OpenAI has announced that it would now allow third-party developers to incorporate ChatGPT into their applications and services via an API at a substantially lower cost than utilizing its existing language models. Whisper, the company’s AI-powered speech-to-text model, is now available through API, and the developer terms of service have been updated significantly.
OpenAI claims that its ChatGPT API is adaptable for more than just building an AI-powered chat interface, but it does mention some firms that have done so, including Snap’s My AI feature, which was launched earlier this week. The company claims its new gpt-3.5-turbo model family is the “best model for many non-chat use cases.”
Businesses that use ChatGPT for their activities will be charged $0.002 per 1000 tokens (about 750 words). OpenAI will provide organizations with a tool for calculating how many tokens a user inquiry will cost and translating this to meet the scope of their customer queries.
OpenAI claims that its pricing is ten times less than existing GPT 3.5 models, which previously powered ChatGPT. “ChatGPT API users can expect continuous model improvements and the option to choose dedicated capacity for deeper control over the models,” OpenAI said.
It’s worth mentioning that the model may not be the same as the one used by Bing, which Microsoft has described as a “new, next-generation OpenAI big language model” that’s “even faster, more accurate, and more capable” than ChatGPT and GPT-3.5. But, considering the amount of money the corporation has put into OpenAI, it’s not surprising that it has access to technology that the typical developer does not. Microsoft is also incorporating some of its own technology into Bing.
According to the company, developers will be able to obtain a dedicated version of ChatGPT if they are processing a massive volume of data via the API. In keeping with the post, doing so will give you greater control over the model you’re using, how long it takes to reply to requests, and how long discussions with the bot can last.
OpenAI has also introduced a new API for Whisper, their speech-to-text model. According to the firm, it can transcribe or interpret audio for $0.006 per minute. Theoretically, the Whisper model is open-source, so you can run it on your own hardware for free.
Whisper has received widespread acclaim from the developer community, but it may also be “difficult to run,” according to OpenAI. “The company has now made the large-v2 model available through our API, which provides convenient on-demand access priced at $0.006 / minute (of audio input),” the company claimed.