Pogbagram

This is a Samsung phone that "needs to be sold at a higher price".

With the official release of the Galaxy Z TriFold last December, the tri-fold phone category finally welcomed its second long-awaited competitor.

After reviewing the Galaxy Z Fold7, Mate XTs and Mate X7 last year, my main feeling after switching to the Galaxy Z TriFold is: it's finally here.

The second leading player in the industry.

Design: Advanced yet crude

As a long-time foldable screen user, I think the Samsung Galaxy Z TriFold is the first foldable phone that I can truly say I can enjoy using for a long time.

The greatest market significance of the Samsung Galaxy Z TriFold is not only that it enriches the tri-fold product category, but also that it finally brings a long-standing question to the forefront:

What are the uses of a tri-fold phone?

As the most expensive and technologically advanced mobile phone product at present, tri-fold phones are usually associated with keywords such as "premium" and "high-end", making it difficult to notice other factors.

But Samsung's tri-fold design seems to be an attempt to shake off that impression.

After all, both user feedback and Samsung's marketing efforts have aimed to make the Galaxy Z TriFold an "efficiency tool," a solution where 1+1+1>3.

The bad news is that Samsung's tri-fold phone is indeed extremely understated in its design, available only in black, with few decorative elements, and so simple that it's almost plain.

The good news is that, stripped of all factors related to face, status, and symbolism, the Samsung Galaxy Z TriFold does indeed prove one thing:

This is the most utility-oriented foldable phone to date.

First of all, the Galaxy Z TriFold has very obvious utility attributes in terms of hardware CMF (Color, Material, Finish).

Whether it's the "polymer" back panel with its carbon fiber texture or the third section of the body that's been widened for easier unfolding, both designs are quite different from other foldable phones on the market.

In other words, the Galaxy Z TriFold is somewhat like an extreme track car like the Ariel Atom— sacrificing most aesthetic pursuits for hardware attributes, resulting in a feeling that is both advanced and rough.

The Z TriFold has the same problem as other racing toys: its performance is undeniably excellent, but when it's put on a poster, it certainly doesn't attract as much attention as the big brands next door.

▲ Image｜Top Gear

Meanwhile, this pursuit of functionality extends to the Z TriFold's configuration choices, bringing out the highest-spec combination Samsung can currently fit into a foldable screen: a specially tuned Snapdragon 8 Elite processor, 16GB of RAM, and the soon-to-be-signature 200-megapixel main camera—

Although this combination essentially copied the hardware specifications of the Z Fold7 , Samsung was at least honest and did not cut the configuration due to excuses such as "the space is more cramped in a tri-fold body".

It even features the Ultra series' exclusive 45W SuperCharge 2.0 and a 5600mAh battery, matching that of its competitors .

However, compared to the back panel that attracts fingerprints no matter how you hold it, the weight of the Galaxy Z TriFold is even more noticeable.

Not only does Samsung's signature right-angled frame design make it quite heavy, but the Z TriFold's bare weight (including screen protector) reaches 310 grams , and with the official stand case, it gets close to 350 grams.

Even just by placing the Z TriFold and Z Fold7 side by side and weighing them, you can clearly feel the weight of the three-fold design, which in turn enhances its heaviness.

Therefore, during the time I was using both the Z TriFold and the iPhone Air, the weight was the most uncomfortable aspect.

Efficiency: Fold once more, double the profit.

However, if we set aside the weight and simply look at how much this 10-inch screen improves the daily user experience, the results are equally significant—

The efficiency improvement of a tri-fold design compared to a double-fold design is even more tangible and noticeable than that of a double-fold design for a candybar phone.

The main reason is that the aspect ratio of a triple fold is completely different from that of a double fold.

Although the Z TriFold's 10-inch screen may not seem significant on paper compared to the Z Fold7's 8-inch internal screen, the aspect ratio has been increased from 10:9 to 4:3, meaning the display area has increased by a full 50% , from 204cm² to 305cm².

The most direct result of this "non-proportional expansion" is that when watching the most common widescreen videos, the Z TriFold has an actual display area that dual-folding and candybar phones cannot match:

Even some websites with extremely poor mobile user experience—such as a well-known academic paper website—become perfectly compliant on the Z TriFold's size and proportions, and its usability is even better than double-folding:

What many people overlook is that the Z TriFold offers an additional "portrait screen" usage scenario compared to the Z Fold7—not using the external screen, but turning it vertically as a 10-inch mini tablet:

At this point, many PDF documents that were too small to view on a mobile phone or not fully displayed on a double-fold screen can be fully displayed on a 4:3 aspect ratio screen, and the details can be seen clearly without zooming in.

For a "research-oriented" editor who frequently needs to consult technical standards documents, the Z TriFold PDF reader experience is near perfect. The only drawback is that it no longer supports S-Pen.

Besides viewing PDFs or browsing Threads timelines, having a 10-inch YouTube Shorts that you can easily unfold is also a very unique experience.

However, this kind of enjoyment also has its drawbacks, as it makes it easier to become addicted to watching short videos.

Furthermore, Samsung's One UI plays an indispensable role in the user experience of the Z TriFold. After all, Samsung possesses the complete Android underlying system and has a great deal of freedom in providing third-party solutions for incompatible apps.

▲ Bilibili HD (left) and PiliPalaX (right)

More importantly, after several system updates, Samsung added all the GoodLock plugins to the Z TriFold, with the fully-featured One Hand Operation+ (OHO+) playing an unprecedented role.

In short: OHO+ has for the first time made it possible to use a tri-fold phone with one hand.

With a simple setup, you can quickly bring up a small touchpad by swiping from the side. You can then tap buttons and links on the diagonal of the screen with just your thumb, without having to change your grip or lift your hand .

The OHO+ convenient menu bar has the same function. As a "control center" that can be accessed with one finger and covered by the thumb, it completely solves the problem of having to frequently reach out to open the control center.

However, it must be said that the Z TriFold, as Samsung's first mass-produced tri-fold product, still retains the immaturity characteristic of first-generation products in the most crucial aspect: "screen quality."

This is rather stingy for a giant that stands at the top of the global flexible screen supply chain .

The Samsung Z TriFold's internal screen resolution is 1584 × 2160, with a PPI of only 269, which is quite unsatisfactory by 2025 standards. I think it should at least maintain the same 368 PPI as the Z Fold7's internal screen.

Meanwhile, OneUI still retains Samsung's long-standing problem of blurry system icons.

Fortunately, the Chinese version of OneUI 8 is one of the few Android systems that allows users to freely change icon packs and launcher icons . After changing the icon pack, the blurry issue will be significantly improved.

At a normal viewing distance on a 10-inch screen, 269 PPI doesn't actually make it noticeably blurry. It might be a bit off-putting , but it doesn't affect usability.

However, it's really frustrating that the Z TriFold's LTPO internal screen doesn't support Always-On Display (AOD) at this resolution and PPI.

Is a tri-fold phone that costs 20,000 yuan absolutely necessary to buy?

Every time I open the Z TriFold, I can't help but think about one question:

The current double-fold design is nearing maturity. While the triple-fold design has obvious advantages in certain scenarios, is it worth sacrificing price, weight, feel, and durability?

Having reviewed both the Huawei Mate XTs and the Samsung Galaxy Z TriFold, my answer to this question is: Yes, but not entirely.

First of all, the increased screen size itself is an improvement in efficiency , which is undeniable in today's "visual"-dominated online world.

Whether it's a more utilized video screen or a properly formatted desktop webpage, the significant improvement of tri-fold over double-fold is worth the trade-off in weight and feel.

In addition, the tri-fold design makes some scenarios that were previously difficult for mobile phones more practical, mainly the "split-screen mode that requires switching back and forth between operations".

In situations like this, the Z TriFold truly delivers on its promise of being "better than three in one":

▲ A classic use case for tri-fold designs: left side for searching, middle for price inquiries, and right side for reviews.

In addition, the Z TriFold also includes Samsung's exclusive technology: Samsung DeX.

As the only manufacturer still persisting in promoting the "mobile PC mode," Samsung's DeX mode has reached a peak of maturity on the TriFold, freeing users from the need for an external monitor.

Furthermore, DeX is far more similar to Windows PCs than the desktop modes on the Z Fold7 and Mate XTs, which are based on a "small window logic".

This is the closest PC-like experience you can get on your phone without messing around with Moonlight streaming; you can even connect an external monitor to use it.

Since Samsung has updated many DeX features on the Z TriFold, due to space limitations, we will bring you a dedicated review of the DeX mode later.

In short, after actually touching, using, and even dropping the Samsung Galaxy Z TriFold, I can feel that the tri-fold design itself is a very "future-proof" technological concept.

It is both exquisite and appears to be "powerful and powerful".

The problem is that the Z TriFold itself belongs to an even smaller niche within the already relatively niche foldable screen category.

Therefore, if you don't have very specific, firm functional needs that "can't be solved without a triple fold," then the dual-fold phones of the last two years are undoubtedly a better choice.

After all, as of the end of the review, Z TriFold still required pre-ordering on major e-commerce platforms and offline stores also charged extra for its products.

For most consumers, paying 19,999 plus an additional 2,000 yuan to buy Samsung's "first-generation product" is a very uneconomical deal .

At the same time, we need to note that 2026, due to the rumored existence of the iPhone Fold, will undoubtedly be a turning point for the accelerated maturation of foldable screen phones.

Although Apple is unlikely to venture into the tri-folding market in the short or long term, buying a foldable screen this year will certainly offer a great experience .

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

ifanr | Original Link · View Comments · Sina Weibo

Doubao 2.0 is rumored to be launching before the Spring Festival.

According to LatePost, Doubao 2.0 may be launched before the Spring Festival this year.

The report states that the upcoming Doubao 2.0 model is the most significant output of Wu Yonghui (head of basic research at ByteDance Seed) in his first year at the helm. It is a multimodal model similar to Gemini, with 1 trillion parameters, making it the largest model trained since Seed's inception.

According to reports from Seed, the model encountered infrastructure challenges during training. They analyzed that during Seed's two-year pursuit of excellence, it relatively neglected building fundamental capabilities, leading to instability when scaling parameters during the training of Beanbag 2.0, which at one point hindered progress.

According to reports, Weng Jiayi, head of RL Infra at OpenAI, said in a podcast that every model team's Infra has bugs. The essence of model companies is the speed at which they can fix bugs in the Infra, which determines the number of ideas that can be validated per unit of time. And ideas can be solved by increasing the density of talent.

The report points out that for the Seed team, rebuilding the Infra system is even more difficult.

It is understood that Seed's Infra team has hundreds of people, supporting the research and development and experimentation of dozens of models within Seed, and is considered by senior management to be the best in China. "To restructure, a lot of manpower and resources need to be invested, and a considerable trust cost must be borne," said a Seed employee, adding that they can only "fix the wheels while driving."

It is reported that after encountering problems during the training of Doubao 2.0, multiple teams worked together for three months to solve the problems, mainly focusing on model architecture and training data, to ensure that the model could be launched before the Spring Festival.

In addition, ByteDance's latest video generation model, Seedance 2.0, has recently been launched on platforms such as Xiaoyunque and Jimeng, and has received widespread attention.

Seedance 2.0 is reportedly capable of creating cinematic-quality videos from text or images. It employs a dual-branch diffusion transformer architecture, generating both video and audio simultaneously. With just detailed prompts or an uploaded image, Seedance 2.0 can generate a multi-shot video sequence with native audio within 60 seconds.

It's worth mentioning that this model's unique multi-camera storytelling feature can automatically generate multiple interconnected scenes based on a single prompt. The AI ​​automatically maintains consistency in characters, visual style, and atmosphere across all scene transitions, eliminating the need for manual editing.

Yesterday, Feng Ji, producer of *Black Myth: Wukong* and founder of Game Science, also published an article reviewing Seedance 2.0:

• Leading: The most powerful video generation model currently available, bar none;
• All-around capabilities: AI's ability to understand and integrate multimodal information (text, images, video, and audio) has made a remarkable leap forward.
• Production capacity explosion: The production cost of general videos will no longer be able to follow the traditional logic of the film and television industry, and will gradually approach the marginal cost of computing power. The content field will inevitably usher in unprecedented inflation, and traditional organizational structures and production processes will be completely restructured;
• Low barriers to entry and ubiquitous (AI) video: Any presentation method that previously required consideration of production costs will be easily converted into video, such as e-commerce advertisements and pre-recorded videos. Although the core aspects of game development have not yet been directly eroded by AI, video itself will gradually evolve towards customization, real-time interaction, and gamification, perhaps becoming a completely new form of entertainment in the future.

Feng Ji also pointed out that with increasingly sophisticated video models, the "proliferation of fake videos and trust crisis" will worsen. He stated that realistic fake videos will become virtually effortless, posing an unprecedented challenge to existing intellectual property and censorship systems. In response, Feng Ji suggested:

Please remind your parents and relatives who are unfamiliar with AI as soon as possible that any video content lacking official verification (especially those containing personal images and voices) may be fake. Please verify information through multiple channels to avoid being scammed.

Shangjie Auto refutes rumors that AI generated winter test images

According to Shangjie Auto's news yesterday, it recently released pictures of the Shangjie Z7 undergoing winter testing, but the pictures have sparked heated discussions among netizens due to their suspected AI generation.

In response to this situation, Shangjie Auto stated:

All winter test images released by Shangjie Auto are real photos, not AI-generated. To maintain confidentiality regarding product design details, we have camouflaged the vehicle body and blurred certain details.

Shangjie Auto stated that the winter test of the Shangjie Z7 was a real-world test in extremely cold conditions, aiming to verify the vehicle's performance in low-temperature environments and bring a more reliable and safer product experience to everyone.

Previous images showed that the Shangjie Z7's winter test photos contained images of "wheels rotating but tires remaining stationary," leading some netizens to question whether the images were generated by AI.

OpenAI: The so-called leaked AI hardware is fake!

According to The Verge, there were recent reports of a Super Bowl ad claiming to be "OpenAI's first AI hardware," but OpenAI responded yesterday morning that it was "fake."

OpenAI President Greg Brockman responded to the "fake news" claim on X, while OpenAI spokesperson Lindsay McCallum Rémy stated that it was "completely fake."

It has been reported that the aforementioned false information has been deleted. The report stated that the content described an employee complaining that an advertisement they helped produce was not aired, and that they had accidentally leaked the purported advertisement video.

The screenshots show the advertisement featuring Alexander Skarsgård, the protagonist of "Killing Bots," wearing OpenAI's first hardware device. The product he's wearing is a bright silver clip-on TWS earphone with an oval-shaped silver charging case.

The Verge analysis noted that the accounts found to have leaked the ads were very new, and the change of profession of the account owners who leaked the false information was also very sudden.

Additionally, Gillian Follett, a reporter for the foreign media outlet AdAge, was also affected. She pointed out on social media that she had not written any reports related to "OpenAI's first AI hardware being revealed," nor had she published any article about "OpenAI being removed from the Super Bowl."

Honor deploys system-level AI recording minutes

According to Blue Whale News, Plaud, a hardware company specializing in AI-powered meeting minutes tools, is developing a native AI recording function for Honor's OS system, enabling the phone's native app to directly generate meeting minutes.

According to sources, Plaud only provides software-level support, and related functions can be implemented by upgrading the Magic OS version later.

According to previous information, Plaud has sold over 1 million NotePin units globally since 2023, each priced at $159. Coupled with its $99 annual AI transcription subscription service, Plaud's annual revenue reaches $250 million, maintaining a gross margin of nearly 25%, comparable to tech giants like Apple.

BYD's sulfide solid-state batteries are expected to enter small-batch production next year.

According to information obtained by CLS from BYD's investor relations department, BYD is exploring multiple approaches in the field of solid-state batteries, with sulfide solid-state batteries as an important technological direction.

The report states that BYD's sulfide solid-state batteries have achieved breakthroughs in areas such as battery life and fast charging, and are expected to achieve small-batch production in 2027.

In the field of sodium batteries, we are already in the development stage of the third-generation product technology platform and have developed sodium battery products that can withstand 10,000 cycles. The mass production timeline will depend on the actual market and customer demand.

Xiaohongshu is developing AI products for video editing.

According to the Science and Technology Innovation Board Daily, Xiaohongshu's technology team is developing a video editing AI product called OpenStoryline (version 1.0.0), which is still in testing and may be open-sourced later.

It is reported that OpenStoryline can be edited in a dialogue style by uploading a few pictures, and the video has a storyboard and independent segments. It also has some content interpretation and can be accompanied by text and background music. The overall effect is similar to that of Little Lark from Jimo and Seko 2.0 from SenseTime.

According to blogger "天天开柒", because Xiaohongshu did not train a base model with trillions of parameters, the entire OpenStoryline platform uses the DeepSeek open-source large language model and the Qwen3 series of open-source multimodal models with 8 billion parameters.

Lei Jun's pre-New Year's Eve live stream

Xiaomi founder Lei Jun announced that he will hold a live broadcast at 8 pm tonight to celebrate the eve of the Lunar New Year.

Lei Jun stated that this live stream will take viewers on a tour of the Xiaomi Auto Custom Service Center. It's worth noting that this is Lei Jun's fifth live stream in 2026.

In the first four live streams this year, Lei Jun addressed numerous questions and rumors circulating online:

• In response to the "small print marketing" controversy, Lei Jun stated that using small print for labeling/explanation is indeed a common industry practice, primarily for legal compliance. He emphasized at the time that this is also a bad habit in the industry and needs to be changed immediately.
• In response to "taking things out of context": "Everyone is maliciously attacked, and then their words are taken out of context to smear them. It's very difficult to speak online. And I believe that taking things out of context and spreading rumors to smear others is not just limited to Xiaomi; many of our competitors are doing it too."
• Xiaomi has put more than 100 self-built laboratories into operation in four cities: Beijing, Shanghai, Nanjing, and Wuhan. At the same time, Xiaomi cars are also being tested in four extreme environments: Turpan (high temperature), Heihe (high altitude and cold), Kunlun Mountains (high plateau), and Hainan (high temperature/high humidity).
• In response to the incident of Xiaomi contacting an unsuitable KOL: The final resolution of the incident was not because the KOL criticized Xiaomi or offended Mi Fans, but because he had been attacking, belittling, and cursing Xiaomi users, "which we certainly could not tolerate."

SF Express launches a 200 million yuan spring revenue-boosting plan: average income to increase by 500 yuan.

According to SF Express, it has officially launched a "Spring Income Increase" special plan for frontline employees nationwide, investing 200 million yuan to help all couriers increase their income, with an average increase of about 500 yuan.

It is understood that the first phase of this income increase plan will last for 3 months and will cover all frontline delivery workers nationwide. The increased income will be directly distributed in the form of subsidies calculated per delivery. This is also the first time in the industry that a nationwide special plan to increase the income of delivery workers has been launched.

SF Express stated that this is not only a heartwarming reward for the hard work of its couriers during the peak season for Spring Festival deliveries, but also an important measure to care for its employees during the Spring Festival. At the same time, it is also an important measure to respond to the national call for stabilizing employment and retaining workers, and to increase the income of couriers and enhance the sense of gain and happiness of frontline workers through special incentives, in response to the traditional off-season characteristics of the express delivery industry in the spring.

In the future, SF Express will continue to launch income-increasing plans for frontline couriers and continuously iterate and adjust existing plans to gradually form a stable and long-term income-increasing incentive mechanism, so as to continuously promote the optimization and upgrading of customer service quality through the increase in employee income.

Anthropic CEO: Parental control over minors' use of AI is the right thing to do.

In a recent interview with ABC News, Anthropic founder and president Daniela Amodie stated that parental intervention in the use of AI by minors under the age of 18 is a very correct first step.

Amodai said that many parents, like herself (Amodai has two children), have mixed feelings about AI – on the one hand, they think AI can bring their children a healthier and happier life, but on the other hand, they also need to worry about their mental development.

He added that, from a societal perspective, AI still has a lot of work to do to ensure that humans can develop this technology thoughtfully and safely.

Amodé mentioned that minors' minds are still developing. Therefore, children under 18 should likely have their ability to interact directly with AI models limited or controlled by parents, and this is a good start.

In addition, Amodé stated in the interview that AI can perform extremely limited tasks without relying on humans. She said that even the tasks that humans excel at and that have the highest cognitive difficulty are better suited to be completed jointly by humans and AI.

They believe that collaboration between humans and AI can create more meaningful, challenging, and engaging work, while also leading to higher productivity and expanding opportunities for more people. Furthermore, future employment may not necessarily require a strong technical background.

Preview of the cheapest MacBook: 12.9-inch non-notch screen

According to 9to5Mac, citing Bloomberg reporter Mark Gurman and analyst Ming-Chi Kuo, the rumored "cheapest MacBook" is expected to launch in the first half of 2026.

In terms of positioning, this MacBook will be directly named "MacBook" and will be an entry-level product, targeting users who do not have high performance requirements and value portability and budget. For this purpose, the MacBook will use the same A18 Pro processor as the iPhone.

Regarding the screen, the entry-level MacBook will feature a 12.9-inch display, slightly smaller than the 13.6-inch MacBook Air, but there's a chance it will ditch the current notch design. Notably, the MacBook will be available in four colors: blue, pink, yellow, and silver, similar to the entry-level iPad.

In terms of pricing, the new machine is expected to start at $699 (approximately RMB 4,838), lower than the MacBook Air's starting price of $999 (RMB 7,999 in China).

Alibaba's next-generation model, Qwen 3.5, has been revealed.

Yesterday, Linux Do platform user "BunnHack" pointed out:

According to the latest updates from the Hugging Face Transformers codebase, a PR titled "Adding Support for Qwen3.5" has been submitted, officially paving the way for the upcoming Qwen 3.5 series models.

According to the new content in the PR, the Qwen 3.5 series will be a large family, covering plain text, multimodal vision, and hybrid expert models (MoE) .

• Multimodal vision capabilities: Includes Qwen3_5VisionModel and Qwen3_5ForConditionalGeneration, and is equipped with an Auto Image Processor.
• MoE Architecture: The code explicitly lists Qwen3_5Moe and Qwen3_5MoeText, confirming that Qwen 3.5 will continue to use Moe.

Based on the above, BunnHack speculates that Alibaba's next-generation Qwen 3.5 platform model is about to be released.

According to CBN, Qwen3.5 adopts a brand-new hybrid attention mechanism and is very likely a native VLM-like model capable of visual understanding. Qwen3.5 may open-source at least a 2B dense model and a 35B-A3B MoE model.

The Lantu Taishan Black Warrior Edition will be released in March.

On February 9th, Voyah Automobile officially announced in a video released on its official social media platform that the Voyah Taishan Black Warrior Edition will be launched in March this year, adopting a limited-sale strategy.

In the official video, Hu Jun, a close friend of the Voyah brand, appeared and, with his signature snap of his fingers, the Voyah Taishan Black Warrior Edition made a stunning appearance, showcasing its continuation of the iconic front face design of the Voyah Taishan and its high-end driver assistance features such as LiDAR and intelligent driving status indicator lights.

According to Shao Mingfeng, CBO of Voyah Automobile, the Voyah Taishan Black Warrior Edition features a full black finish and is equipped with exclusive exterior kits. The combination of elements such as high-gloss black grille, waterfall grille, exclusive 22-inch black forged wheels and red brake calipers creates a low-key yet powerful overall style.

The interior offers exclusive options that emphasize individuality and sophistication, featuring exclusive perforated textures on the door panels and seats, hand-embroidered headrests, and exclusive starry sky headliner and other details.

In the official video released this time, the Voyah Taishan Black Warrior Edition showcases a cool black front face, with a continuous light strip and a light waterfall grille featuring an illuminated brand logo, giving it a powerful and highly recognizable overall look.

The Legion Y700 5th generation features a 165Hz refresh rate.

Yesterday, Lenovo Legion released more information about the fifth generation of its AI tablet, the Y700.

The new machine will be equipped with an 8.8-inch gaming screen with a 165Hz ultra-high refresh rate and a 2640Hz instantaneous sampling rate (480Hz for multi-finger sampling), a resolution of 3040 × 1904, 408 PPI, and 800 nits of global brightness.

The Legion team also stated that the 165Hz ultra-high refresh rate is to eliminate screen jitter caused by screen movement, so that users will have a smoother and more comfortable experience when performing actions such as turning their heads in FPS games or dragging maps in MOBA games.

In addition, an ultra-high refresh rate can make the edges of moving objects smoother, and the display engine scans the screen in a shorter period of time.

It is reported that the Legion Y700 5th generation will be equipped with a rear RGB light and is expected to be powered by the Qualcomm Snapdragon 8 Ultra mobile platform, offering up to 24GB+1TB of memory.

The once-popular internet sensation "Zihaiguo" has filed for bankruptcy investigation.

According to Qichacha, a business information platform, Hangzhou Jinlingyang Enterprise Management Consulting Co., Ltd., an affiliate of Zihaiguo, recently had a new bankruptcy review case filed by Ma, with the court handling the case being the Yuhang District People's Court of Hangzhou.

Information shows that Zihaiguo was founded in 2018 and is a self-heating instant hot pot brand. Its products include spicy beef self-heating hot pot, self-heating soup pot, self-heating dry pot, and self-heating cooking pot.

According to CBN, "Zihaiguo" was once a leading brand in the popular self-heating hot pot industry, founded in 2018 by Cai Hongliang, the founder of Baicaowei. Cai Hongliang previously founded the snack brand "Baicaowei," which he sold to "Haoxiangni" in 2016, cashing out 960 million yuan and leaving the company. "Zihaiguo" was his second high-profile entrepreneurial project.

To date, Hangzhou Jinlingyang has six records of being subject to enforcement, involving more than 140 million yuan, and Cai Hongliang has also been restricted from high-level consumption.

With its precise positioning as a "single-person meal" and "lazy economy" and heavy advertising and marketing, "Zihaiguo" quickly became a dark horse in the market at the time.

In 2020, with a surge in demand for home cooking, the sales of Zihaiguo (a self-heating hot pot brand) skyrocketed, making it a phenomenal internet sensation. It not only set a record of selling 5 million units in 10 minutes, but also achieved over 100 million yuan in sales on its Tmall flagship store in just 21 minutes during the Double Eleven shopping festival that year.

However, as the demand for "staying at home" decreased sharply after 2022, the popularity of "Zihaiguo" (a type of hot pot) gradually weakened. In March 2023, Lotus Health, a listed company known as the "MSG King," intended to acquire a portion of the equity of Hangzhou Jinlingyang, but the deal fell through five months later.

12306 responds to "passengers without standing tickets being asked to leave quiet carriages"

According to Jimu News, a netizen recently posted that he bought a standing ticket for a high-speed train, but because the regular carriages were crowded, he stood in the quiet carriage. Although he didn't make a sound, staff still reminded him that standing passengers are not allowed to stay in the quiet carriage.

The incident sparked heated discussions among netizens. One netizen commented, "Are people not allowed to stand in second-class seats? I'm willing to stand for a ticket, but I need to be given a place to stand." In response, another netizen explained, "It's not that people aren't allowed to stand in second-class seats, it's that people aren't allowed to stand in the quiet carriages. Tickets without assigned seating aren't sold in the quiet carriages anyway."

According to reports, the netizen who posted the content said that they bought a standing ticket for another carriage, but the carriage they were in was too crowded, so they went to a quiet carriage, and then a staff member came over to remind them. "But it may also be a matter of how strict the supervision is."

Regarding the above situation, the 12306 operator stated that when passengers purchase tickets, the system will display the specific carriage in which they are booking a standing ticket, and standing tickets are not issued for quiet carriages.

Another staff member from the 12306 platform said that passengers without seats are advised to avoid riding in quiet carriages, but there is no hard and fast rule and the situation on site shall prevail.

Dingdong Maicai: Everything is running smoothly.

On February 5, Meituan announced on the Hong Kong Stock Exchange that it would complete the acquisition of 100% equity of Dingdong Maicai's China business for an initial consideration of approximately US$717 million.

During the transition period, Dingdong Maicai will continue to operate according to the pre-transaction model.

Yesterday, Dingdong Maicai responded, saying, "Please rest assured, everything at Dingdong Maicai is operating smoothly and steadily."

Dingdong Maicai stated that all of its business partners, including product development, delivery personnel, and quality control teams, have remained steadfast in their posts as always.

Dingdong Maicai, founded in 2017, is a leading fresh food instant retail platform in China, with its main feature being "delivery in as little as 29 minutes".

The company went public on the NYSE in 2021. In the third quarter of 2025, Dingdong Maicai achieved revenue of RMB 6.66 billion, a record high for a single quarter; net profit was RMB 80 million, marking the seventh consecutive quarter of profitability under GAAP standards.

The first trailer for "Minions 3" has been released.

(Universal Film)

Yesterday, the first trailer for the Despicable Me spin-off film, Minions 3: The Minions and the Monsters, was released, and it was announced that the film will soon be released in theaters worldwide.

The film tells the story of Minions who venture into Hollywood to make a monster movie, but accidentally summon real monsters, causing chaos and adventure.

"Minions 3," directed by Pierre Coffin, written by Brian Lynch, and starring Pierre Coffin, will be released in North America on July 1.

"The Beaver Transformation Project" is scheduled for release in March.

Pixar's official Weibo account announced yesterday that its new original animated film, "The Beaver Project," will be released in theaters nationwide on March 20.

The film tells the story of a young girl named Mabel who, in order to explore the animal world, uploads her consciousness into the body of a robotic beaver. Later, she befriends King George of the beaver kingdom, and together with the animals, they thwart a real estate developer's scheme.

The film is directed by Daniel Chung, written by Daniel Chung and Jesse Andrews, and features the voices of Pepper Kodak, Jon Hamm, and Bobby Moynihan. It will be released in the United States on March 6, 2026.

The pre-sale box office for new films scheduled for the 2026 Spring Festival has exceeded 80 million yuan.

According to Lighthouse Pro, as of 10:48 PM yesterday (February 9th), the total pre-sale box office for new films scheduled for the 2026 Spring Festival has exceeded 80 million yuan, with "Pegasus 3", "Silent Night", and "The Swordsman: The Wind Rises" currently ranking in the top three of the pre-sale box office chart.

It is reported that six films scheduled for the Spring Festival have already started pre-sales, including "Pegasus 3", "Silent Night", "The Swordsman: The Wind Rises in the Desert", "Boonie Bears: Every Year", "Panda Project: Tribal Adventure" and "Star River Dream".

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

ifanr | Original Link · View Comments · Sina Weibo

Who would have thought that the first hot-selling device of 2026 would be the Mac Mini, which was propelled to new heights by OpenClaw—a product that had always been lukewarm in Apple's lineup, but was suddenly brought up to prominence.

The versatility of OpenClaw needs no further explanation, as it has been discussed extensively recently. However, the combination of OpenClaw and Mac Mini raises a perplexing question: if an agent like OpenClaw can enable computers to automatically reply to emails, manage schedules, and control everything, why doesn't Apple, with Siri and its vast ecosystem, create its own ?

With its hardware, ecosystem, and "out-of-the-box" capabilities, Apple could easily launch an official "Super Siri," perhaps even charging a few hundred dollars more per device, like AppleCare. Users would flock to it.

This is also the view of Jake Quist, a partner at Y Combinator: if Apple possesses an AI agent layer, they can build the most formidable moat in the tech world; the more an agent knows about its users, the smarter it becomes. Apple already possesses almost all of its users' data, applications, and even devices. They could have built an agent that could seamlessly operate across iPhones, Macs, iPads, and Apple Watches—something no other company could do.

But what if Apple actually elevated Siri to the S-level strategic importance and quickly created something similar to OpenClaw?

They will receive lawyer's letters from Microsoft and Meta the next day.

You want Siri to post an Instagram story? Sure, that's like having the material delivered right to your door. Instagram's terms of service explicitly prohibit automated operations.

Want Siri to post a job application on LinkedIn? Here we go! LinkedIn is one of the world's strictest platforms against automation, with Article 8.2 of its User Agreement explicitly prohibiting a range of automated operations.

"User agreements" are often just gentlemen's agreements, especially when users are personally involved in web scraping and automated operations, which fall into a gray area. Platforms usually turn a blind eye. For individuals running agents, the platform might at most ban the account, but it's difficult to sue every user. Going too far would be unsightly: LinkedIn once had a lawsuit in 2019 against a small data analytics company, hiQ, which sued LinkedIn for blocking its web scraping tools. The case went all the way to the Supreme Court, but hiQ ultimately lost, and the two parties reached a settlement.

But if Apple does this as a company, the situation is different. If Apple builds an agent into iOS/macOS, it's equivalent to Apple, as a commercial entity, systematically bypassing the platform's API firewall. This clearly constitutes "unfair competition" and "interference with business relations." Go ahead and sue; they'll just stay silent.

But what if a user buys a Mac Mini and runs open-source OpenClaw code on it? Then whatever they do is none of Apple's business. Apple is simply selling computer hardware; what the user runs on it—whether it's editing videos or "stripping ads through automated scripts"—that's the user's own doing.

Everyone thought Apple had missed a golden opportunity, watching the personnel shake-up in their AI department and mocking their strategic slackness. But seeing OpenClaw boost Mac Mini sales, Apple has a new path to choose from: hiding its strength behind the scenes.

Apple's "past offenses"

Apple's lukewarm approach in recent years has made people forget its past achievements. In 2021, Apple introduced App Tracking Transparency (ATT), allowing users to choose not to share IDFA tags starting with iOS 14. From then on, a pop-up window would appear when opening an app, asking if users allowed app tracking. The result was that 90% of users clicked "decline."

IDFA is a unique identifier used by advertisers when targeting ads on third-party platforms like Facebook. With its cut off, Meta can no longer track whether users made a purchase after seeing an ad. This makes Meta's ad targeting extremely inaccurate, leading advertisers to cut their budgets on Facebook and Instagram.

This move directly resulted in Meta losing over $10 billion in revenue that year, and its stock price plummeted. Chief Financial Officer David Wehner explicitly stated in the earnings call that it was related to AT&T, which at the time accounted for 8% of Facebook's annual revenue. The stock price fell by 26%, wiping out approximately $232 billion in market capitalization. This set a record at the time for the largest single-day market capitalization loss in US stock market history.

Meta was the most affected at the time because it was the most dependent on IDFA. However, other tech giants were also affected to varying degrees; for example, the combined revenue loss for Snapchat, Twitter, and YouTube was around $3.2 billion.

ATT created a time lag for Apple's own advertising business. According to data from the marketing platform Branch cited by the Financial Times, six months after the new policy was implemented, Apple Search Ads' market share tripled, accounting for half of iOS app install ads.

Meanwhile, Wall Street investment bank Evercore ISI predicts that this strategy will help Apple's advertising revenue reach $30 billion by 2026—most of which originally belonged to Meta and Google.

You might ask, isn't this simply because Facebook itself is no longer popular, so it suffers terribly when its control is restricted, while TikTok remains unaffected? Nothing lasts forever; this is merely an inevitable result of the changing landscape of social media platforms.

The problem is that, according to this logic, Facebook's Android and OS systems should have experienced simultaneous shocks. However, six months after the AT&T regulations took effect, advertising spending on Android devices did not drop significantly. Looking at it from another perspective, Snapchat, which also targets young users, released its financial report after the AT&T regulations were introduced. Not only did it see a 42% year-on-year revenue increase in the fourth quarter of that year, reaching $1.3 billion, with daily active users reaching 319 million, but it also added 54 million new users, a 20% increase compared to the same period of the previous year.

All of this is because Snapchat began preparations a year ago, helping advertisers switch to Snap Advanced Conversions, or Apple SKAdnetwork—Apple's own ad tracking solution.

"Respecting user privacy" is such an impeccable statement, occupying the moral high ground and allowing Apple to strike at the foundation of its competitors without any psychological burden, without shedding a drop of blood.

Apple's open conspiracy

Essentially, Openclaw's reliance on "MacMini + local AI" will face the same problem as the pre-AT&T era advertising model: it depends on an infrastructure from Apple.

If users, by building and using services like OpenClaw, allow AI to filter ads, bypass platforms, and accomplish their desired tasks, while Apple not only avoids losses but also sells more hardware, then they will naturally be happy. This perfectly aligns with Apple's consistent strategy of combating "data capitalism."

The groundwork for this began long ago. In the last two years, Apple did something very cunning with the Mac: it promoted a unified memory architecture and made memory extremely expensive.

Before the AI ​​era, ordinary users didn't necessarily need 128GB or 256GB of RAM, and Apple's "price-cutting" strategy was always criticized. But times have changed. Large LLM (Limited Module Model) is extremely demanding on VRAM/RAM. Apple's M-series chips are currently the best hardware on the market capable of running large-parameter models at a consumer-grade price (compared to NVIDIA H100), without the complexity of a cloud environment, offering only the convenience of quickly accessing various folders and code repositories.

To complement its hardware, Apple released the MLX framework (a tool specifically designed to enable open-source AI to run efficiently on Apple chips). This isn't just a gesture of goodwill; it's a blatant invitation to developers: "Stop buying Nvidia, buy Mac Studio instead. We're the best place to run Llama 3 and OpenClaw."

However, providing "hardware that can run AI" is only the first step. Apple's greatest strength lies in its risk isolation—which is reflected in its completely different management scales for iOS and macOS.

The iOS ecosystem is a walled garden that is well protected. If OpenClaw were to appear as an app today, Apple would absolutely not allow it. The reasons would be nothing more than "undermining system security" or "unauthorized automation." In short, Apple would not allow something that could lead to antitrust lawsuits and privacy disasters to be listed on the App Store.

However, Apple has always maintained the "universal computer" attribute of macOS. You can install any unsigned software on your Mac (by granting it security permissions yourself) and run any open-source code. If problems arise, such as AI randomly sending messages or placing orders, Apple can simply shrug and say, "This is the user's personal behavior on an open system, and it has nothing to do with us."

This is a perfect form of isolation, much cleaner than stepping in as an agent that could easily spiral out of control and cause problems.

This divide-and-conquer approach to openness has become Apple's deepest moat. Only locally can users relinquish that most crucial permission—Root access.

It's been a while since OpenClaw was released, and an increasing number of mishaps have occurred, including but not limited to API overspending, AI helping users apply for loans, and the deletion of important local files. These mishaps further highlight the importance of a Mac Mini —it provides a physically isolated, fully user-controlled security sandbox . Users are willing to grant OpenClaw root access on their Mac Mini because the code is open-source, transparent, and runs on their own desk in their study, not on some unknown cloud server.

In this dimension, Apple has already gained an advantage in the AI ​​race: they may not have the smartest models, but they already have hardware that carries the trust of users.

The back of the apple is hidden

Right now, what might really make Apple hesitate is another giant: Google.

Google pays Apple approximately $20 billion annually just to be the default search engine for Safari. If the Mac Mini + AI truly allows users to skip the "search" step altogether, causing a sharp drop in Google's search traffic, Apple's $20 billion "protection fee" will be in jeopardy.

Whether to sell their own products to earn the money back, or continue to sit back and collect $20 billion, is truly a sweet dilemma.

The existence of this money precisely reveals the transition between two eras. The default search engine is the imprint of the old era on various internet giants. Google wants you to search manually and see the results list so it can display ads. LinkedIn requires you to log in and endure a cumbersome interface before it can show you job postings from sponsors. Instagram requires you to scroll down your feed so it can insert sponsored links between your friends' photos.

In the previous era, platforms used every means possible to keep users within their own sphere of influence, capture attention, and then monetize it. You could say this business model was inefficient, but it was also effective, supporting the entire golden age from Web 2.0 to the mobile internet.

But now you suddenly say, no, no need. From now on, my agents will handle everything for me: browsing news feeds, posting job ads, and searching for answers. They're efficient, ruthless, and get straight to the point; they won't be swayed by short videos. I just need to choose a good piece of hardware for them.

In a series of closed business loops, "people" disappear.

Advertisers want to buy human attention and exploit human weaknesses, not AI crawlers. If half of Instagram's traffic turns into clicks from agents, what's the difference between that and buying fake traffic? Are you kidding me?

To this day, advertising remains an unshakeable revenue pillar for internet businesses. In 2023, Alphabet, Google's parent company, had total revenue of $307.4 billion, of which advertising revenue (Search + YouTube + Network) was $237.8 billion, accounting for 77%. In the same year, Meta's total revenue was $134.9 billion, of which advertising accounted for $131.9 billion, a staggering 97.8%.

Even OpenAI, with its 700 million monthly active users, has come up with a way to monetize, but isn't it still advertising?

But children make choices; adults want it all. Apple wants to reap the benefits of being a giant in the old era while also being a shovel-maker in the new. As for the future, whether Google continues to dominate the advertising world or local AI completely reshapes how we access information—it doesn't matter.

Tactically defend, strategically prepare—this is the essence of concealing one's blade, so that it can be drawn at any time.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

ifanr | Original Link · View Comments · Sina Weibo

The hardware industry had a tough 2025, and will undoubtedly have a tough 2026 as well.

The RTX 60 series graphics cards, which were highly anticipated by gamers, have been met with disheartening news: their release date has been significantly delayed until the end of 2027.

According to the latest disclosure by The Information, Nvidia has decided to shift its focus entirely to AI products in 2026, and the RTX 60 series gaming cards have completely gone from being a hot commodity in the "pre-AI era" financial reports to being sidelined.

▲ Image｜Tom's Guide

Worse still, this means that the graphics card market will face a substantial update vacuum for the next two years.

Old Huang's resource balance no longer tilted, but rather tipped over—

To appease most gamers, Nvidia released two major updates, DLSS 4 and 4.5.

Compared to DLSS 3, DLSS 4.5 unprecedentedly supports the entire range of RTX graphics cards, which is essentially a way to "extend the lifespan" of 20-series and 30-series graphics cards.

▲ Image｜Nvidia

However, DLSS 4.5 is not a charity, and moreover, some older graphics cards will experience a certain degree of frame rate reduction after using DLSS 4.5.

With both hardware upgrades and production capacity hitting a wall, Huang could only rely on software algorithms to keep things afloat—the so-called "backward compatibility" seemed more like a fig leaf to cover up the hardware stagnation.

Especially considering the recent news that the RTX 3060 would be put back into production , the entire DLSS 4 seems to have been "premeditated":

Don't get too excited when you find that a five-year-old graphics card can still run the latest features. It means that the hardware has stagnated, but you'll have to pay a higher price for an upgrade.

▲ Image｜The Information

Besides the graphics card delays, another shocking piece of news came from Intel CEO Lip-Bu Tan. Following his comments on memory price increases last year, the godfather of the semiconductor industry dropped another bombshell at the beginning of the year.

At the Cisco AI Summit on February 3, Chen Liwu clearly stated:

The current round of memory price increases will not end until at least 2028, and there are currently no signs of easing in the market.

▲ Photo｜Bloomberg

Chen Liwu, who led Cadence for many years, has extremely solid industry knowledge and is deeply involved in the chip manufacturing field. His predictions often represent the underlying trends of the entire semiconductor industry.

Chen Liwu bluntly pointed out that AI is draining all storage.

This has led to an extremely absurd situation:

Memory was originally a component of terminal products, and it only cared about its own rapid growth. However, in 2026, from mobile phones to computers, they all faced a "stagflation" situation with soaring costs, declining configurations, and shrinking profits.

The performance improvements promised by Moore's Law have already been offset as chip manufacturing processes approach the 1nm limit.

Now, memory chips, which were originally a supporting player, are beginning to threaten to shatter Moore's Law.

▲ Photo｜The Verge

Let's first clarify the logic—

Modern large-scale AI chips (such as Nvidia's Rubin platform) have an insane thirst for high-bandwidth memory (HBM).

Samsung, SK Hynix, Micron, and other major memory manufacturers have invested almost all of their R&D and production resources in HBM memory in pursuit of the extremely high profits in the AI ​​industry.

This has directly led to a squeeze on the production capacity of civilian DDR4/5 and LPDDR series, and the procurement costs of basic components for mobile phone and computer manufacturers are visibly spiraling out of control.

▲ Photo｜The Irish Times

To make matters worse, chip manufacturing processes that are approaching physical limits will also face challenges in 2026.

Not only did TSMC's first-generation N2 process not show significant performance and energy efficiency improvements compared to the more mature 3nm process, but the advanced N2P process traded nearly 20% of the cost for a 5% performance improvement.

Since 2nm chips combined with expensive memory would make end products prohibitively expensive, manufacturers have no choice but to take a conservative approach.

In 2026 we will see many products that have been repeatedly "refined" using the 3nm process.

▲ Photo｜TechSpot

This is the truth behind Moore's Law's failure to predict time.

It's not that the technology can't do it, but because resources are being devoured by AI, manufacturers have no choice but to "stand still" in order to maintain even the most basic profits.

We can speculate that Qualcomm and MediaTek will both release more products this year with "upgraded names and downgraded specifications," such as the Dimensity 9500s released recently.

These new processors sound like flagship iterations, but the architectures that have been disassembled are mostly familiar.

▲ Image｜MediaTek

In this context, the "refinement" strategy for mobile phones, tablets, laptops, and even PCs in 2026 will undoubtedly become an industry consensus.

Consumers have only two choices when buying new products: either minor improvements and a small price increase , or a major upgrade and a significant price increase .

For Qualcomm, 2026 will be an especially tough year.

Not only does it have to compete with Apple for N2P process production capacity, but it also needs to push its X2 Elite processor into the laptop market, putting both of its supply chains in a state of crisis.

▲ Image｜Stuff

In short, as ordinary users, we need to realize that:

The entire digital industry is entering a technological "long holiday," during which new products no longer represent performance improvements, but rather peripheral premiums.

Especially if you plan to build your own computer in 2026, you will likely be disappointed to find that apart from being more expensive and having less memory, the actual experience of the whole machine has not made any qualitative leap.

▲ Image｜YouTube @Toasty Bros

This is the "technology tithe" that everyone will have to pay in 2026.

In order for major companies to establish a foothold in the AI ​​field, the basic experience of ordinary consumers has become the most vulnerable area to be exploited .

On the other hand, the disappearance of the hardware benefits directly brought by SoC has actually made digital products more financial.

For example, the last month or two may be the last time for the PS5 Pro to be offered as a "free console with hard drive purchase".

▲ Image｜Eurogamer

The consumer electronics market in 2026 is like a marathon without an end. Manufacturers are waiting for the storage cycle to end, and we are waiting for the performance explosion to return.

Looking at the beginning of the year, with prices rising, performance stagnating, and costs soaring, every purchase needs to be made with more caution than ever before.

After all, even Nvidia's graphics cards have been delayed, so your wallet should wait too.

▲ Photo｜Financial Times

Perhaps the most rational choice in 2026 is to "pat and mend for another three years," and the 3nm products that have been repeatedly refined are not unusable.

All we need to do is adjust our expectations from "always making progress" to "stagflation is an objective reality".

Until 2nm technology truly matures and memory prices return to a rational level, we don't need to pay for expensive mediocrity.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

ifanr | Original Link · View Comments · Sina Weibo

Of all the major sporting events, the Winter Olympics is probably the most "aloof".

This aloofness stems not only from the fact that Milan, the host city, is nearly 10,000 kilometers away from us, and only a few people can experience the thrill of speeding on ice; even fewer athletes actually give their all on the ice; and even the very nature of "ice and snow sports" itself means that half the world's population will never have the opportunity to experience them.

Most viewers can only watch those glorious moments of victory through the screen; their imagination runs wild, rather than their sweat and heart rate.

The intervention of digital technology is eliminating the barriers caused by physical distance.

Lin Hsiao-jun, the world champion in short track speed skating, recently made his "debut" on the snowboard big air platform, which went viral on social media and quickly became a trend among netizens and KOLs.

In the video, the athlete performs a textbook 1440-degree spin dozens of meters in the air. It's hard to imagine that this fluid movement can be achieved with just a few taps of the fingertips.

Behind this seemingly "unexpected" crossover lies a small technological shock brought to the Winter Olympics by Alibaba Cloud AI.

As the official cloud service partner of the Olympic Games, Alibaba Cloud has truly lowered the barrier to entry for the Winter Olympics to everyone – with just a little "imagination" and a Qianwen App, those dazzling highlights can instantly become your exclusive memories.

No kidding, this might be the closest I've ever been to the Winter Olympics.

In the past, to generate a usable AI face-swapping video, one not only had to input a long string of detailed prompts, but if one wanted to generate intense sports scenes such as short track speed skating and snowboard jumping, the AI ​​was also very prone to "crashing," either growing extra arms and legs or violating the rules of physics. It was inevitable to generate multiple "card draws" to get the desired effect, making it only a toy for a few geeks and enthusiasts.

But with Alibaba Cloud AI, shining on the ice and snow track like Lin Xiaojun is not that complicated at all.

Open the Qianwen App, select "AI Video" at the top of the chat box, and you can see Lin Hsiao-jun's highlights in snowboard big air or short track speed skating. Click "Make the Same" to upload your own photo, wait 3-5 minutes, and your exclusive Winter Olympics moment will be generated.

A little-known fact: Despite his busy schedule, Elon Musk has been a fan of winter sports since childhood. Skiing is one of his few vacation activities, and the outdoor skiing brand RENOUN even customized a Tesla red ski for him.

Now that Lao Ma is getting older and has more things to do, it may be difficult for him to have the opportunity to ski on the slopes again. No problem, we can use AI to send him to the Milan competition venue to challenge the extremely difficult ski jumping.

Just like in the business world, he has a bright and confident look, with a subtle smile playing on his lips. Alibaba Cloud has captured Musk's subtle expressions perfectly, not only in appearance but also in spirit.

Besides skiing on the high platform, we can also have Lao Ma switch to ice blades and compete against Lin Xiaojun on the short track speed skating track. The consistency of the characters and the overall feel of the video are quite natural.

For a national event like the Winter Olympics, participation shouldn't be confined to the screen. In addition to creating digital sub-venues in the Qianwen App, Alibaba Cloud also brought the Winter Olympics experience to Hangzhou Xiaoshan Airport and even "invited" Tina & Milo, the mascots of the Milan Winter Olympics, to host an AI interactive space that "invited everyone to become athletes."

Here, visitors can not only take photos with the mascots and feel the strong atmosphere of the Winter Olympics, but also participate in AI Winter Olympics interaction on-site from February 9th to 11th to generate their own highlight moments.

No professional equipment or complicated operations are required. During the short time you are waiting for your flight, you can immerse yourself in a Winter Olympics moment that belongs to ordinary people, turning the event from a distant live broadcast into a real and tangible participatory experience during your journey.

How did the Olympics become a new battleground for AI?

This is not the first time the Olympics and AI have collided.

As early as the Paris Olympics two years ago, Alibaba Cloud's AI-enhanced "360-degree real-time replay technology" was used in the television broadcasts. Viewers watched the games like watching a movie, and could see the slow-motion and time-freezing effects of the athletes' movements from multiple angles in an immersive way.

Although the effect was stunning, the audience was still only "watching," which simply amplified the visual tension of the sports competition in the traditional viewing mode.

With the development of image-generating models like Nano Banana last year, we can easily put ourselves into the competition venue, and even generate photos with sports stars such as Gu Ailing and Su Yiming, or directly swap faces, making the sense of participation even greater than before, but it still lacks the thrill of "personally participating in the competition".

Alibaba Cloud has directly integrated a cyberpunk "big air" and "speed skating rink" into the Qianwen App. There are no rules or referees here, and everyone can become a sports star speeding on the track.

In fact, becoming a "champion" in the digital world is not so simple. Whether it is short track speed skating or snowboard big air, the speed of movement is extremely fast and the range of movements is large, making it easy for AI models to fail.

To bring millions of users to the "arena," Alibaba Cloud has deployed a powerful combination of industrial-grade technologies.

After receiving photos uploaded by users, Qianwen App will first perform "refinement" by completing facial recognition, posture adjustment, and lighting optimization according to the competition environment to match the competition environment.

Next, Alibaba Cloud's WAN 2.2 VACE large-scale model took over the "camera movement".

If you simply paste the user's face onto the screen, the face-swapping video will look quite disjointed. WAN 2.2 can achieve natural and coherent dynamic synthesis under drastic changes in posture and various lighting conditions. Even the reflections on the ice surface and the splashing snow particles will be automatically adjusted according to your position, making the synthesized video look seamless.

The entire budgeting process is completed collaboratively by the cloud and the application, and what is ultimately presented to the user is the Qianwen App's minimalist, instant-get experience: with just three taps and uploading a photo, a video with a pretty good effect can be generated, allowing the Olympic spirit to truly achieve "everyone's participation".

AI enables everyone to be "higher, faster, stronger".

The essence of technology is often to bridge distances. Telescopes allow us to see further, cars allow us to get closer, and generative AI is eliminating the invisible distance between people and the competition arena.
The Olympics have always carried a sense of contradiction. In name, it belongs to all the people of the world, but the vast majority of people are far removed from the competition field, making it an extremely centralized narrative field.

That distance is not just a geographical separation of thousands of miles, but also an insurmountable barrier built by sweat and talent, ensuring that the most essential experience always belongs to only a few, and that the spotlight only shines on the 0.1% of people on the podium, while the remaining 99.9% can only contribute their cheers and viewership ratings from behind a screen.

This slightly "looking up" barrier has never been truly broken down, and for a long time it has even made people feel a little tired: you can see the glory, but you can't touch it.
This attempt by Alibaba Cloud may be a signal that things are starting to really change.

Some might argue that the allure of sports lies in the real physical pain and the triumphs it brings. Isn't generating a highlight video using AI a form of "psychological victory"?

I don't see it that way. This precisely demonstrates the humanistic side of AI technology: it allows elderly people to rediscover the energy of their 20s in the digital world; it allows children from the south who have never seen snow to leap across the Milan stadium.

This is the ideal form of technology in our minds; it doesn't necessarily have to be for efficiency, for growth, or for replacing humans.

Before the opening of the Winter Olympics, International Olympic Committee President Kirsty Coventry announced that thanks to Alibaba Cloud's big data model technology, this Winter Olympics would be the "most intelligent" Olympics in history, and the first official big data model in Olympic history was created based on Alibaba's Qianwen platform.

On the field, the International Olympic Committee launched the "National Olympic Committee AI Assistant" on its staff website, providing instant answers to questions; during broadcasts, a recognition solution based on Qwen-VL operates simultaneously, capturing and labeling key events such as goals and fouls in real time…

Off-field, Alibaba Cloud AI also blurred the invisible boundary between spectators and athletes, allowing spectators who could only cheer from behind a screen to finally step onto their own track and "compete" in another sense.

Technology is no longer just a behind-the-scenes hero; it has become a bridge, bringing the audience from the stands to the center stage. "Public participation" is no longer an empty slogan, but a series of real digital experiences.

In this era where AI is developing at an alarming pace, Alibaba Cloud has shown us another facet of technology:

It can also help every ordinary person like you and me realize that the distance between us and "flying" is not that far.

The Winter Olympics is just the beginning. In the future, those grand events that could only be viewed from afar on screens, whether it be the magnificent moments of manned spaceflight or other global sporting events and cultural celebrations… AI is turning these "experiences" that were once only available to a select few into real and tangible exclusive memories on your phone.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

ifanr | Original Link · View Comments · Sina Weibo