OpenAI ChatGPT-4o: All-around AI model surpasses GPT-4 and provides realistic interactive experience

🚀【AIWorld Revolution] OpenAI’s new favorite GPT-4o was unveiled, and its voice and video processing capabilities shocked the audience! 🌟

GPT-4o, a brand-new AI model, not only has the wisdom of GPT-4, but also can perform real-time interpretation, allowing people of different languages ​​to communicate smoothly. Its emergence will completely change the way we interact with AI. Come and experience the charm of future technology brought by GPT-4o! 🌐

OpenAI releases new GPT-4o model

The day before Google I/O kicked off, OpenAI unceremoniously stole the limelight and took the lead in releasing a new generation model-GPT-4o. This new model not only inherits the wisdom of GPT-4, but also has more powerful voice and video processing capabilities, giving users the feeling of almost interacting with a real person.

The special feature of GPT-4o can be seen from the name. The “o” here stands for “omni,” which means “omnipotent,” indicating the new model’s all-round capabilities in text, audio, and video reasoning. "We are proud to introduce GPT-4o, our new flagship model capable of processing audio, video and text in real time," OpenAI said in a statement.

GPT-4o's response ability is close to that of humans, "like the AI ​​in the movie"

Although GPT-4 can also recognize images and perform text and speech conversion, these functions have been scattered in different models in the past, resulting in long response times. GPT-4o integrates these functions into one model, which is called the "all-in-one model". Compared with the previous generation flagship GPT-4 Turbo, GPT-4o performs similarly in English and programming languages, but has significantly improved performance in other languages, faster APIs and up to 50% lower cost.

OpenAI指出，GPT-4o的回应时间接近人类，能提供更自然的沟通体验，最快可在232毫秒（0.232秒）、平均320毫秒（0.32秒）内响应问题。作为对比，GPT-3.5和GPT-4在语音模式下的回应时间分别为2.8秒和5.4秒。

In OpenAI's demonstration, GPT-4o was able to interpret in real time, allowing two people in different languages ​​to communicate without barriers. Or when you ask GPT-4o to tell a bedtime story, it can tell it vividly with a fuller and more emotional voice; or it can use a near-human tone to teach you how to solve simple mathematical problems.

According to OpenAI, GPT-4o can "read" the user's expressions and tone, know when and how to respond, and can quickly switch between different tones, from a cold mechanical sound to a cheerful song. . OpenAI's technical director Mira Murati said that the development of GPT-4o was inspired by the human conversation process, "When you stop talking, it's my turn to speak. I can read your tone and Response. It’s just so natural, rich and interactive.”

OpenAI CEO Sam Altman said in a blog, "The new voice and video modes are the best computer interfaces I have ever used, just like the AI ​​in the movie. I can't even believe it. Really, it turns out how dramatic the changes in response times and expressiveness are to reaching human levels."

Although not everything was perfect during the demonstration, GPT-4o sometimes interrupted others during the demonstration and even commented on the host's clothing without being asked. However, it quickly returned to normal after the presenter corrected it.

Mulati revealed that through the power of the all-round model, GPT technology will be further improved in the future. For example, it will explain the competition rules to users after watching the broadcast of sports events, and it will no longer be limited to simple tasks such as translating pictures and text.

OpenAI said users can now use GPT-4o in the free version, while paying subscribers will enjoy five times the message limit of the free version. The GPT-4o-based voice service is expected to be available to subscribers in beta next month. The free provision of GPT-4o also reflects OpenAI's achievements in reducing costs.

However, due to concerns about abuse, the voice function will not be available to all API users for the time being, and will first be available to some trusted partners in the next few weeks.

ChatGPTPC version of the program is now available

While GPT-4o has greatly enhanced its voice and video functions, OpenAI also announced an update to the web version of ChatGPT UI, claiming to have a more conversational main interface and message presentation. Mulati emphasized that although the models are becoming increasingly complex, she hopes that the interactive experience between users and AI will be simpler, clearer, easier and more natural, so that users do not need to worry about the UI, but focus on collaboration with ChatGPT.

OpenAI also announced a computer version of the ChatGPT program. The MacOS version is expected to be launched first, and the Windows version will be launched later this year. It is worth noting that there were earlier rumors that the negotiation between OpenAI and Apple on AI technology cooperation has come to an end. At this time, the Mac version of the program was first launched, triggering various associations from the outside world.

If you register OpenAI in mainland China, a prompt will appear:OpenAI's services are not available in your country."▼

Because advanced features require users to upgrade to ChatGPT Plus to use,In countries that do not support OpenAI, it is quite difficult to open ChatGPT Plus, and you need to deal with complicated issues such as foreign virtual credit cards...

Here we would like to introduce you to an extremely affordable website that provides ChatGPT Plus shared room accounts.

Please click the link address below to register for Galaxy Video Bureau▼

Click the link below to view the Galaxy Video Bureau registration guide in detail ▼

🔗Where is the registration address of Galaxy Video Bureau? How to log in and purchase ChatGPT Plus after registration?

Tips:

• IP addresses in Russia, China, Hong Kong, and Macau cannot register for an OpenAI account. It is recommended to register with another IP address.