One day before Google’s I/O, OpenAI tried to steal the show from its big competitor. And their demos definitely caused quite a stir.
The focus was on their latest AI model GPT-4o, where the “o” stands for “omnimodel”. This is to indicate that this version does not only process text, but also e.g. image and sound information. GPT-4 was already able to do this to some extent, but here it seems to be integrated directly into the model.
GPT-4o is noticeably faster. And best of all, this latest model will soon be available on the free version of ChatGPT.
However, it was the extremely realistic and at the same time tipsy-sounding female voice of the voice assistant that caused the most excitement and reactions. Parallels to the movie “Her” were drawn quickly.
In the demos, the answers came at lightning speed and the AI performed a wide variety of tasks. But a demo never represents reality. So we’ll have to wait and see how well it actually works in the real world. But I found the progress impressive regardless.
You can find many examples on the official website.