May 14, 2024 - Technology

OpenAI debuts new, voice-focused model

Illustration of two pixelated speech bubbles containing binary code clashing against each other.

Illustration: Gabriella Turrisi/Axios

OpenAI yesterday unveiled a new flagship AI model with more powerful capabilities — including smarter, faster real-time voice interactions.

Why it matters: OpenAI, whose ChatGPT kicked off the generative AI race, is now trying to hold its lead with rivals like Google in hot pursuit.

The big picture: OpenAI CTO Mira Murati said the new GPT-4o — that's a letter "o" for "Omni" — is a "huge step forward" for ease of use and speed.

Zoom in: In one demo, OpenAI showed one of its workers getting a real-time deep breathing tutorial.

  • Another showed ChatGPT reading an AI-generated story in different voices, including super-dramatic recital, robotic tones and even singing.
  • In a third, a user asked ChatGPT to look at an algebra equation and help the person solve it, rather than simply providing an answer.

OpenAI also showed the new chatbot working simultaneously across languages — helping translate between English and Italian, for instance.

  • In all the demos, GPT-4o showed considerably greater personality and conversational skills than previous incarnations.

What's next: A bigger update to the underlying model is due to be unveiled later this year, Murati tells Axios.

🖖🏻 Alex's thought bubble: Universal translator, here we come.

Read the rest.

avatar

Subscribe for more Axios What's Next in your inbox.

Read the full edition