March 31st.
NeusAI's presentation will be held at 10:00 AM United States West Coast.
However, CEO Ryan is not in a good state right now.
Of course, it's not just because of the press conference in a little while. In fact, he's held many press conferences like this before, so this kind of thing doesn't make him nervous anymore.
What he's worried about now is OpenAI's press conference in two hours.
Yes, on the same day, NeusAI and OpenAI will hold press conferences one after the other, but neither of them has released any information about the content to be released.
Ne. usAI's press conference announcement was released a week ago, and on the same day, OpenAI scheduled its press conference for two hours later on the same day.
Ryan knew that this was OpenAI CEO Sam targeting him.
Ryan had worked under Sam and knew his style. Sam was not only an AI godfather, but also a successful businessman, speaker, and speculator. Sam would never miss an opportunity to promote himself. He would treat every press conference like a battle.
It was precisely because Ryan knew Sam so well that he felt so uncertain. Even the fact that Neus's newly trained Venus model had proven to be the best Large Model in the world didn't give Ryan enough peace of mind.
Sam and his OpenAI may have some big surprises in store.
Over the course of the week, Ryan used his connections in Silicon Valley to inquire from various sources, including former OpenAI colleagues and investors—basically everyone except Sam himself—but to no avail. This time, Sam had done an excellent job of keeping things secret.
But the die was cast, and Ryan couldn't postpone his NeusAI launch event for an OpenAI event whose content he had no idea about; otherwise, he would become a laughing stock in the industry.
John, the CTO of Ne. usAI, watched Ryan pacing around the room and suddenly said, "Don't worry too much. We all know Sam. He might just be bluffing."
Ryan stopped in his tracks upon hearing John's words and looked at his old friend and business partner. He knew why John had said that; with the press conference imminent, Ryan couldn't let this pressure break him.
"Don't worry, I'm waiting to see Sam's miserable face when the Venus data comes out," Ryan joked. Neither of them liked Sam, which was one of the reasons they got along.
Soon, it was time for the NeusAI launch event. Ryan left the preparation room and went to the launch hall.
The press conference was held in a building in Palo Alto, rather than NeusAI's own offices, mainly because the event was large-scale, with many media outlets invited, and required a separate stage and large screen to demonstrate the new model's effects.
Ne. usAI's office couldn't hold them all.
The press conference hall was now set up, with a temporary stage in the center and a large electronic screen behind it displaying a sentence.
The new frontier of intelligence.
(Smart New Frontier)
The background is a light blue planet.
Marketing director Michael was standing in the audience, holding a tablet computer with the press conference agenda displayed on it.
The event opened with the unveiling of a new model, the release of test results, a demonstration of capabilities, and an update to the security strategy.
The entire press conference lasted about an hour, followed by a Q&A session with the media.
Michael greeted Ryan as he approached.
Ryan nodded, straightened his tie, and suppressed his unease.
"Venus is the best," he told himself.
As the lights dimmed, Ryan walked to the center of the stage, and applause erupted from the audience.
Ryan didn't speak; he stood there until the applause lasted for more than ten seconds before he raised his hand.
The hall fell silent.
"The Artificial Intelligence industry has experienced a lot of debate over the past year."
Some say that Scalinglaw is nearing its end.
Some say that growth in modeling capabilities is slowing.
Some say that the next generation of AI needs a new path.
"But today, I will prove that this road is not yet over. Please allow me to present you with NeusAI's most cutting-edge model— the Venus model."
As Ryan spoke, the words behind him gradually faded away, and the light blue planet slowly rotated out of the frame. At the same time, a larger planet, radiating golden light, moved from off-screen to the center of the frame.
Then a line of large characters appeared.
NeusVenus.
This is Venus, and also the name of the Roman goddess of beauty.
Starting today, it will also represent the highest level of Large Model capabilities in the AI industry.
"This is our answer: not to slow down, but to keep moving forward."
Rows of data begin to appear on the screen, eventually forming a complete table.
A table comparing multiple models.
Only four models are listed above: Venus, Mercury, GPT-4, and Gemini.
Ne. usAI is benchmarked against the world's top models.
In the various rankings, Mercury is slightly weaker than GPT and Gemini, while Venus ranks first with a significant advantage.
First.
Applause erupted from the audience again, this time even more enthusiastic than at the beginning. Ryan raised his hand to quell the applause, but couldn't.
A reporter on X provided live text and images, posting the comparison photo with the caption. "Neus has just been crowned."
After Ryan finished speaking, he invited Sarah to the stage. She demonstrated the capabilities of the NeusAtlas Agent based on Venus. Complex problems that were originally thought to be solvable only by senior engineers were fixed on the spot by the Venus -based Atlas.
This demonstration proves that Venus is not just a figurehead for boosting rankings, but a genuine source of productivity.
Not only are the models powerful, but the accompanying tools are also provided.
A reporter commented on X: "OpenAI is in trouble now."
After Sarah finished her demonstration, Ryan returned to the stage.
He announced the price for model calls.
It's more expensive than Mercury, more expensive than Gemini, and more expensive than OpenAI.
But nobody at the scene cared, because when it comes to the best models, price is never the primary concern.
Ryan looked down at the audience and began the final segment.
"We believe that the most advanced Artificial Intelligence capabilities should serve the most trusted partners."
Security policy updated; this message appeared in large print on the screen.
Then a world map appears, with North America, Europe, Israel, Japan, South Korea, and Australia marked in light colors.
The remaining areas are marked in red.
"We must face the fact that cutting-edge AI capabilities are becoming a strategic resource."
"Over the past few months, we have observed large-scale anomalous calls, model Distillation attempts, and potential requests for high-risk military and intelligence purposes from certain regions."
A murmur arose from the audience, which included journalists of various nationalities and ethnicities. They had many questions about NeusAI's sudden statement, which even overshadowed the release of the new model.
"Therefore, we have decided to implement a new user policy starting April 1, 2027."
As he finished speaking, several text rules appeared on the screen.
Access from China IP addresses will be restricted.
Entities with more than 50% China capital holding a controlling stake will be prohibited from directly or indirectly using NeusAI's Large Model APIs. Overseas subsidiaries, proxy calls, and third-party resales will also be subject to restrictions.
Accounts that violate the rules will be terminated immediately.
"Ne. usAI has a responsibility to ensure that the latest AI capabilities are not misused. We will continue to work with developers, businesses, and governments in trusted regions, but NeusAI will not assist any organization that attempts to copy, Distillation, or weaponize our technology."
As Ryan finished speaking, the phones, cameras, and video recorders in the audience continued to film.
This is the real big news today.
While NeusAI topped the list, it also put forward clear directions for the regulation of AI.
During the Q&A session, almost everyone raised their hand.
Ryan called on Bloomberg reporter.
"Mr. Miller, is this new policy targeted at any specific company?"
Ryan frowned almost imperceptibly and replied, "Of course not. We made this decision out of a responsible attitude towards the use of AI, and out of consideration for humanitarianism and counter-terrorism. It is not targeted at any specific company or individual."
Then he called on another reporter, who stood up and asked, "Mr. Miller, there was a China software called Yufeng that reportedly also used the Mercury Model and was in the same market segment as Atlas. They didn't release it in the United States, but it had a very high reputation in the community. What impact do you think this new policy will have on them?"
Ryan suppressed his anger and calmly replied, "That's beyond our consideration. I believe the company can find its own solution."
Then, without waiting for the reporter to continue, he asked, "The next question is for those who have questions about the new model. If it's about the new user policy, please put your hands down first."
Ryan then answered a few more questions from the media, and the press conference came to an end.
Ryan returned to the lounge, where John, Michael, and Sarah were already there. They had prepared a bottle of champagne, just waiting for Ryan to open it.
Seeing Ryan come in, John smiled and said, "Well done, old buddy."
Michael handed over the champagne.
Ryan waved his hand, not taking the glass, but instead walked to the table, poured himself a glass of water, and drank it all in one gulp. Only then did he speak, "I'm not in the mood right now. Any news from OpenAI?"
Michael glanced down at his phone; it was 11:32.
"There are still thirty minutes until their press conference."
Ryan was silent for a moment, then suddenly said, "Open the live stream link. Let's not go back to the company yet. Let's watch their press conference here, otherwise I won't be able to relax."
After hearing this, Sarah opened the laptop on the table and started working on it.
Outside, reporters were hurrying out, rushing to the OpenAI launch event.
Interestingly, OpenAI also chose Palo Alto for its launch event, rather than its headquarters in downtown San Francisco.
This was certainly convenient for reporters, allowing them to quickly get to the next event after attending one. However, reporters also keenly sensed a targeted element; OpenAI's press conference was clearly aimed directly at Ne. usAI.
I just don't know what kind of content they will release.
Thirty minutes passed quickly, and soon the press conference livestream started. The screen showed a stage, but it was connected to Neus...
Unlike AI launch events, this stage was brightly lit and didn't have a huge screen. There was only a long sofa and a coffee table in the center of the stage, with a television set next to them.
At the start of the press conference, OpenAI CEO Sam Altman came on stage with two people, both of whom were of Asian descent.
The three of them walked over to the sofa and sat down, introducing themselves. Both of them were senior researchers at OpenAI, each leading the development of their respective models.
Wait a minute or two?
The reporters in the audience had already keenly realized what was going on.
At this moment, Sam spoke up, "The first product we're bringing you today is our latest model—GP."
TV.
The TV next to me was turned on, but instead of a promotional video or a pre-recorded product introduction, it showed a clean, almost rudimentary chat window.
There are no dazzling animations, no grand planetary backgrounds, just one line of text.
GPT-V.
Sam sat on the sofa, leaning slightly forward, holding only a cell phone instead of a remote control. He glanced at the audience, then at the two researchers beside him.
"For the past few years, we've been pursuing one goal: to go further in ScalingLaw. We don't believe it's the only direction, but it's still worth exploring."
A leaderboard appeared on TV, the same one that appeared at the NeusAI press conference a little over an hour earlier, showing scores for various benchmark tests and comparisons between different models.
It even includes data on Venus that NeusAI just released.
However, there's an extra column of data.
GPT-V narrowly surpassed the Venus model to take first place.
The media reporters below held their breath, and countless people were pounding their fists on the table in excitement in front of their screens.
It's so dramatic! Same Joe, two press conferences, only two hours apart. Venus had just finished his presentation, announcing that the model's capabilities had reached the top, and two hours later, GPT-V proved him wrong.
OpenAI is OpenAI; it's completely outclassing Ne. usAI.
Did Sam know this would be the outcome? Of course he knew! This is a blatant humiliation!
Ruiling stood in the lounge, her face ashen. Michael and Sarah stood to the side, barely daring to breathe. John opened his mouth, but no words came out. At this point, saying anything would be pointless.
The real-time comments on X refreshed quickly.
"Wait, did OpenAI just surpass Ne. usAI's Venus?"
"Silicon Valley is so lively today."
After the applause subsided slightly, Sam said, "This is GPT-V, OpenAI's latest achievement. But that's not the focus of our Jinqiao press conference; we have another small thing to announce."
The audience understood the Steve Jobs reference he used and became excited.
In NeusAI's lounge, Michael muttered under his breath, "From death."
Sam turned to the side and looked at one of the researchers next to him.
"Next, we at Jinqiao are very excited about something. We believe that future human-computer interaction should not be limited to text boxes, so we've brought GPT-VOmni."
The TV next to them was connected to Sam's webcam and displayed his webcam screen.
Sam pressed a button in the ChatGPT app, the dialog box disappeared and replaced with a live video feed. The video showed the view from the front-facing camera of the device— Sam, two researchers, the sofa behind them, and the background.
The researcher took the scanner, held it up, and scanned the area.
What can you see now?
A dull, monotonous female voice came from the television.
"I saw a press conference stage. Three people were sitting on the sofa, the one in the middle was Sam Altman. There was a glass of water on the table, and a TV next to it was showing my live feed. There should be an audience, because I could hear ambient sounds and occasional voices."
A huge commotion broke out at the scene.
Real-time interaction – this is an unprecedented form of interaction.
The researcher pointed the camera at Sam.
Why does he look so serious right now?
ChatGPT continued without pausing, "He seems a little nervous, probably waiting for some feedback; given our current situation, maybe he's waiting for the audience's reaction."
Gao Xia laughed, Sam laughed too.
Ruiling stared at the live stream on the screen, silent for a long time, his finger tapping rhythmically on the chair.
It was being tapped rhythmically.
At the press conference, after the applause subsided, Sam spoke again.
"GPT-VOmni can understand text, speech, images, and video in real time. It's not just piecing together multiple systems; it's the world's first end-to-end multimodal model."
The audience, both in person and online, has already understood the meaning of 0mni. It is a prefix derived from Latin, meaning "all" or "total". This model is an all-around model.
"GPT-VOmni will be gradually rolled out to developers and enterprise users in the coming weeks. We will also integrate some of its capabilities into ChatGPT, allowing more users to experience natural multimodal interaction."
Finally, the television screen froze on the names of the two models.
GPT-V
GPT-VOmn i
A flagship general-purpose model.
A native multimodal model.
When the two models appeared side by side, OpenAI had completely taken control of the atmosphere at the press conference.
Sam then casually concluded, "We respect all our peers' exploration of AI capabilities, but we also acknowledge that simply focusing on improving rankings will not help achieve AGI."
This statement didn't name names, but everyone could tell who it was targeting.
Ne. usAI's Venus model focuses on dominating leaderboards, but it still can't beat GPT-V, while OpenAI has brought the world's first native multimodal model.
OpenAI truly lives up to its name; its parent company is still its parent company.
This press conference was even shorter than Ne. usAI's, and it quickly moved on to the Q&A session.
The first reporter to stand up was from The Verge.
"Sam, two hours ago Ne. usAI announced that its Venus model surpassed GPT and Gemini on multiple leaderboards, claiming it represents the current pinnacle of AI capabilities. What are your thoughts?"
This is a provocative question, but it's exactly Sam's type.
"I believe everyone can see who currently represents the highest level of AI. I hope NeusAI continues its efforts to develop even better models."
Sam said this with that familiar smile on his face.
The second reporter is from Bloomberg, the same reporter who asked Ruiling the first question.
"Ne. us also announced a new usage policy, restricting China IP and China capital holding companies from using their APIs. Will OpenAI take similar measures?"
Sam's smile faded, and he said, "We will comply with all applicable laws and regulatory requirements, and will strictly follow the guidance of the United States. government."
The implication is that we are not targeting anyone in particular; we only target those whom the United States government tells us to target.
After answering a few questions, OpenAI's press conference came to an end, leaving the world's onlookers who are always following the latest developments in the AI industry thoroughly entertained.
That's so satisfying! Getting slapped in the face right there and then—even novels wouldn't dare write something like that.
Tech media outlets are releasing news flashes one after another, each with an increasingly eye-catching headline.
OpenAI reclaims the top spot two hours after Neus's coronation.
From a single-rhythm model to a new future direction, OpenAI's counterattack.
<United States AI companies are simultaneously initiating restrictive policies; where is the AI competition headed?>
Within hours of the press conference, someone had made a comparison video, editing the two press conferences into a single video.
LeBian is standing in front of the huge golden planet, announcing Venus's ascension to the top.
On the right is Sam sitting on the sofa, holding a phone and showing his GPT-VOmni live footage of the dance.
Just two hours ago, NeusAI was the most dazzling new king in Silicon Valley.
Two hours later, it became the backdrop for the OpenAI press conference.
In NeusAI's lounge, the live stream had ended long ago, but no one spoke.
The champagne on the table hadn't been opened yet. Because it had been out of the ice bucket for too long, a layer of dense, trailing water droplets had condensed on the outside of the bottle, slowly sliding down the glass.
Ruiling sat in the chair, looking at the blacked-out laptop screen, in which his face was reflected.
His uneasy premonition came true.
OpenAI certainly didn't disappoint.
John finally spoke, "Even with this one, Venus is still the second-best model. And, of course, we still have Atlas."
Ruiling nodded. OpenAI's Jinqiao has only demonstrated its model capabilities; they haven't made any moves in the Agent field yet. This is NeusAI's opportunity.
After experiencing the ups and downs of life, and hitting rock bottom, he felt even more determined.
It would be more interesting if OpenAI could defeat this guy.
"Back to the company," Rui Ling said in a low voice. "Everyone stay tonight. Our next step is to make the Venus and Atlas combination number one in the industry. After Ming Qiao, Riding the Wind from China will no longer be a threat; we will be the leader in the Agent field."
After he finished speaking, he pushed open the door and walked out.
In the corridor outside, workers were still packing up the materials from the press conference.
There was a standee that hadn't been taken down yet, depicting a huge golden planet with the word "Venus" written on it.
Venus.
God of beauty.
It is also the brightest star at dusk.
Unfortunately, it's still dark in Silicon Valley.
At this very moment, on the other side of the ocean in China, it is still the early hours of April 1st. Apart from a few night owls, most people are still asleep.
However, bloggers who follow the latest developments in AI could not contain their excitement and compiled the two Silicon Valley press conferences and the new user policies into news articles, which they posted on Weibo, Little Red Book, and other AI communities.
Public opinion is already slowly spreading, and it's just waiting for Qiao Liang to be mentioned before it explodes, causing a huge uproar in China's AI community.