After reaching over 100 million users in less than two months, OpenAI, the company behind the viral chatbot service, announced yesterday that it would release ChatGPT’s AI model to the public at a price ten times lower than its predecessor.
ChatGPT is built on OpenAI’s GPT 3.5 model, a refresh of OpenAI’s GPT 3 learning model. And the APIs (application programming interfaces) that OpenAI released today are based on the company’s GPT 3.5 Turbo model, which is identical to the current ChatGPT model.
The newer March 1st GPT 3.5 Turbo model is now priced at $0.002 per 1000 tokens, ten times less than OpenAI’s previous GPT-3.5 models. OpenAI also claims that GPT 3.5 Turbo is the company’s best model for many non-chat use cases, citing early clients who migrated from GPT 3 (text-DaVinci-003) to GPT 3.5 Turbo (gpt-3.5-turbo) with only minor changes to their prompts.
Whisper API accepts a variety of formats and is accessible via the organization’s transcriptions (transcribes in the source language) or translations (transcribes into English) endpoints (m4a, mp3, mp4, mpeg, mpga, wav, webm). OpenAI has set the price at $0.006 per minute.
In addition, the launch today revealed that OpenAI has already made its latest March 1st API available to early clients such as Snapchat with its new ChatGPT AI-powered chatbot.
To summarize the announcement, OpenAI hopes that the new AI models made available to developers today will lead to a plethora of applications that everyone can benefit from.