On Monday, OpenAI made headlines when they unveiled GPT-4o, or GPT-4 Omni, their newest flagship model. This new model, which boasts faster and more intelligent real-time voice interactions, promises improved capabilities for all users.


Why does this matter? Tech behemoths like Apple, Microsoft, and Google are all making a shift to generative AI in the future. OpenAI is leading the way and is committed to staying at the forefront of this ever changing industry.

The unique quality of GPT-4o, in CTO Mira Murati's opinion, is that it makes the powerful intelligence of GPT-4 available to all users, even those on the free tier. Murati stressed during a livestream presentation that GPT-4o—represented by the letter "o" for "Omni"—represents a major advancement in terms of speed and user-friendliness.

GPT-4o, the newest model from OpenAI, is pushing the envelope by smoothly producing output in all three formats—text, audio, and images—while receiving input in any combination of these forms. Not only that, but this sophisticated AI can also recognise emotions, so it can interpret the meaning behind your input. It may also be interrupted in the middle of a sentence, and it responds to commands very quickly—nearly keeping up with human speech. GPT-4o is transforming the way we connect with AI with its diverse set of skills and human-like reactivity.

Introducing GPT-4o, our newest flagship model that has real-time text, visual, and audio reasoning capabilities.

API and ChatGPT are now accepting text and image input; voice and video will follow in the upcoming weeks.

OpenAI, May 13, 2024 (@OpenAI)
However, the fun doesn't end there. Murati promised more developments and alluded to an even more significant update to the model—think GPT-5—that would be revealed later this year.

OpenAI demonstrated ChatGPT's real-time voice assistant in a demonstration, emphasising quicker answers and the ability to easily interrupt the AI. This look into AI-powered interactions in the future emphasises how revolutionary GPT-4o has the potential to be in the way we interact with technology.

In one presentation, OpenAI provided a live lesson on deep breathing to show how GPT-4o may be used to provide useful guidance.

Another example showed off ChatGPT's adaptability as it read an AI-generated novel in a variety of voices, including theatrical recitals, robotic tones, and even singing.

In a third example, ChatGPT's ability to solve problems was demonstrated by helping a user solve an algebraic equation rather than just giving them the solution.

Compared to earlier versions, GPT-4o demonstrated noticeably improved conversational and personality skills during the demos.

Additionally, OpenAI showcased the chatbot's smooth language switching, making real-time translations between English and Italian possible.

These examples highlighted ChatGPT's multimodal features, which include text, voice, and visual interactions. The AI assistant could read written messages and even try to read people's emotions by using the phone's camera.

In a larger sense, the online gathering fell on the same day as Google's next I/O developer conference, which is anticipated to feature a major focus on generative AI developments.

Additionally, OpenAI has announced the launch of ChatGPT's desktop version, which is primarily intended for Mac users and is now available to premium users. Given that Mac users had a larger user base than Windows users, OpenAI decided to give Mac users priority over Windows users in the beginning.

Furthermore, OpenAI disclosed its intentions to provide free user access to its GPT store and customised GPTs. The upcoming weeks will see the gradual implementation of these features.

GPT-4o's text and picture features are now being rolled out to ChatGPT Plus and Team subscribers who have paid for them; Enterprise customers will soon be able to access them as well. Access will also be available to free users, though at a slower pace.

Moreover, GPT-4o's audio version is scheduled to launch "in the coming weeks," extending its functionality beyond text-based exchanges.

The text and vision modes of GPT-4o will be used by developers, while audio and video features will soon be made available to "a small group of trusted partners."

OpenAI revealed the name of the mysterious GPT2-chatbot, which was discovered on a benchmarking website, in a tweet.