In the early morning of May 14, OpenAI officially released a new generation of flagship generative model GPT-4o, desktop App, and demonstrated a series of new capabilities at the "Spring New Product Launch Conference".
Unlike in the past, external public opinion is no longer bowing to the whole group, but quickly divided into two camps.
Many people in the industry have the same conversation: "OpenAI is no longer far ahead. ”
01
No GPT-5
There is also no AI search engine
OpenAI's previous new product launch conference was basically two routines:
One is to leave suspense and amplify moves, such as GPT4; The second is a surprise attack, such as Sora.
This time, OpenAI revealed information one after another before the meeting, but the deep meaning of the early briefing was more focused on reducing user expectations.
On May 11, OpenAI officially announced that it will broadcast some demonstrations about ChatGPT live at 1:00 a.m. on May 14, with the theme of "Spring Upgrade", and GPT-5 will not be released on that day, nor will it release the AI search engine.
On the same day, Ultraman also said in a podcast that OpenAI will continue to improve and improve the quality of ChatGPT's voice functions, and expressed his belief that voice interaction is an important way to lead to the future of interaction.
In other words, the content of the "Spring New Product Launch" has long been known to the outside world, and there is no suspense or easter eggs.
This means that OpenAI has indeed slowed down.
As an industry insider said, "If GPT-5 is released, OpenAI is still far ahead." If it's AI Search or a voice assistant, it means that OpenAI is in decline. ”
02
GPT4+新壳=升级版Siri
What new features have been released in the "Spring New Product Launch"?
To put it bluntly, two new products have been released: GPT-4o and desktop App.
OpenAI的新模型名叫GPT-4o,字母o是"Omnimodel"的首字母,意思是全能模型。
What is an all-round model?
In fact, GPT-4o covers all modalities such as text, visual, and audio.
In terms of overall performance, GPT-4o is more like "GPT4 + new shell = upgraded Siri".
OpenAI CTO Back Murati 表示,GPT-4o 提供了「GPT-4 水准」的智能。
This means that GPT-4o does not bring a disruptive upgrade, at best, it adds a new shell.
Judging from the live demonstration function, users can have a real-time conversation with GPT-4o, just like chatting face-to-face with friends.
Such as reacting in 232 milliseconds; You can interrupt at any time to start a new topic; Feel the rhythm of the user's breathing; Explain the reasoning process of equations and diagrams; Identify the emotions of the people in the picture, etc.
These, if it is iPhone16 equipped with new functions after the self-developed large model, it is more in line with the expectations of users, after all, Apple has been so "patching" in recent years.
The so-called desktop App, more like the mobile phone WeChat finally has a computer version, which makes people look forward to it, but not much.
For OpenAI, its user expectations have long since risen to the point where they can't manage. This also made the press conference a little bleak.
To this end, OpenAI's PR team further explained the following:
"A key part of our mission is to make very powerful AI tools available to people for free (or at a discounted price). I'm very proud that we have the best model in the world for free in ChatGPT, with no ads or anything like that.
When we started OpenAI, our initial idea was that we wanted to create AI and use it to create various benefits for the world. Instead, now it looks like we're going to create artificial intelligence, and then other people are going to use it to create all sorts of amazing things that benefit us all. ”
03
OpenAI commercialization
It's always been a problem
"Some time ago, I chatted with the person in charge, and I felt that Ultraman was bragging, GPT 5 is definitely not so amazing, or it doesn't matter if GPT 5 is amazing or not. Because GPT 4 already meets most of the business needs, the upgrade has not changed so much for the business model itself. ”
On May 8, Zhu Xiaohu and Fu Sheng had such a conversation. In this regard, Fu Sheng believes that GPT 5 has not been released for so long, either because the performance has not improved much, or the cost is too high.
Behind this, we can see at least two problems: one is the cost problem; The second is the issue of business model.
As we all know, OpenAI's technological capability leadership has always been proportional to the speed at which it burns money. However, its commercial income has not increased significantly.
According to data released by Sensor Tower, a third-party market analysis platform, ChatGPT has been downloaded 7 million times in the App Store worldwide in the past month, and its subscription revenue has been $12 million. The global Google Play market has 90 million downloads and $3 million in subscription revenue.
Currently, ChatGPT Plus is available for $19.99 in both app stores. Extrapolating from the subscription data, ChatGPT Plus has 750,000 subscribers who paid through the app store in the past month.
Although ChatGPT Plus still has a large number of direct paying users, from the perspective of the revenue of the mobile phone, the annual income is only 200 million US dollars, and several times more will also cover its money-burning speed, and it is even more difficult to support OpenAI's valuation of nearly 100 billion.
From the perspective of the overall environment of China and the United States, many large model capabilities on the market are rapidly pulling together this year.
For startups that focus on large-scale model research and development, they must either be far ahead, for example, it is difficult for competing products to break through your technical barriers within a year;
Either make a single-point breakthrough and find AI application scenarios in advance. Otherwise, it will soon be surpassed by large manufacturers with a large number of application scenarios.
Previously, the industry had looked down on Google because of OpenAI's disruptive innovation. But in fact, Google's stock has been rising recently, and even its market value has returned to $2 trillion.
This means that with the integration of large model capabilities, whoever can overcome the application scenario first will be more likely to advance to the king of the AI era.
Therefore, Zhu Xiaohu has repeatedly mentioned recently, "At this point in time next year, AI applications will definitely explode." ”
Whether OpenAI can catch up with the big wave of AI applications is still difficult to say.
In addition, GPT-4o has just been announced to be free for everyone, which actually makes its commercialization even worse.
04
Some industry insiders believe that what GPT-4o may do next:
Challenge all GUIs (Graphical Interfaces) and those who want to work on LUI (Voice Interfaces).
If the UI interface of the taxi app becomes a sentence "give me a taxi", then it is no longer important for users to open Didi or AutoNavi.
The rest is nothing more than a more brutal price war between merchants.
This change may overturn the user volume advantage and brand voice accumulated by all apps on the market in the past, and everyone has to start from scratch.
To put it bluntly, in OpenAI's view, since the subscription income in the direction of To c is limited, it is better to adopt the strategy of "rural areas surrounding cities".
Use a large number of C-end users to force the B-end to make subversive reforms and create a more imaginative application market for themselves.
What are your thoughts on this?
[The author of this article is a dark horse, an entrepreneur who originally wrote.] If you need to reprint, please contact the WeChat public account (ID: iheima) for authorization. ］