OpenAI’s ChatGPT Launches ‘GPT-4o,’ Desktop App

This article originally appeared on InformationWeek.

OpenAI on Monday announced major improvements to its flagship chatbot, ChatGPT, including improvements to voice, text and vision capabilities that would make “faster” services available to all users, and a new desktop app.

Mira Murati, OpenAI’s chief technology officer and (very briefly) former CEO, during a video briefing said the company’s latest large language model, GPT-4o, would let developers and other users utilize real-time conversational speech, text, video, and audio. While use limits will still separate the company’s enterprise offerings, all users will have access to the new model’s capabilities, Murati said.

“This is the first time that we are really making a huge step forward when it comes to ease of use,” Murati said.

In a blog post, CEO Sam Altman said the new model advances the company’s mission to provide advanced AI tools to everyone. “We are a business and will find plenty of things to charge for,” he wrote. “ … and that will help us provide free, outstanding AI service to (hopefully) billions of people.”

The announcement could have been a bit of a letdown to those excited by recent reports of a potential deal with Apple and a rumored search engine feature to rival Google. Altman shot down the prospect of those developments being part of today’s announcement in a post on X. Still, the new features offer the biggest product announcement from the Microsoft-backed GenAI firm since its ChatGPT Enterprise announcement last year.

Related:Microsoft’s Hedge Against OpenAI Makes Perfect Sense

“ … the new voice (and video) mode is the best compute interface I’ve ever used,” Altman wrote. “It feels like AI from the movies; and it’s still a bit surprising to me that it’s real.”

Murati said the new model will offer improved speed in up to 50 different languages and will be offered via OpenAI’s API to developers starting today. “So, developers can start building two times faster, 50% cheaper with five times higher rate limits,” Murati said.

OpenAI team members also sat down to demonstrate the new model’s capabilities, showing off the model’s ability to judge the users emotional state, and the ability to interrupt to clarify questions. OpenAI researcher Mark Chen showed how the model could even be used to calm your nerves with detailed instructions on breathing -- even correcting Chen when he purposefully hyperventilated. "Whoa, slow down!” the ChatGPT voice said.

The model’s ability to help with code and simple math were also demonstrated. Interestingly, the coding demonstration was performed on a MacBook. Bloomberg had earlier cited sources saying the company was close to a deal with Apple that would bring ChatGPT capabilities to its next operating system.

Related:UAE Releases New Falcon AI Model To Challenge Meta, OpenAI

According to Bloomberg Intelligence, GenAI will be a $1.3 trillion market by 2032 as OpenAI, Meta, Google and its parent Alphabet, and Microsoft all jockey for position to create increasingly sophisticated large language models. According to Statista, AI’s global market will reach $184 billion in 2024.