Okay, let's talk about it for you. Remember when? It's 1 a.m. on May 14, 2024, no, OpenAI woke up a bunch of us night owls, saying that there was something big to announce. It's been a long time since it last surprised the market - the release of the "Wensheng video model" Sora. This time, they showed their faces at the spring conference, and no, OpenAI's boss Mira Murati stepped forward and showed us a wave of ChatGPT-related updates. The highlight of this time is "GPT-4o"! That's right, it's the big multimodal model that's faster and more affordable than its older brother GPT-4 Turbo.
The first is the debut of GPT-4o, which is claimed to be faster and more cost-effective than Big Brother. The second thing, the operation is as fierce as a tiger, it turns out to tell us that ordinary free users who use ChatGPT can also enjoy the treatment of GPT-4o, and those who can only use GPT-3.5 in the past are nothing!
In addition, Mira Murati also said that in the future, our Plus users will be able to experience an improved voice experience based on GPT-4o, which means that the current GPT-4o does not include voice functions. Apple computer users are even more happy, OpenAI has made them an exclusive ChatGPT desktop application, hey, Windows users have to wait a little longer.
Not to mention, at this press conference, those functional demonstrations are simply amazing! From real-time interpretation to detecting people's emotions, and the bridge where GPT acts as a math teacher, it really makes everyone look stupid. Especially the one who could detect Mark Chen's nervousness when he exhaled excessively, and teased "Mark, you're not a vacuum cleaner", I almost didn't laugh out loud. GPT-4o can follow you through math problems and read emotions — a little bit harder, but it's pretty sharp.
OpenAI is not bragging, GPT-4o was born, taking a big step towards a more natural human-computer interaction. It responds quickly to audio input, even about the same time as a human chat, and understands not only your text, but also your voice and expressions. Other models, stupidly still divide the work, this GPT-4o, one can top three, knows everything, it's really our all-round little expert!
However, although it is good and good, there are always those concerns and worries, such as the technology is freshly baked, the market has not been tested, or the problem of technology communication bias. Hey, how to say it, any new technology comes out, someone has to worry about it and pick this thorn. But just such a thing, bringing us so many possibilities, is still quite exciting.
Okay, leave a little suspense at the end: with so many new features and new experiences, what new changes can GPT-4o bring to us? How will OpenAI address those concerns and problems? Ouch, don't worry, all the answers will probably have to be revealed slowly in the future.