OpenAI introduces ChatGPT and Whisper APIs for developers
OpenAI made developer APIs available for its popular ChatGPT and Whisper AI models, allowing app developers to extend the capabilities of their apps with OpenAI technology.
An API (Application Programming Interface) is a set of protocols that allow different computer programs to communicate with each other. The usage fee for integrating the models into their app is based on usage.
OpenAI launches API for Whisper
OpenAI on Wednesday released an API for Whisper, the AI research company to open-source the speech-to-text model in September 2022. The API is accompanied by ChatGPT, which produces consistent text in multiple styles. Whisper can help you transcribe spoken audio into text.
API Updates for Developers
The company announced that developers now have access to state-of-the-art language and speech-to-text capabilities through the ChatGPT and Whisper models over APIs.
In addition, it shared that the cost has come down by 90% since December, so users can now use the open-source Whisper Large-V2 model in the API with much faster and cost-effective results. It provides additional features such as:
- continuous model improvement
- Choose dedicated capacity for deeper control
- Enhanced API Terms of Service for Developers
Beginner users of ChatGPT and Whisper API
crackle: The creators of Snapchat earlier this week unveiled My AI for Snapchat+, an experimental feature powered by the ChatGPT API. With this friendly, customizable chatbot, Snapchatters can get personalized recommendations and more.
Quizlet: Leveraging OpenAI’s GPT-3 to power multiple use cases such as vocabulary learning and practice testing. With the launch of the ChatGPT API, Quizlet is introducing Q-Chat: a fully adaptive AI tutor that interacts with students using relevant study material, providing a fun, chatty experience.
Instacart 2.0: Now with ChatGPT, Instacart is taking its app to the next level, introducing ChatGPT and Instacart’s own AI to provide what it calls “inspirational, shoppable answers” to customer questions about food. With access to over 75,000 retail partner store locations, customers can find whatever they want to satisfy their open-ended shopping goals.
Shopify Shopping: Used to power the shop’s new shopping assistant, powered by the ChatGPT API, which makes personalized recommendations based on shoppers’ search requests.
Speak: is the AI-powered language learning app that creates the smartest path to spoken fluency. It is the fastest growing English app in South Korea, and is already using the Whisper API to develop an AI speaking companion product.
Introducing the ChatGPT model family, GPT-3.5-Turbo, which is the only model used in the ChatGPT product. It costs $0.002 per 1k tokens, which is 10x cheaper than our GPT-3.5 model and our best model for many non-chat use cases. Early testers have seen successful migrations from the TEX-Davinci-003 to the GPT-3.5-Turbo with minimal adjustments to their signals.
The ChatGPT model consumes a sequence of messages and metadata, which are still presented to the model as a sequence of “tokens” using a new format called Chat Markup Language (“ChatML”).
For those who require deeper control over model versioning and system performance, we now provide dedicated instances. Our API runs on Azure, and developers pay for a reserved allocation of compute infrastructure over a period of time.
With a dedicated instance, they get full control over the instance’s load, the option to enable features such as longer context limits, and the ability to pin model snapshots. This can be economically beneficial for those moving more than ~450M tokens per day, as it can dramatically reduce costs compared to shared infrastructure.
Speech-to-text made easy Get fast on-demand access to the Large-V2 model of the Whisper API, which costs only $0.006/min. Enjoy optimized performance and support for various formats (m4a, mp3, mp4, mpeg, mpga, wav, webm).
OpenAI said it has been focusing on our developers over the past 6 months, gathering feedback and making concrete changes to provide a better service:
- API data submission number unless the organization opts in.
- Default 30-day data retention policy, with options for stricter retention.
- Pre-launch review done.
- Better developer documentation.
- Simplified Terms of Service and Use Policies, including data ownership terms.
Starting today, you can use the ChatGPT and Whisper APIs – for more information, see here.
Announcing the update, OpenAI shared this
We believe AI can provide incredible opportunities and economic empowerment to everyone, and the best way to achieve this is by allowing everyone to build with it. We hope that the changes we’ve announced today will lead to a myriad of applications that everyone can benefit from. Get started building next generation apps powered by ChatGPT and Whisper.