What you need to know
- OpenAI announced GPT-4o at its live stream today.
- GPT-4o is faster than GPT-4 and improves upon its predecessor in several areas.
- GPT-4o promises "GPT-4-level intelligence" while being available to free users.
- Paid users will continue to have five times the capacity of free users.
OpenAI had three pieces of news to share at its spring event. The company started its event by unveiling GPT-4o, the latest flagship model from OpenAI. GPT-4o delivers "GPT-4-level intelligence" to everyone, including free users. It also improves on its predecessor in several areas, including being faster.
GPT-4o is half the price as GPT-4 turbo but is twice as fast and has five times the rate limit, as highlighted by OpenAI CEO Sam Altman.
The live stream was relatively short, only lasting around 25 minutes. During that time, OpenAI showcased real-time voice communication, video capabilities, and ChatGPT being able to detect and replicate emotion.
OpenAI will roll out the new features unveiled for ChatGPT over the coming weeks. The new voice mode in ChatGPT will first roll out to ChatGPT Plus users.
Real-time audio capabilities
OpenAI also demonstrated its real-time audio capabilities. The demo showed ChatGPT giving feedback on how to give a live presentation. The tool seemed conversational and to work without much delay, though the demo was short. One thing that differentiates the latest ChatGPT version is that you can interrupt it. The tool can also sense your emotion to some extent, such as identifying someone's breath indicating stress (when asked).
A second demo highlighted how OpenAI can add emotion to its speech by sharing an over-the-top bedtime story on request.
Later during the presentation, OpenAI translated back-and-forth between Italian and English in real-time.
Visual capabilities
You can also interact with ChatGPT using video. A demo during the event shows a user filming himself working through a math equation and ChatGPT walking him through the problem. The tool also correctly determined a doodle that said "I 💗 ChatGPT."
The presenters of the live stream had ChatGPT look at a person's face and identify his emotions. ChatGPT said the presenter had a "big smile and a touch of excitement." ChatGPT then said the presenter was making it blush when told the reason for the happiness was presenting ChatGPT.
ChatGPT in action
Altman live-tweeted the live stream on X. Here are some of his posts:
and with video mode!! pic.twitter.com/cpjKokEGVdMay 13, 2024
especially at coding pic.twitter.com/EYC3dMCmNpMay 13, 2024
audience request to act as a translator pic.twitter.com/E2qbfhyVmXMay 13, 2024