If you’ve been following the news around OpenAI, you’ve probably seen reports that the company is working on a search product that was going to be announced this week. That didn’t happen, but instead it revealed something arguably more impressive — it launched its latest model, GPT-4o.
The “o” stands for “omni,” alluding to the fact that the model can handle speech, text and video. Speech is the big addition to GPT-4o and to the popular chatbot it powers, ChatGPT. Where ChatGPT previously supported voice interactions through text-to-speech, it’s now closer to a real human assistant, with capabilities like detecting the user’s emotion and responding with emotion of its own. It can even sing.
On top of that, GPT-4o also has improved vision, for example being able to analyze a photo to identify an item of clothing or a desktop screen to analyze software code.
In a variety of new video demos, GPT-4o users show off different capabilities, such as tutoring a student on math, complimenting a dad joke, providing guidance to a user whose vision is impaired and even chatting with other versions of itself, GPT-4o to GPT-4o. And unlike some other OpenAI announcements, where there’s extremely limited access to the new technology, GPT-4o is available in the free tier of ChatGPT today.
Hit play, then let us know if you’re excited!
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
Comment