of 1000 tokens will be $ 0.002. A token is equivalent to 4 characters of an English word.
For example, if ChatGPT were to write this news text, a total of 1203 tokens would be spent (2,379 characters). This is almost equivalent to only 4 cents (excluding this paragraph and the rest).
To calculate the cost, you can access OpenAI’s own tool from this link.
42287
