ChatGPT API announced

OpenAI has stated that ChatGPT integration into third-party apps and services is now possible via an API, and that doing so will be substantially less expensive than utilizing its current language models. The business is updating its developer terms of service and making Whisper, its AI-powered speech-to-text technology, accessible through an API.

Even while it lists some businesses that have been utilizing it for that purpose, such as Snap's My AI feature, which was unveiled earlier this week, OpenAI claims that its ChatGPT API can be used for more than just building an AI-powered chat interface. According to the business, the gpt-3.5-turbo model family is the "optimal model for various non-chat use cases."

A model that Microsoft has described as "new, next-generation OpenAI big language model" that is "even faster, more accurate and more capable" than ChatGPT and GPT-3.5" is probably not the same one that Bing is utilizing. Yet, it's not surprising that the corporation has access to technology that is not available to the typical developer considering how much money it has put in OpenAI. For Bing, Microsoft also makes good use of its own technology.

1,000 tokens are available for $0.002, which, according to OpenAI, is "10 times less expensive than our current GPT-3.5 models," in part because of "a series of system-wide optimizations." Although 1,000 may seem excessive, it's important to keep in mind that providing just one text piece for the API to respond to could require numerous tokens. To forecast what text should be output next, the system divides sentences and words into blocks of text known as "tokens."

"ChatGPT is wonderful!" requires six tokens, according to OpenAI's specification; its API separates it into "Chat," "G," "PT," "is," "great," and "!". A typical rule of thumb is that "one token generally equates to 4 letters" in English, according to the business, which offers a tool for calculating how many tokens it will take to decode a string of text.

According to the business, developers that are processing enormous amounts of data through the API will also be able to obtain a dedicated instance of ChatGPT. It claims that by doing this, you will have more control over the model you use, how quickly you want it to respond to your requests, and how long discussions with the bot can last.

ChatGPT demonstrates that AI is now widely used, and things will only grow stranger from here on out.

Even while ChatGPT is most likely to draw notice, OpenAI has also unveiled a new API for Whisper, its speech-to-text technology. According to the business, you may use it to translate or transcribe audio for $0.006 per minute. The Whisper concept can technically be operated on your own hardware for free because it is open source software. However, OpenAI probably has access to more potent hardware, so using the API may be the best option if you need a quick turnaround or need to do transcription on low-powered devices like phones.

Additionally, OpenAI is introducing certain policy adjustments that it claims are a result of developer input. One significant one has said that it will no longer train its models on data provided over the API unless users specifically consent to that use.

To put it another way, it is switching from an opt-out to an opt-in system. This adjustment may allay some worries about incorporating confidential information into the bot, as some businesses have prohibited staff from utilizing the technology at all. Trade secrets shouldn't be entered if it's learning from user input because there's a chance it will spew that information back out to someone else.

The business also claims that it is aiming to increase uptime and that "stability of production use cases is now the engineering team's top focus."

Although many developers have created workarounds to include chat functionality into their applications, including using OpenAI's normal GPT API, which has been around for a while, the launch of an official ChatGPT API feels like it could be the tipping point. While many businesses are developing their own AI chatbot models, most developers cannot afford such sophisticated technology. They will now be limited to using OpenAI's technology.