Inspiring:
In the early morning of May 14, 2024, OpenAI held its first "Spring New Product Launch Conference", where they launched a new flagship product - the GPT-4o generative model and desktop app, and demonstrated some new features.
The launch was unusual because OpenAI’s technology changes the form of products we are familiar with and sends an important message to global technology companies.
While other tech companies are still playing catch-up with large models and multimodal features, trying to integrate features like summarizing text and editing images into mobile apps, OpenAI has directly set off a huge wave of technological revolution. The products they released feel like they came out of a sci-fi movie, and even Sam Altman was amazed by it!
Enter the text: 👇
Mira Murati, technical director of OpenAI, expressed three important things:
●First, she announced that OpenAI’s future products will prioritize free use so that more people can benefit.
●Second, to achieve this goal, OpenAI released desktop versions of the program and an updated user interface (UI) to make it easier for people to use these tools and make them feel more natural.
●Finally, after Mira announced GPT-4, OpenAI launched a new version of the big model, called GPT-4o. What’s special about this new model is that it interacts with users in an extremely natural way, bringing intelligence comparable to GPT-4 to everyone, even free users.
After this update, ChatGPT's large model can receive any combination of text, audio, and images as input, and generate any combination of text, audio, and image output in real time - this is the future of interaction.
Four highlights of GPT-4o:
Highlight 1: Simulated dialogue + mathematical solution
GPT-4o's response speed is quite fast, taking only 320 milliseconds on average, and even 232 milliseconds at the fastest to answer your questions, which is about the same speed as human conversation. This means that when talking to GPT-4o, you don't need to wait too long to get an instant answer. Moreover, after GPT-4o's response speed becomes faster, its interaction with you becomes more natural, a bit like chatting with a real person! It can even imitate the robot's tone, tell you bedtime stories, and sing songs!
Not only that, GPT-4o also improves visualization technology, through the camera, it can help you solve math problems step by step! This means learning math becomes more interesting and easier to understand.
Highlight 2: A new era of intelligent communication
This time, GPT-4o's voice technology has been improved not only in terms of reaction speed and emotional perception, but also in terms of instant language translation. At the press conference, they talked in Italian and English respectively. Then GPT-4o was able to quickly translate their conversation into the other party's language, demonstrating its powerful instant translation capabilities.
Highlight 3: A new era of intelligent perception of user emotions
In the past, models like GPT-3.5 and GPT-4 were made up of three separate parts that were responsible for converting audio to text, processing text, and then converting text back to audio. This process was a bit cumbersome, and the design prevented GPT from accurately capturing details such as tone and emotion.
However, OpenAI has made special adjustments to the GPT-4o model. Now, both the input and output are processed by the same neural network, which can better perceive the user's emotions. Even if there is an interruption in the conversation, GPT-4o can quickly recover and continue the conversation.
Highlight 4: Free for all users
The GPT-4o model will be free for all users, while the capacity limit for paid users will be increased to five times the original. The upcoming GPT-4o alpha version will include new voice models for ChatGPT Plus users. The "GPT Store", which was previously only available to ChatGPT Plus users, is now also free for all users. In addition, Mac users can now use the desktop version of the ChatGPT App, and a version for Windows users will also be launched this year.
These updates mean that more people can enjoy the power of GPT-4o, and the free and open "GPT Store" also provides users with more choices.
On what’s coming up at OpenAI:
In addition to introducing new models, OpenAI also revealed their business strategy at the press conference.
First, they said the query cost of the new model is 50% lower than GPT-4, which means they can afford to make the new model available to all users. This shows that OpenAI is more focused on attracting a large number of users rather than getting more people to pay for subscriptions.
The large number of users is good news for the GPT store, which has been online for several months. Like the iOS store, OpenAI hopes that the creator community and developers will work together to promote the development of the GPT store.
For example, some professors may use GPT to design customized content for classes, or broadcasters may create exclusive GPT content for listeners. Of course, some technologists also believe that many practical tools powered by GPT-4o will appear on smartphones in the future.
Apple is currently looking for third-party partners for this, and its negotiation partners include Google and OpenAI.
What does OpenAI need most? App permissions on the client side and system-level permissions, which only Apple can provide.
What does Apple need most? The best AI technology and the most suitable large language model. GPT-4o is undoubtedly the best candidate.
Moreover, Apple has unique advantages in self-developed chips and closed ecosystem. As I write this, I am even beginning to look forward to Apple’s upcoming WWDC conference on June 10th.
end:
In the field of artificial intelligence, every technological advancement represents a deeper integration of human intelligence and machine intelligence. As OpenAI's latest flagship model, GPT-4o has made great progress in technology. It has added audio comprehension capabilities and pays more attention to user experience, making interactions more natural and convenient.
Although OpenAI still faces many opportunities and challenges, the company has maintained a firm belief and continued innovation. In the future field of artificial intelligence, OpenAI will continue to play a leading role and make greater contributions to the progress of human society.