King Alexander first reveals himself: "Muse Wonder messes up, and Meta goes bankrupt!"

  It's the first time in the history of the world.

It's never been so crazy to bet on a card table of a tech giant.

Recently, in an interview with Alexander King, the current head of Meta AI, he revealed himself to the public for the first time:

It is estimated that around US$10 billion to US$20 billion will be raised.

The giant was stung by a hard-nosed, hard-headed dashboard with no validation on Meta's front-end model!

If the new model fails, Meta will face a serious financial crisis.

The multi-billion-dollar Meta Empire has been tied up on a high-speed AI train headed to the cliffs. Successful, it will be crowned; unsuccessful, it's a hell of a ride ahead.

The AI competition is already so hot?

What's behind this billionaire death row?

Recently, Stripe co-founder John Collison invited Zuckerberg and Alexander Wang into an Irish-style bar in his podcast Cheeky Pint stall.

It's time to go back to the end of last year.

At the time, the core executive team of Meta, in Hawaii, was assembled to form an annual executive team.

On the surface, it was a light-hearted cocktail party. But the whole party was actually dark, and you could even say "cold sweat".

Alexander Wang, who had just been excavated by Meta, had a life-long experience here.

During the build-up gap, Meta CFO Susan Li steadily pulled her into a no-man's-land.

King Alexander reveals: "In private, Susan Li repeatedly warned him harshly, with an extremely serious and even desperate tone:"Listen, if this cutting-edge AI model fails, the consequences will be very, very serious from a financial standpoint!"

The CFO's panic was not without reason, as she had just learned of the crazy decision of CEO Zhao.

At the same time, he called Alexander Wang his predecessor. Even the entire front-end model architecture of Muse Wonder was still in its very early stages, and even the most core technical verification had not been completed.

The war on AI is already raging, and the frenzy of OpenAI and Google is putting pressure on him.

Zuckerberg looked at Wang, personally asking him for his final "bottom line": "Are you sure this can happen?"

In the face of pressure from the boss, Wang gave a positive answer.

After receiving this sentence, Zuckerberg's eyes closed and he immediately issued a cheque for an infrastructure investment of $100 billion to $20 billion!

The money will be turned into GPUs, data centers, cooling towers, and amazing electricity enough to light up several cities.

Alexander Wang's explosion completely exposed the wreckage of Meta.

This is not a rational plan at all, which suggests that Meta has placed the fate of the entire company on an AI project that has not yet been developed.

If you make a mistake, the company is facing a direct disaster; if you win, the stock price can go up.

Under heavy pressure, Alexander Wang and his team embarked on months of hell-level research and development under the CEO's "100-billion-blind shot".

What's with the money?

A year ago, when Meta announced that it was investing heavily in acquiring Wang's team, the entire network was mocking the "big head".

However, the run-rate data from Muse Wonder can be described as eye-catching.

In terms of integrated intelligence, the full version of Wonder Full took a staggering 56 points! directly over Fable 5.1, behind Opus 5.5.

In terms of long-range coding capabilities, Wonder has demonstrated a dominant level. The DeepSWE test has a passing rate of about 79%.

From open-source Llama to directly killing off the closed-source top Agent track, Meta has officially declared its dominance over long-cycle software engineering at the forefront!

As the netizens commented: "God damn it, you have to look at the layout right now, this guy on the eve of the biggest win of his life, he's barbecuing at ease!

However, as soon as it was released, Muse Wonder was met with a fierce battle of wits.

Many people say that the reality of what it's testing is stretching across the board.It's called "benchmaxxed".。

In the Lava lamp test, there were no lights, no glass, no lava, and even a red cylinder.

It's a spectacular effect!

In response to this, Alexander Wang personally came down hard.

He responded ruthlessly to the X:

When we developed Muse Wonder, our goal was to build a model that would be useful for the real world, for real humans to perform real missions. We focused on things that didn't look so "bright" on Twitter -- like not going to those fancy 3D game shows.

That's the kind of thing that makes Vesuvius think he's better than a man!

Big V Token Gremlin says, "Muse was so bad when he wrote Three.js! Even a simple Minecraft clone could not be tolerated. It was disappointing".

A sharper question comes from pricing strategies.

One developer noted that the "bloodless version" of the open API called Muse Wonder currently has only 52 runs, but the single price is the same as the Opus 5.5 with 58 runs.

Wonder Full is open to subscribers only.

In the Cheeky Pint anime podcast, Zuckerberg and Wang detail the real reason why Muse Wonder is called a "monster" - Active Memory and Active Focus.

As some of the developers have said:It's not a one-sided run-off that's really worthy of attention, and when the model really interferes with your e-mails, schedules, and personal documentation, its Active Memory can be terrifying.」

Active Memory: It has super-time memory. "We've put a lot of effort into making sure the model remembers what's important to you for days, weeks, or even months", the official Meta said. "It's no longer a "goldfish memory" chatbot, but a deeply remembered cybernetic part of all your preferences and habits".

Active Focus: A.I. of the past will get messed up. But Muse Wonder has been injected with a lot of "real-world failure scenarios". When it gets stuck doing tasks, it will actively look for alternatives; it will stop and come to you for confirmation before performing critical operations.

In the demonstration, Muse Wonder was able to continuously search online for rental information, filter based on personalized conditions, and even independently contact the landlord!

This is a real super personal assistant working 24 hours a day for you.

So what's Meta going to do with all these crazy chips?

In the podcast, Zuckerberg gives his judgement:"AI is definitely not a winner".

He did not believe that the world would be monopolized by a "Great Model of God" such as OpenAI. On the contrary, he firmly believed that the future would belong to a myriad of super-intelligents.

This also explains why Meta created a "super-intelligent lab" and why they came up with a hardware vision of "smart business" and "beyond smart phones".

Imagine that in the future you're wearing Meta's smart glasses and living inside the Muse Wonder with Active Memory.

Your personal agent will communicate with the merchant's agent, cut prices, and buy goods.

If it really happens, Meta will completely reshape the global digital business model - that's the future that the $20 billion wants to buy!

Now in Silicon Valley, you're either going to burn billions of tons tomorrow, or you're going to die with billions in cash.

Zuckerberg picked the first one. He pushed the Meta over the grill, put the gun on the grill, and unlocked the grill with the Muse Wonder.

In the meantime, the baby is in the middle of a headache.

Meta did not go to financial disaster, but instead succeeded in killing off the top tracks of the cutting edge model, bringing personal superintelligence to the hands of billions of users.

What's wrong with Muse Wonder? Let's keep the model open.