The News
OpenAI has announced the latest version of the AI model powering ChatGPT, revealing the AI can now handle real-time voice and video chats with users.
Called GPT-4o, it is a more advanced version of the existing GPT-4 model, and OpenAI said it would roll out to users over the next few weeks. The most significant update is the ability for users to talk in real-time with ChatGPT — it can respond in 320 milliseconds on average, OpenAI said, which makes for more natural interactions.
Users can also show it live video through the ChatGPT app, or via a screen-sharing function using the desktop app.
OpenAI also upgraded the model’s ability to communicate in non-English languages; during a live demo, the chatbot acted as a real-time English-to-Italian translator.
The new version will be available to all ChatGPT users, but paid subscribers will have higher message limits.
OpenAI is at the fore of AI development, but the new model signals the company’s push to stay on top. Voice recognition and response is an especially competitive sector: OpenAI’s competitor, Google, is holding an event Tuesday in which it could announce a new virtual voice assistant, while Apple is expected to debut a more conversational, AI-powered version of Siri in June.