Competition among tech giants is becoming less and less about martial arts.

Originally, OpenAI's “Spring New Product Launch” was scheduled for May 9, but as a result, it has been delayed until now.

Why is that? Because Google has a developer conference tomorrow...

I just want to ride my face to output, and not give friends and merchants any way to survive!

So at 1 a.m. this morning, Sam Altman said, “Something like magic,” finally unraveled the veil.

It's not GPT-5 as you might imagine, but an iterative version of GPT-4, GPT-4o.

o is an abbreviation for omni, derived from the Latin word omnis, which means omnipresent, omniscient, and omnipotent.

It sounds divine; it's quite a bit like Buddha or God.

GPT-4O, or omnipotent model.

omnipotent? That's very interesting.

Where are the top 01?

Almighty certainly doesn't mean omniscience and omnipotence.

At least not right now.

If a man-made “God” were actually created, we could all lie flat, eat, and die right away. Dominating the world or liberating the world would all be in this guy's mind.

Currently, GPT-4o can achieve the full modality of any combination of text, image, audio, and video.

OpenAI's original statement is: GPT-4O is the first model to combine all modes, and we're only scratching the surface of its capabilities.

Although it is only superficial, it is also extremely powerful.

Let's look at the horizontal review first.

The short summary is: faster, stronger, cheaper.

The first is efficiency. GPT-4o's processing speed is twice that of the GPT-4 Turbo, the rate limit has been increased by five times, up to 10 million tokens/minute, and the price has been reduced by half.

Next is performance. GPT-4O is more powerful than GPT-4 in all language benchmarks and can be seamlessly translated between more than 50 languages.

And then the most important point: multi-modal input and output.

GPT-4o can process any combination of text, audio, and image inputs, and generate corresponding outputs in real time to interact with users.

Note, it's real time! Real time!

Let's take a look at its specific performance.

At the press conference, the tester said to the phone:This is my first time to live stream and I'm a little nervous.

ChatGPT answers immediately：You can take a deep breath.

The man did the same.

ChatGPT immediately joked again:You're not a vacuum cleaner, so don't breathe.

When they hear that the other person is finally breathing smoothly, it actually encourages them.

Seeing this, Apple phone users can quickly notice the difference.

The voice assistants we used before, such as Siri, had slow feedback; you'll have to wait until it's finished before you can have the next round of conversations.

Very rigid and a waste of time.

Actually, this is normal; after all, it's just a very original program.

Before we had a conversation with AI, we had to go through 3 steps:

1. When people talk, AI converts audio into text code;

2. AI answers this text translated by itself;

3. Convert the content of the reply to an audio output.

This is equivalent to a back-and-forth round system. There will be delays no matter what. Currently, the fastest response speed in the industry is 2 seconds.

Without saying anything else, at least it's hard for users to have a sense of immersion in real communication.

However, with GPT-4o, the average response time is only 0.32 seconds. Basically, as soon as you finish asking, it can answer you right away, which is no different from live chat.

What's more important?

Because conversations between people are full of all kinds of immediate reactions, such as, uh, ah, all kinds of expected auxiliary words, gestures, pauses, drooling, etc.

However, when you chatted with AI in the past, these factors didn't exist at all, and even if the AI answers were perfect, you still couldn't have a sense of immersion.

Now, not only can you interrupt GPT-4o at any time, it can even judge your emotions based on your speed of speech, tone of voice, breathing, and even facial expressions, and express the corresponding emotions in sequence.

That's very nice.

More than just voice responses, all of GPT-4o's text, audio, and video inputs and outputs are processed by the same neural network.

In other words, it can perform equally well in all dimensions.

Simply put, GPT is more “user-friendly” in terms of being able to see, hear, and speak.

It doesn't necessarily really understand emotions, but it can imitate them.

At this stage, just being able to imitate is enough; it's enough for commercial use.

What do you think all of the above means?

This means that ChatGPT has made another huge breakthrough in interactivity.

For example, before going to bed, you can ask GPT to use the voice of a goddess or lick a dog to tell stories and sing songs to make you fall asleep.

Another example is that you can send your daily data to GPT and let it generate work and life plans based on daily weather, emergencies, etc.

and even tutoring kids to write homework, etc...

Never underestimate interactivity; its value far exceeds your imagination.

02 Why is it free?

In addition to being powerful, what is more interesting about GPT-4o?

It's free!

Not only is GPT-4o free, but what's more exciting is that GPT Store and Vision (including code interpreters, networking features, etc.) will be opened one after another.

In order to make it easier for users to use, the new version of ChatGPT has also opened a desktop version.

On this point, Sam Altman blogged specifically:

One of OpenAI's core missions is to provide top AI tools to humans for free, create all kinds of benefits for the world, and benefit everyone. In the future, everyone will have free access to GPT computing power, which can be used, resold, or donated.

You used to criticize me for not being open source, but now I'm free, and I don't even need to sign up. Is there anything else to say?

Judging from our business logic, isn't this pure charity?

Certainly not, at least not entirely.

First, the new model is smaller, and operating costs have been drastically reduced.

As mentioned earlier, the GPT 4o doubles the processing speed and is only half the price of the GPT 4 Turbo.

The original price for entering and exporting 1 million tokens was $10 or 30, but now it's only $5 or 15 dollars.

Second, there is the commercial logic of giving up before getting it.

There is a limit to what is free.

As stated in the official documentation, free users can currently only use 10 GPT-4Os every 3 hours, and fall back to the GPT-3.5 version when used up.

10 rules, what use is it?

Wanna keep playing? Wanna have some fun? Charge me!

For only $20 per month, you can become a Premium Plus member and enjoy 80 GPT-4O per hour!

That's simple! It's so unprofitable!

As far as the current situation is concerned, for the vast majority of people, you can try to play anything as long as you're not bored; 10 pieces of content every 3 hours is completely enough.

Following OpenAI's approach, now loyal users of ChatGPT (which used to be free can only use GPT-3.4) will probably not be able to recharge any money.

Why does OpenAI take the risk of losing paid members to provide free services to all?

Actually, we can look a little bit farther.

If you think about it, everyone can use high-quality AI for free, what does that mean?

If you look at it pessimistically, this is likely to have a major impact on the current division of labor structure in society, causing widespread unemployment.

Looking at this, I'm afraid to say anything else; we can confirm at least one thing: children won't have to learn English again unless they're interested.

GPT-4o is fully capable of all kinds of interpretations, simultaneous interpretations, and even with emotion and understanding.

In addition to this, a large number of ordinary tutors, programmers, designers, etc. will basically be replaced.

This is unavoidable.

However, just like previous technological revolutions, while some industries die out, new wealth outlets will surely emerge, and the total wealth of society as a whole will inevitably increase.

Looking at it from an optimistic point of view, it's a different story.

03 The explosion of wealth

“I Ching” “Qian” Gossip “Use 9”: Seeing that the dragons have no head, good luck.

****ming also said, “Everyone has Zhong Ni in their hearts. The conscience of the heart is sacred.

A truly great era should be where everyone is equal, everyone is like a dragon, and everyone is sanctified.

From ancient times until now, this has been nothing but fantasy, but if we can make good use of AI and tools, we may not be able to get close to this level.

In fact, from GPT-3.5 → GPT-4 → GPT-4O, we can clearly conceptually feel what OpenAI wants to do:

They want GPT, an artificial brain, to meet more and more “human” standards.

What is a person?

People are more than just a labor force; no matter when or what tools are used, talent is the main source of wealth creation.

As multi-modality becomes more and more perfect, how will some of the existing industries be changed?

Our main focus should be on entertainment.

Because manual labor in the material world is bound to get farther and farther away from humans, the direction of human wealth creation will definitely focus on the spiritual side at an accelerated pace.

As can be seen from the previous investment in Description, OpenAI has long intended to introduce AI technology into the field of film and television creation.

Even if they don't do that, other film and television companies will definitely do it.

Because the future trend is “interactive media.”

You can think of this model as a short video. Everyone is a creator; no one is more professional.

Various short video platforms are now filled with a large amount of AI-generated content. We could see it before, but now it's getting more and more realistic.

In the future, as long as you make good use of large multi-modal models that mimic human emotions, the content you create will also be completely devoid of “mechanical sense.”

Everyone is the best director, so it's no problem to make more than a dozen blockbuster movies in one day.

If you want to get out of the game, it depends on whose ideas are more innovative and more suited to the tastes of the audience.

In contrast, all kinds of film and television companies, including a large number of current influencers, will have no room for development.

In the future, only platforms and countless individuals will survive.

In addition to film and television, any field of entertainment with consumer value, including music, animation, games, etc., will become the same:

Decentralized.

Everyone is a perfect musician, cartoonist, game designer, as long as you have enough patience.

Can you imagine how big of a market these will catalyze?

Take gaming as an example. By 2025, there will be 3.53 billion gamers worldwide. Billions of people, how many bizarre ideas?

Previously, 99% of people were limited to technology; they were simply players; they were reaped; only game companies made money.

From a market perspective, this development is very inefficient.

After that, it would be equivalent to these 3.5 billion people paying each other's bills, and the speed of money circulation only increased tenfold!?

Take social media, for example.

There used to be no technology. When a netizen shared his interactive experience with games, movies, and music, other users had no follow-up other than reviews.

But later, we can use this as a basis for AI to customize our second experience, whether in the form of voice, video, or comics, and share it with others.

Then others saw it, customized it, shared it...

That's how it went viral.

These descriptions are very similar to the Web 3.0 concept that was hyped up at the end of last year.

Its purpose is to create a decentralized and interactive Internet world and break the existing shackles of the Internet, which has reached its peak.

Its driving force is a large multi-modal model, and even more advanced AI that actually perfects the five senses in the future.

GPT, which has five senses, is not only a technological advance, but also a complete entertainment, consumption, and social revolution in modern business society.

All in all, allowing everyone to use top AI for free is equivalent to empowering everyone with productivity. Everyone's value will be further highlighted, and the entire Internet world will also create greater value.

By the time you discover the wealth effect, most people will probably have to live in such a big environment in the future...

When AI actually becomes an important tool for everyone to create wealth, using 10 GPT-4O in 3 hours, do you still think that's enough?

The monthly membership fee of 20 US dollars is still expensive, don't you think?

Even if it's ten times more expensive, you're willing to rush to buy it!

What we should really worry about should not be this fuzzy thing, but rather: AI technology is advancing so fast, are you aware that you need to adapt to the new era?

Don't be one of those people who are lagging behind.