Pogbagram

Priced at approximately $2,000, the foldable iPhone is rumored to be named "iPhone Ultra".

According to Bloomberg, Apple's upcoming foldable iPhone will most likely be named "iPhone Ultra" and priced at around $2,000.

According to Bloomberg reporter Mark Gurman, this device will feature new technologies such as an extra-large inner screen and under-display sensors, and its pricing and specifications will surpass those of the existing iPhone product line.

However, Gurman added that "Ultra" is currently the most likely naming scheme to be adopted in internal discussions at Apple, but the possibility of using a more straightforward name such as "Fold" cannot be ruled out.

Regarding the next generation of AirPods, Gurman said it will be equipped with a vision camera to provide visual intelligence data for Siri, and the price is expected to exceed that of the current AirPods Pro. "AirPods Ultra" is one of the naming options under discussion.

The next-generation high-end MacBook Pro will introduce a touch OLED display. Referring to Apple's past precedent of increasing the starting price of iPads by about 20% after switching to OLED, the new model is expected to be priced higher than the existing M5 Pro and M5 Max models. The naming scheme of "MacBook Ultra" is also under consideration.

At the manufacturing process level, Apple's manufacturing design team is exploring the possibility of 3D printing aluminum, with the goal of improving the production efficiency of Apple Watch casings and even future iPhone bodies. This is an extension of the 3D printing titanium process used in the previous Apple Watch Ultra 3.

Regarding the Mac product line this year, Gurman predicts that the upgraded Mac Studio will be launched in the middle of the year, the iMac will have a new color scheme, and the high-end MacBook with touchscreen and OLED is expected to be released around the end of the year.

Apple's official website has been updated with information on three new executives, covering the design, legal, and interface departments.

Apple recently updated its leadership page on its official website, adding biographies of three new executives and adjusting the titles of existing executives.

• Jennifer Newstead appears on the page as Senior Vice President and General Counsel.

She officially succeeded Katherine Adams on March 1, who had held the position for nine years since 2017.

Newstead previously served as Chief Legal Officer at Meta for six years and is a seasoned legal executive in the technology industry. Meanwhile, Katherine Adams' title has been updated to Senior Vice President and Head of Government Affairs.

• Molly Anderson, VP of Industrial Design, also appears on the updated page.

Anderson joined Apple in 2014 and has been effectively leading the industrial design team since 2024, responsible for the appearance design of all Apple products, accessories, and packaging. She is the latest design head after Jony Ive and Evans Hankey, but Apple has not disclosed the specific date of her formal promotion to vice president.

• The addition of Steve Lemay, VP of Human-Computer Interface Design, has also attracted much attention.

Lemay joined Apple in 1999 and has been with the company for over 25 years. He currently leads the Apple platform software interface design team. He succeeds Alan Dye, who left Apple at the end of last year to lead design work at Meta's Reality Labs division.

• In addition, Apple veteran Eddy Cue's title has been updated from "Senior Vice President of Services" to "Senior Vice President of Services and Health," and his official photo has been changed.

This change stems from the retirement of former Chief Operating Officer Jeff Williams last year, after which Cue took over Apple's health and fitness-related business team. Cue joined Apple in 1989 and has served the company for nearly 37 years, making him one of the most senior members of the current executive team.

Sold at a loss? Apple quietly removes the 512GB memory option from the Mac Studio.

According to AppleInsider, Apple has quietly removed the 512GB memory option from the Mac Studio, which previously cost as much as $14,099 (the top-of-the-line configuration in China, including 512GB of memory and a 16TB SSD, costs 108,749 yuan), with the memory upgrade alone requiring an additional $4,000.

At the same time, Apple also raised the price of existing memory upgrade options. For example, upgrading from 96GB to 256GB of memory now costs $400 more.

Regarding delivery times, the Mac Studio with the highest memory configuration is expected to ship from early to mid-May this year; while the basic configuration can still ship within a few days, and the basic memory version with a larger capacity SSD is expected to ship from the end of March to the beginning of April.

The 512GB version of the Mac Studio uses a single 64GB LPDDR5 memory chip, which is reportedly a custom specification for Apple. Currently, few other manufacturers in the industry can provide products with the same density, and its technical difficulty and cost are at the forefront of the industry.

Furthermore, some analysts have pointed out that this removal from the market may be related to Apple's upcoming M5 Ultra version of the Mac Studio—Apple may be stocking up on high-density memory for the new product. If the M5 Ultra version is released as scheduled, 512GB or even higher memory options may return.

ChatGPT's "Adult Mode" Delayed for the Second Time; OpenAI: Higher Priorities Are More Important

According to TechCrunch, OpenAI has once again delayed the launch of ChatGPT's "adult mode." This feature was originally intended to provide age-verified adult users with access to adult content, including pornography.

This is the second delay for this feature.

In October of last year, OpenAI CEO Sam Altman first announced the plan, stating that in December, as the age verification mechanism was improved, pornographic and other content would be made available to verified adult users, and that this would be part of the principle of "treating adult users like adults".

However, in December of last year, Altman reportedly issued an internal "red alert" memo, requiring all teams to focus on optimizing the core experience of ChatGPT, thus delaying the launch date for the first time until the first quarter of this year.

An OpenAI spokesperson told Axios that the company is "delaying the rollout of the adult mode" because it needs to "focus on work that is currently of higher priority for more users," including improving model intelligence, optimizing the personalized experience, and making ChatGPT "more proactive."

The spokesperson also stated, "We remain committed to the principle of treating adult users as adults, but it will take more time to perfect this experience."

Lei Jun: AI will not eliminate jobs; in the future, two hours of work per day will be enough.

According to China News Weekly, Lei Jun, founder and CEO of Xiaomi Group, said in a recent interview that the AI ​​era means "many rules will be rewritten," but at the same time it will also create a large number of new jobs.

He advised the public to embrace this more advanced era with an open mind and to make bold predictions about future work patterns:

People won't need to work eight hours a day or five days a week; they might only need to work three days a week or two hours a day.

He believes that the overall quality of life and work will be greatly improved at that time.

The creator of OpenClaw refuted rumors about an "official Weibo account": He has never used Weibo.

Yesterday, Peter Steinberger, the developer of "Lobster" OpenClaw, publicly stated on X that the "OpenClaw Official Weibo" account that recently appeared on Weibo is not an official account, and he has never used Weibo.

The incident began when a user posted a question to Peter on X, stating that OpenClaw had opened an official account on Weibo and attaching a link to the account.

After Steinberger stated that the account was unofficial, the blogger "Afternoon Nap" posted on Weibo, "I asked around, and they said the materials were given by the person in charge of their project team in China."

Steinberger then responded, "If it's not under my control, then what kind of official Weibo account is it?"

It is worth noting that this account not only obtained Weibo's blue V certification, but also interacted with the official Weibo accounts of AI products such as GLM and Kimi, and claimed to have been invited by Weibo officials to join.

The first OpenClaw benchmark test results are in: lightweight models outperform flagship models across the board, and domestically developed AI has strongly entered the first tier.

Yesterday, PinchBench, a benchmark test designed to evaluate the performance of large language models in the OpenClaw task, was officially released. It tested 32 mainstream large models at once and compared them horizontally in terms of success rate, speed, and cost.

Related reading: The first The Lobster Model Ranking is Here! Two Chinese AIs Break into the Global Top Three – A Must-Read Before Starting Lobster Farming

• In terms of success rate, Google's Gemini 3 Flash Preview ranks first with a success rate of 95.1%.

As a "lightweight" version of the Gemini series, it outperformed its own flagship Gemini 3 Pro (91.7%), and also surpassed Claude Sonnet 4.5 (92.7%) and GPT-4o (85.2%).

Domestic models also performed impressively, with the MiniMax M2.1 ranking second with a success rate of 93.6%, followed closely by the Kimi K2.5 with 93.4%. These two domestic models together occupy two of the top three spots globally.

Anthropic's flagship model, the Claude Opus 4.6, had a success rate of only 90.6%, ranking seventh and lagging behind many mid-range models.

• In terms of speed, the MiniMax M2.5 completed all the test tasks in 105.96 seconds, leading the second-place Gemini 2.0 Flash by a narrow margin of 0.09 seconds, and won the speed championship.

In comparison, Claude Sonnet 4 took 137.66 seconds, while Gemini 3 Pro took a whopping 239.55 seconds, roughly twice the time of the champion.

• In terms of cost, the GPT-5 Nano is the lowest-priced option at $0.03 per mission, with a success rate of 85.8%.

The Gemini 2.5 Flash Lite followed closely behind with a cost of $0.05 and a success rate of 83.2%. The Claude Opus 4.6, however, cost a whopping $5.89 to complete the test—nearly 200 times the cost of the GPT-5 Nano—yet its success rate was more than 3 percentage points lower than the MiniMax M2.1.

PinchBench's scoring mechanism includes three methods: code execution verification (automated checks), quality assessment (judged by Claude Opus), and a combination of both. All questions and answers are open-source on GitHub. The complete leaderboard can be viewed at pinchbench.com.

Samsung seeks more AI collaborations; its multi-model strategy targets Apple.

According to the Financial Times, Samsung Electronics co-CEO and head of consumer devices, Roh Tae-moon, said in an interview that the company is open to "strategic partnerships" with more AI companies, including OpenAI.

He pointed out that Samsung's research shows that consumers are increasingly using multiple AI services simultaneously, rather than relying on a single platform.

Consumers are not limited to one AI platform; they are using multiple AI models… We remain open to all options… I believe this is the key to Galaxy AI’s appeal to consumers.

At the macro market level, Counterpoint Research predicted last week that global smartphone shipments will decline by 12% year-on-year in 2026, the lowest level since 2013.

With hardware upgrade space increasingly narrowing and the overall market continuing to shrink, the ability to integrate AI functions is becoming a core bargaining chip for major manufacturers to compete for users' willingness to upgrade their devices.

Behind the popularity of "AI-powered lobster farming," the Ministry of Industry and Information Technology (MIIT) has highlighted security risks associated with OpenClaw.

According to CCTV News, the Cybersecurity Threat and Vulnerability Information Sharing Platform of the Ministry of Industry and Information Technology recently detected that some instances of the open-source AI agent tool OpenClaw pose a high security risk under default or improper configuration, which could easily lead to cyberattacks and information leaks.

The Ministry of Industry and Information Technology pointed out that OpenClaw has the problem of "fuzzy trust boundaries" when deployed, but it has the characteristics of autonomous continuous operation, autonomous decision-making, and calling system and external resources.

Without effective access control, auditing mechanisms, and security hardening, this tool may be subject to unauthorized operations due to misleading instructions, configuration defects, or malicious takeover, resulting in a series of security risks such as information leakage and system control.

In response to the aforementioned risks, the Ministry of Industry and Information Technology recommends that relevant entities and users fully verify public network exposure, permission configuration, and credential management when deploying and applying OpenClaw, disable unnecessary public network access, and improve security mechanisms such as identity authentication, access control, data encryption, and security auditing. They should also continue to pay attention to official security announcements and hardening recommendations.

According to China News Service, Wang Jian, a member of the National Committee of the Chinese People's Political Consultative Conference and an academician of the Chinese Academy of Engineering, said in an interview that the "crayfish farming" phenomenon will soon become cheaper and more widespread , and no one in the industry will be unaware of the existence of OpenClaw.

Gao Wen, a deputy to the National People's Congress and an academician of the Chinese Academy of Engineering, also quoted Ma Huateng's view during his speech: "Ma Huateng said that even he did not expect that this 'lobster' would become so popular."

Shenzhen's Longgang District plans to introduce policies to support crayfish farming.

Yesterday, the Shenzhen Longgang District Artificial Intelligence (Robotics) Bureau released a draft of "Several Measures of Shenzhen Longgang District to Support the Development of OpenClaw & OPC (Draft for Solicitation of Opinions)" for public comment.

The document focuses on the AI ​​intelligent agent framework OpenClaw and the OPC ecosystem, and introduces ten systematic support measures, ranging from deployment and development, data services, computing power supply to talent introduction, with the highest single subsidy reaching 10 million yuan.

• At the deployment and development level, Longgang District encourages platform carriers to launch "Lobster Service Areas" to provide free OpenClaw deployment services, and provides subsidies of up to 2 million yuan for projects that contribute key code to the international mainstream community and develop embodied intelligence combined with applications;
• On the data side, Longgang will open up high-quality de-identified public data in areas such as low-altitude airspace, transportation, medical care, and urban governance. Related development service fees will be offered at a 50% discount, and companies that purchase self-developed AI NAS ("Lobster Box") can receive a subsidy of 30% of the market price .
• Regarding the procurement of intelligent agent tools, Longgang District will implement the "OpenClaw Digital Employee Application Voucher" program. Enterprises that purchase or build their own OpenClaw intelligent agent solutions can receive subsidies of up to 40% of the total project investment, with an annual limit of 2 million yuan per enterprise.
• For in-depth application demonstration projects in scenarios such as intelligent manufacturing, smart government affairs, smart parks, and smart healthcare, a one-time reward of up to 1 million yuan can be obtained. AIGC enterprises that use leading domestic multimodal large models for creation and production can enjoy a 30% subsidy on the usage fee, with a maximum of 1 million yuan per enterprise per year;
• In terms of computing power guarantee, newly registered companies in the OPC community will receive three months of free computing power resources; for demonstration projects with industry-leading significance, support of up to 4 million yuan will be provided based on 50% of the actual investment.
• Newly registered PhD, Master's, and Bachelor's degree holders in Longgang can receive a settlement subsidy of up to 100,000 yuan, depending on their qualifications; newly registered or relocated OPC companies can enjoy up to two months of free accommodation and up to 18 months of preferential office space.
• Longgang District will also provide equity investment support of up to 10 million yuan for seed-stage OPC projects through the district's science and technology innovation "seed fund", Longgang Cloud Map Industry Fund and Artificial Intelligence Industry Fund of Funds.

In addition, the measures cover two dimensions: support for overseas expansion and rewards for events. These measures are proposed to be implemented from 2026 and will be valid for three years.

OpenAI's head of robotics has officially announced his departure.

Yesterday, Caitlin Kalinowski, head of OpenAI's robotics division, publicly announced his departure on X.

I have resigned from OpenAI. I deeply value the robotics team and what we have accomplished together… I have great respect for Sam and the entire team, and I am proud of what we have achieved together.

Regarding the number of internal departures, Kalinowski stated in a subsequent interaction with netizens that she could not disclose any internal details. When asked about her next move, she responded, "I will be taking a break for a while, but I remain very focused on building responsible entity AI."

Kalinowski joined OpenAI last November to lead the robotics and consumer hardware engineering team.

Previously, she served as a technology lead at Apple, participating in the development of the Mac Pro, MacBook Air, and the first-generation unibody MacBook Pro; later, she joined Meta Reality Labs, where she led the VR hardware and AR glasses hardware teams and spearheaded the prototype AR glasses project codenamed "Project Orion".

The background to this resignation is the cooperation agreement signed between OpenAI and the U.S. Department of Defense (Pentagon).

PS5 heavily modified Linux PC: GTA V Enhanced Edition ray tracing can achieve a stable 60 FPS.

According to NotebookCheck, security engineer Andy Nguyen recently successfully ported a Linux system to the Sony PS5 console and ran GTA V Enhanced Edition on Steam, achieving a stable 60 frames per second at 1440p resolution with ray tracing enabled.

It is reported that Andy used a PS5 running firmware version between 1.0 and 2.0 (an early version released about five years ago) and bypassed the virtual machine management program deployed by Sony on the console by exploiting a vulnerability called "Byepervisor".

As seen in the publicly available demonstration video, the modified console is a PS5 Slim running a full version of Linux, with no functional reductions. The modified console supports 4K HDMI video and audio output, and all USB ports are fully functional.

It's worth noting that the GPU acceleration functionality is currently incomplete, with some features still having limitations, and the entire project is still in the early exploratory stage. Furthermore, due to its reliance on a specific firmware vulnerability, this solution currently only works on devices running older firmware, making it difficult for ordinary users to reproduce the issue.

Alibaba's latest paper: Large AI models "spontaneously jailbreak" during reinforcement training, hijacking GPUs for cryptocurrency mining.

Alibaba Group's research team recently published a technical paper on arXiv, disclosing a security incident that occurred during the training of the open-source intelligent agent large model ROME:

During the reinforcement learning (RL) training phase, the model spontaneously performed a series of dangerous behaviors without any human instruction, including hijacking GPU computing power for cryptocurrency mining, establishing a reverse SSH tunnel to bypass firewalls, and actively probing internal network resources.

The paper explicitly states that none of the above behaviors appeared in the task prompts, nor were they necessary steps to complete the predetermined goals.

The research team believes this is a byproduct of the model's spontaneous emergence under the pressure of optimization in reinforcement learning. In order to maximize rewards, the model "autonomously invented" these cheating paths during multiple rounds of tool calls and code execution, without issuing any notification to humans throughout the process.

The research team took immediate action after identifying the problem. Nevertheless, the team acknowledged that the current model is "still seriously inadequate" in terms of safety, reliability, and behavioral alignment, and called on the community to continue to focus on AI safety and controllability issues.

Full text of the paper: https://arxiv.org/abs/2512.24873

OPPO Find N6 Appearance Revealed: Hasselblad Quad Camera and New Color Options

Recently, well-known leaker Evan Blass (@evleaks) posted renders of OPPO's upcoming foldable flagship Find N6 on X.

The images show that the Find N6 retains the slim and lightweight design of its predecessor, the Find N5, with a circular rear camera module and the flash located outside the lens module.

The external screen uses a centered punch-hole design, resulting in a relatively slim and elongated screen ratio. In terms of color options, the Find N6 may add a "golden orange" color, in addition to deep black and original titanium (or gray/silver) colors.

Other specifications are as follows (as reported online):

• Equipped with a Snapdragon 8 Elite Gen 5 processor;
• It features an 8.12-inch 2K LTPO UTG main screen and a 6.62-inch secondary screen;
• The rear camera setup includes a 50MP main sensor, a 50MP ultrawide sensor, and a 2MP depth sensor, plus a 2MP multispectral sensor; the front camera features two 20MP selfie lenses.
• The battery capacity is approximately 5850~6000mAh;
• It offers 12GB/16GB + 1TB storage options, weighs approximately 225g, and will also launch a BeiDou satellite communication version;

As previously reported, Zhou Yibao, head of OPPO Find series products, had released a video of the Find N6, highlighting the creases on the device after it was taken out of the water, with the creases being "almost invisible" to the naked eye.

Xiaohongshu's open-source image editing model, FireRed-Image-Edit v1.1

Yesterday, Xiaohongshu's Super Intelligence team officially released and open-sourced its image editing model, FireRed-Image-Edit v1.1. Reportedly, this update significantly improves core capabilities such as ID consistency editing, multi-element fusion, portrait beautification, and font style reference.

The biggest highlight of v1.1 in terms of editing performance is its ability to integrate OOTD (Outfit of the Day) elements. The model supports the free combination of more than 10 elements and is equipped with Agent intelligent automatic cropping and stitching functions, which can handle complex composition scenarios with multiple image inputs without the need for cumbersome prompting text engineering.

Regarding portrait consistency, the team positions it as "the best role identity preservation technology in the open-source field," ensuring that the subject remains recognizable even after complex editing. Furthermore, the model covers dozens of portrait makeup styles, as well as high-fidelity font style references and old photo restoration capabilities.

GitHub: https://github.com/FireRedTeam/FireRed-Image-Edit

Hugging Face: https://huggingface.co/spaces/FireRedTeam/FireRed-Image-Edit-1.1

Technical report: https://arxiv.org/abs/2602.13344

The "PC-grade" WPS is now available on the App Store.

Yesterday, the official version of Kingsoft WPS for Pad was first released on the Apple App Store in Hong Kong, Macau, and Taiwan. Users without TestFlight testing privileges can download it directly now.

Last December, the productivity version of WPS Office for Pad announced the launch of TestFlight open beta, and at the same time, it also launched corresponding products adapted for Huawei tablets, Honor tablets, Vivo tablets and Xiaomi tablets.

According to the official introduction, this final version achieves the same core functions as the Windows version, is perfectly adapted to iPadOS 26, supports full professional capabilities such as new Office formulas, complex pivot tables, and LaTeX formula editing, and is fully compatible with mobile touch gestures.

Regarding subscription pricing, WPS for Pad offers in-app purchases for WPS members, with a 36-month Super Membership priced at HK$368 (approximately RMB 325.6).

McDonald's CEO's food tasting video was ridiculed, and Burger King seized the opportunity to add insult to injury.

According to Fortune and Business Insider, McDonald's CEO Chris Kempczinski recently faced widespread ridicule on social media after a video of him tasting a burger unexpectedly sparked a "burger war" between fast-food brands.

The incident began in early February this year when Kambosinski posted a promotional video for McDonald's new product, "Big Arches," on his personal Instagram account.

In the video, he repeatedly referred to the burger as "product" and, after a moment's hesitation, said he "didn't know how to eat it." He then took only a small, cautious bite and promised to "enjoy the rest of his lunch" off-camera.

The video has now garnered nearly 11 million views on Instagram, boosting Camposinski's personal account follower count by approximately 30%. Comments include, "This guy looks like he eats salads, drinks fruit and vegetable juices, and runs marathons every day," and "He doesn't look like someone who eats McDonald's."

After the incident escalated, competitors quickly jumped in. Burger King President Tom Curtis appeared in a 13-second video wearing an apron, taking a big bite of a Whopper, with the caption "Let's replay this."

Wendy's then joined the fray, with its US president, Pete Suerken, enjoying a Baconator burger and captioning it, "This is what it looks like to truly love your own 'product'."

Honor of Kings officially announced its crackdown on "boosting" services.

Recently, the official "Honor of Kings" released a comprehensive optimization announcement for its security system, stating that it will use AI technology to identify and combat a series of violations such as account boosting, AFK (away from keyboard), and intentionally feeding.

The core of this upgrade is the restructuring of the trial system. The new system adopts a dual-track system of intelligent AI + human review, which can analyze in-game behavior 24 hours a day and supports real-time monitoring of malicious behaviors such as AFK, intentionally feeding, and passive play.

Those with minor offenses will receive a warning, while those with serious offenses will be kicked out of the game. At the same time, the system has added a "second review" mechanism to conduct a thorough review of reported content, reducing misjudgments and omissions.

Regarding the identification of account boosting services, the official statement said that it will anchor and compare account operation habits. Even if the booster deliberately imitates the account owner's game style, the system can still detect the disguise through micro-operation differences such as combo rhythm and reaction speed. Once the boosting behavior is confirmed, corresponding penalties will be imposed according to the severity of the violation.

Regarding the governance of illegal content, the Honor of Kings security team stated that it will continue to expand the illegal word database to cover multiple dimensions such as text, voice, and images, and will carry out special governance of issues such as "making inappropriate jokes".

JPG Coffee launches new "Energy Cookies"

Yesterday, JPG Coffee officially announced the launch of its new coffee beverage, "Energy Cookie," which is now available at its JPG Coffee and JPG Canteen stores.

According to the official statement, the new "Energetic Cookies" blends "salty buttery milk" with the brand's signature dark roast coffee liquid, aiming to create a rich aroma and silky smooth flavor experience; it also adds "caramel cookie flavor" to increase the layers of flavor, which may include caramel cookies, butter and salty cheese.

vivo releases a short documentary film titled "Becoming Yourself" to celebrate International Women's Day.

Yesterday, vivo released a new documentary short film, "Becoming Yourself," on Weibo. Using International Women's Day as a starting point, the film uses the vivo X300 Pro to record the true story of a young creator named "185 Student."

The short film, which is about 4 minutes long, focuses on themes such as choice, courage and self-identity, emphasizing the importance of firmly following one's own path beyond the gaze of others.

Filming for "The Wandering Earth 3" wrapped in May, and it will be released during the Spring Festival next year.

According to China Film Report, Fu Ruoqing, chairman of China Film Group, revealed yesterday that "The Wandering Earth 3" is expected to wrap up filming in May this year. The film is directed by Guo Fan, and features the return of the original cast including Wu Jing, Andy Lau, Qu Chuxiao, and Zhao Jinmai. It is adapted from Liu Cixin's novel of the same name.

Unlike the previous two films, *The Wandering Earth 3* is presented in two parts for the first time. The first part is confirmed for release during the Spring Festival next year, while the second part is planned for release during the summer of next year or the Spring Festival of 2028. The producers emphasize that the two parts are interconnected and share a unified narrative.

The film continues the macro-background of the solar crisis, focusing on the life-and-death rescue stories of the younger generation during the advancement of the "Wandering Earth" project in 2500.

#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

Last night, XPeng held a press conference in Guangzhou, launching a G6 equipped with a 60-liter fuel tank to the market.

The new car is named XPeng G6 Super Extended Range Edition, with a starting price of 186,800 yuan. This entry-level 1704 Max version comes standard with XPeng's self-developed Turing AI chip. For users with higher computing power needs, XPeng also offers two other versions:

For an additional 12,000 yuan, you can upgrade to the Ultra SE version with dual Turing chips; for an additional 20,000 yuan, you can upgrade to the Ultra version with three Turing chips.

It has been exactly three years since the first-generation G6 was launched.

Back then, the domestic auto market was fiercely competing on the energy consumption performance of pure electric platforms. Today, the competitive logic in the 200,000 RMB-level new energy SUV market is completely different.

Consumers still crave user-friendly driver assistance systems, but they also care about the convenience of avoiding long charging stops at service areas when driving long distances during holidays. They want the low cost of electric vehicles for urban commuting and the ease of driving long distances with gasoline cars – a combination that today's consumers desire.

This is why XPeng launched the G6 Super Extended Range Edition.

The XPeng G6 is more than just a large fuel tank.

This time, XPeng has equipped the G6 with a 55.8 kWh battery and a 60-liter fuel tank.

Combining these two components results in a combined range of 1704 kilometers under CLTC conditions. Looking at the pure electric range alone, it can travel 430 kilometers. For daily commutes in the city, we can basically drive it like a pure electric vehicle.

The battery life has increased, and the charging speed hasn't suffered either.

Many range-extended vehicles on the market, considering cost, often adopt a 400V platform, or differentiate between low-end 400V and high-end 800V models within the same series. However, the G6 Super Range Extender comes standard with 800V plus 5C high-voltage fast charging across the entire series.

According to XPeng's test data, it only takes 12 minutes to charge from 10% to 80%. The overall energy consumption is also low, at 13.9 kWh per 100 kilometers.

To address the noise generated when the range extender kicks in, XPeng has incorporated a dual-source active noise cancellation system, including ENC and RNC, which enhances cabin quietness.

Having discussed the three core electric components, let's look at XPeng's long-standing core competitive advantage – assisted driving.

The entry-level Max version, priced at 186,800 yuan, comes standard with one Turing chip. If you're willing to pay extra, the Ultra SE version will have two chips, while the top-of-the-line Ultra version will offer three.

In the Ultra version's hardware configuration, two chips are responsible for processing driver assistance data, while another is dedicated to the smart cockpit. The maximum on-board computing power of 2250 TOPS significantly increases the vehicle's data processing capacity.

The hardware configuration is in place, and the software iteration plan is also clear.

The second-generation VLA visual language model is the key to this product experience upgrade.

According to XPeng's timeline, the second-generation VLA will be fully rolled out to all users of the Ultra series equipped with dual-chip and triple-chip systems in April this year. Max version owners will receive the optimized version a little later, in the second half of this year.

Meanwhile, to improve the intuitive experience of vehicle-machine interaction, XPeng has also deployed a 10B-level parameter VLM large model on the vehicle side, which will make the system respond more quickly to complex voice commands and visual perception.

To cater to the travel needs of more families, XPeng has also made corresponding upgrades to the chassis suspension.

The G6 Super Range Extender comes standard with a front double wishbone and rear five-link independent suspension. In addition, a CDC intelligent variable damping shock absorber has been added.

XPeng stated that, combined with the Taiji hydraulic bushings in the chassis, the vehicle's vibration damping performance when driving over bumpy roads and its body support during high-speed cornering have been optimized. Furthermore, this chassis system also possesses AI self-learning capabilities, allowing it to adjust damping in advance based on road conditions.

Finally, let's look at the vehicle architecture and safety features.

This extended-range version not only retains the front and rear integrated aluminum die-cast body of the pure electric model, but also increases the maximum trigger speed of AEB active braking and AES active avoidance to 130km/h in terms of active safety mechanisms.

In response to the extreme and unexpected situations that are easily encountered when driving on highways, XPeng has also introduced a high-speed tire blowout stability function at 180km/h, which can help the driver maintain the vehicle's posture on slippery roads or curves at the moment a tire blowout occurs.

XPeng, returning to pragmatism

Three years ago, when the first G6 was unveiled, everyone was still debating whether advanced driver assistance systems were actually any good.

At the time, the G6 was launched to the market with an 800V high-voltage architecture, front and rear integrated aluminum die casting, and all-scenario urban assisted driving, and could be called a "pioneering product".

Rewinding to today, the G6 has accumulated 180,000 owners. Based on last year's delivery data, it has secured the top spot in the domestic sales charts for pure electric mid-size SUVs priced between 180,000 and 200,000 yuan by 2025.

Over the past three years, XPeng has pushed out 29 major OTA updates to the G6, completing 287 core function updates.

These continuous software iterations, along with the recent changes to the chassis hardware and the addition of range-extended powertrains, outline XPeng's shift in product definition. They are beginning to shift their focus from simply pioneering technology to providing comfortable family travel and covering all long-distance travel scenarios.

In addition to making adjustments based on feedback from the domestic market, XPeng is also trying to use this car to open the door to the global market.

The G6 has now entered the automotive markets of more than 60 countries and regions worldwide. As the mainstay of XPeng's expansion into Europe, the G6 accounts for nearly 70% of XPeng's sales in Europe. In markets such as Norway and Denmark, it has consistently maintained a relatively leading market share among emerging Chinese brands going global.

To match testing standards in different regions around the world, XPeng did not differentiate between domestic and international vehicle architectures. The core safety hardware, such as body strength and anti-collision beams, remained consistent between the domestic and export versions.

With its established market presence both domestically and internationally, this SUV, which originally remained firmly in the pure electric vehicle camp, has now been equipped with a combustion engine, officially entering the range-extended vehicle market.

The current new energy vehicle market is fiercely competitive, and relying on a single technological advantage is unlikely to sustain rapid growth. Ordinary consumers value the ease of use offered by assisted driving systems, and they also need the confidence to drive long distances without range anxiety.

By incorporating its self-developed Turing chip and second-generation VLA large-scale model into the vehicle, retaining 800V high-voltage fast charging, and adding a 60-liter fuel tank, XPeng has addressed the pain points of pure electric vehicle usage in specific scenarios through an additive approach.

Today, XPeng's product philosophy is undergoing a visible change.

In the past, it has always wanted to reshape the industry with cutting-edge technology, but in the process, technological ideals will eventually collide with real-world car usage scenarios.

In a zero-sum game where only existing players compete, the top priority for automakers is always to embrace a wider audience with a more balanced overall configuration, survive, and seize market share.

#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

Starting at 4599 yuan, the MacBook Neo is coming!

Last night, the MacBook with the same processor as the iPhone, which everyone has been waiting for, was finally released. It's called the "MacBook Neo." Let's take a look at the price:

The 256GB version is priced at 4599 yuan, and the 512GB version is priced at 5299 yuan; both versions come with 8GB of RAM.

The screen is a 13-inch Liquid Retina display with a maximum brightness of 500 nits and supports 1 billion colors, but only the sRGB color gamut; it is equipped with dual speakers and a dual microphone array.

In terms of processor, both versions use the A18 Pro chip (a 6-core CPU + 5-core GPU version) and are equipped with 60GB/s memory bandwidth. According to the official website, this machine will support Apple Smart.

The exterior is crafted from recycled aluminum and comes in four colors: silver, indigo, yellow, and pink. Notably, the keyboard matches the body color. However, keyboard backlighting has been removed, and the touchpad lacks pressure sensitivity.

In terms of interface configuration, the MacBook Neo is equipped with only one USB 3 port (10Gb/s) and one USB 2 port, and has a 3.5mm headphone jack.

Related reading: Over 3,000 yuan after national subsidy! Apple's dopamine-inducing MacBook Neo is here! No notch.

Lei Jun urged: Don't over-rely on driver assistance systems.

According to the Beijing Daily, as the 2026 National People's Congress and the Chinese People's Political Consultative Conference (NPC and CPPCC) are about to open, Lei Jun, a National People's Congress deputy and founder of Xiaomi Group, gave an interview to reporters, sharing his views on how to strengthen the safety defenses of intelligent vehicles.

Lei Jun specifically mentioned that current driver assistance functions in smart cars still heavily rely on human drivers, who must remain focused while driving. He urged the media to increase public awareness, making the public understand that current driver assistance systems have not yet reached Level 4, and therefore drivers must not be complacent.

In the "2026 NPC Deputies' Suggestions" released by Lei Jun yesterday, he also mentioned that the current intelligent technology routes are diverse and difficult to regulate with a unified standard, and the driver training content and assessment standards have not yet been well connected and matched with the new technologies and functions of vehicles.

In response, Lei Jun suggested that efforts should be accelerated to develop intelligent vehicle technology standards and optimize motor vehicle driving test items , increasing the weight of intelligent connected vehicle-related content in driving tests. At the same time, it is also necessary to develop and promote a new generation of driving simulation training equipment that integrates scene cognition, complex road condition simulation, and dangerous situation handling functions.

According to the China Fund News, He Xiaopeng, a deputy to the 14th National People's Congress and chairman and CEO of XPeng Motors, recently submitted three proposals on key areas of technological innovation and industrial innovation, involving industries such as autonomous driving technology.

He Xiaopeng stated that promoting the leap from Level 2 to Level 4 in policy and regulatory systems, and facilitating rapid technological iteration and large-scale commercialization, will help transform my country's accumulated advantages in Level 2 into competitive advantages in the era of Level 4 autonomous driving.

He Xiaopeng offered four suggestions:

• It is recommended that, while maintaining the stable operation of the Level 2 safety regulatory system, autonomous driving policies should be promoted to leap from Level 2 to Level 4, and intermediate steps for Level 3 should be simplified.
• It is recommended to gradually clarify the registration and traffic management system for Level 4 autonomous vehicles and gradually promote the compliant operation of Level 4 vehicles on roads nationwide.
• It is recommended to conduct a traffic regulation compatibility assessment and, under the premise of safety, optimize and form traffic behavior norms applicable to both "human driving" and "machine driving" categories, taking into account the characteristics of autonomous driving.
• It is recommended to grant local pilot management authority for L4 autonomous driving applications in specific scenarios, allowing some cities with mature basic conditions to conduct L4 application pilots in specific low-risk scenarios, and gradually form replicable and scalable experiences.

After Lin Junyang submitted his resignation, Alibaba executives provided urgent answers.

Early yesterday morning, Lin Junyang, the technical head of Alibaba's Qianwen Big Data Model, suddenly announced his resignation on social media. This sparked a heated discussion on the internet.

It is understood that Lin Junyang is the core driving force behind Alibaba's open-source AI model and one of Alibaba's youngest P10s.

According to "Intelligent Emergence," Tongyi Lab held an emergency All Hands meeting yesterday afternoon, where Alibaba Group CEO Wu Yongming frankly told Qianwen employees, "I should have known these things earlier."

Attendees at the meeting included several senior executives from Alibaba, the Qwen team, and other team members from Tongyi Lab. Several responses were given by Alibaba Chairman and CEO Wu Yongming, Alibaba Chief Talent Officer Jiang Fang, and Alibaba Cloud CTO Zhou Jingren regarding key issues such as team adjustments and strategic direction.

The core assessment given by Alibaba's senior management regarding this adjustment is: Qwen is not shrinking; this is a team expansion, unrelated to any political struggles, and on the contrary, it requires more resources.

The report stated that Alibaba's Chief Talent Officer, Jiang Fang, also acknowledged shortcomings in communication: "We are developing rapidly, and this adjustment is to expand our talent pool and provide more resources. The organizational structure wasn't communicated well this time. The introduction of new talent will inevitably bring about changes in our formation, and these are things that will inevitably be involved in the expansion process. We may not have handled this well."

Regarding rumors that Zhou Hao will directly lead Lin Junyang and his team, the report stated that "Zhou Hao's successor position and reporting line are still under discussion."

In response to the issue of "external users having smooth access to Alibaba Cloud computing power while internal teams are struggling with limited computing power and recruitment quotas," Zhou Jingren stated that the team is in a "resource-scarce state," and the internal and external differences have many historical reasons. They are currently making overall plans for the future, but did not elaborate further.

The report also mentioned that Alibaba executives are still in close communication with Lin Junyang, and it is still unknown whether Lin Junyang has confirmed his departure from Alibaba.

Yesterday afternoon, Lin Junyang posted on his WeChat Moments, saying, "Qwen's brothers, let's continue as planned, no problem," without specifying whether he would return.

Nvidia will hold an AI conference this month.

Nvidia recently announced that it will hold its AI and Accelerated Computing Conference (GTC) in San Jose, California, from March 16 to 19 this year.

It is understood that CEO Jensen Huang will still deliver a keynote speech at this AI conference, which will also include more than 1,000 sessions and groundbreaking achievements covering the entire AI stack.

Jensen Huang stated that GTC is the center of the AI ​​industrial era. He also pointed out:

AI is no longer a single breakthrough or application, but an essential infrastructure that every company will embrace. Every country will build it. From energy and chips to infrastructure, models, and applications, all layers of the stack are advancing in synergy, and all of this will be fully showcased at GTC.

According to official sources, Huang Renxun's keynote speech will be held at 11:00 AM local time on Monday, March 16.

Tesla's Cybercab driverless taxi may be entering mass production.

According to TESLARATI, Tesla is accelerating pre-production of its driverless taxi, Cybercab, and plans to begin mass production next month.

According to reports, at least 25 Cybercabs have been spotted at Tesla's Texas Gigafactory, marking the largest single-day exposure of the model to date.

The footage shows 14 vehicles parked at the factory exit, 9 vehicles undergoing structural and safety verification at the crash test center, and 2 vehicles undergoing final testing in the final assembly line area. Some vehicles have also been observed undergoing dynamic road testing within the factory area.

It is reported that the first mass-produced Cybercab rolled off the production line in mid-February 2026 using a proprietary "unpacking process," ahead of the originally scheduled April date.

Based on previous statements from Tesla executives, the report points out that the high frequency of testing and manufacturing signifies a substantial ramp-up in production capacity, with full-scale mass production expected to begin within the next 4 to 8 weeks. The current cautious approach is primarily aimed at validating the vehicle's new architecture and its compatibility with vision-based Full Self-Driving (FSD) technology.

AMD's Lisa Su: Rising memory prices will put significant cost pressure on the PC market.

According to Yicai Global, on the evening of March 3, AMD CEO Lisa Su held a dialogue with Morgan Stanley analysts at the Morgan Stanley Technology, Media and Telecom Conference. During the conference, Su addressed several hot market issues.

When discussing the collaboration with OpenAI, Lisa Su stated that the partnership between AMD and OpenAI is better than ever before. The two parties are actively planning the first gigawatt of computing power installation, and the MI450 has been jointly verified by both parties. The transaction structure between the two parties has not changed.

In response to concerns that "large tech companies' investments in AI infrastructure may not be sustainable in the long term," Lisa Su stated that investments in AI infrastructure can be viewed, to some extent, as investments in productivity and intelligence. Companies are investing ahead of time and taking into account the returns they will receive.

Lisa Su also addressed the CPU supply shortage, stating that the market size is larger than predicted three and six months ago, and the supply chain needs time to meet market demand. She revealed that AMD is well-positioned with its supply chain to meet a large proportion of demand and will expand supply capacity in 2026 and 2027.

Regarding the impact of memory shortages on GPUs and CPUs, Lisa Su stated that the price increases of DDR4 and DDR5 memory, used in consumer products, are affecting systemic pricing. She anticipates that the memory price hikes will put greater cost pressure on the PC market, and that market volatility may be more moderate in the second half of this year, but further observation of changes in the memory market is still needed.

In addition, Su Lifeng also mentioned at the meeting that "in the next stage of AI infrastructure, no single chip can do everything perfectly; it is already a heterogeneous world."

Honda plans to import its all-electric vehicles produced in China back to Japan.

According to Nikkei, Honda plans to import its all-electric vehicles manufactured in China to Japan and begin sales as early as spring 2026.

The report points out that this is the first time a Japanese automaker has imported and sold pure electric vehicles produced in China.

It is understood that Honda will use vehicles from its "e:N" brand of pure electric vehicles, which was launched in China in 2022, as a prototype and adapt them to Japanese specifications before selling them.

According to inquiries, the relevant model may be the e:NP 2 (also known as the e:NS 2), produced by GAC Honda and Dongfeng Honda respectively. Both are pure electric compact SUVs, scheduled to go on sale in 2024, and are built on Honda's pure electric e:N architecture.

The report states that the proposed brand name for returning to the Japanese market is "Insight". Honda launched the first-generation Insight in 1999 and discontinued production of the third-generation model in 2022.

Furthermore, Honda's move is also aimed at rescuing its China factories, which are struggling with declining market performance. According to reports, Honda's new car sales in China in 2025 were 645,345 units, a 24% year-on-year decrease, marking the fifth consecutive year of decline.

Step 3.5 Flash Training Framework for Star Leap (Open Source)

Yesterday, Step Star announced that following the open-sourcing of the Step 3.5 Flash model, it has also open-sourced the pre-training/in-training/training framework for this Agent base model.

It is understood that Step 3.5 Flash is the most powerful open-source foundational model developed by StepStar to date, possessing powerful reasoning capabilities and agent intelligence, and is designed specifically for agents.

StepStar aims to enable developers to use Step 3.5 Flash as a foundation for deeper model customization, creating truly unique agents through more thorough open-source development.

• Base weights: The basic model, suitable for fine-tuning all parameters;
• Midtrain weights: stronger reasoning ability, suitable for secondary development such as agent and tool calls;
• Steptron Training Framework: The official training framework, covering continuous pre-training, supervised fine-tuning, reinforcement learning (coming soon), evaluation (coming soon), and other stages.

Base weight: https://huggingface.co/stepfun-ai/Step-3.5-Flash-Base

Midtrain weights: https://huggingface.co/stepfun-ai/Step-3.5-Flash-Base-Midtrain

Steptron training framework: https://github.com/stepfun-ai/SteptronOss

AI startup Flowith raises tens of millions of dollars in funding

According to Z Potentials, AI startup Flowith recently announced the completion of a multi-million dollar seed and seed+ funding round.

It is understood that the seed round was led by Vertex Ventures and other institutions, while the seed+ round was jointly led by Sequoia China Seed Fund, LongRiver and other top institutions. The funds raised will be mainly used for research and development and global market expansion.

The report states that Flowith gained significant attention upon its launch due to its innovative interactive methods, and pioneered several cutting-edge products and features, including the first general-purpose creative agent framework Oracle, AI Context Knowledge Garden, the infinite-step agent Agent Neo, and the first OS Agent – ​​FlowithOS, acquiring millions of dedicated users both domestically and internationally.

Seedance 2.0 pricing announced

Recently, ByteDance's Volcano Engine officially announced the pricing standard for its Doubao-Seedance-2.0 video generation model:

Video input costs 28 yuan per million tokens; input without video costs 46 yuan per million tokens.

It is understood that generating a 15-second video using Seedance 2.0 requires approximately 308,800 tokens. Based on this calculation, the price for pure video generation (excluding video input) is approximately 1 yuan per second.

It is worth noting that Doubao-Seedance-2.0 does not have API access enabled.

Ant Financial releases AReaL v1.0: One-click integration of agents into RL training.

On March 4th, Ant Group, in collaboration with Tsinghua University, released the stable version of AReaL v1.0, an open-source reinforcement learning training framework. This version emphasizes "one-click integration of agents into RL training": no code modifications are required, it is compatible with various agent frameworks, and allows for out-of-the-box use of reinforcement learning training for intelligent agents.

Since the beginning of 2026, the agent field has continued to gain momentum, with intelligent agent frameworks such as LangChain, Claude Code, and OpenClaw flourishing. However, this has also exposed two major bottlenecks: high access and training costs, and a lack of continuous evolution capabilities for agents.

AReaL is the first fully asynchronous training and inference decoupled large-scale model reinforcement learning training system, which enables agents to obtain feedback and continuously optimize decisions through real-world task interactions.

According to the official introduction, the newly released v1.0 version makes it possible for any agent to access RL training with zero modifications—by adding a Proxy Worker intermediary layer between the agent and the training system, developers only need to modify a request address to access the training.

AReaL v1.0 also introduced the native training engine Archon, which is based on PyTorch's native capabilities to achieve complete 5D parallelism (data parallelism, pipeline parallelism, tensor parallelism, context parallelism, and expert parallelism), reducing the installation and debugging threshold, while providing multiple backend options on the training and inference sides, making it easy to deploy flexibly in different environments.

It is reported that such a complex distributed system was implemented from scratch and its correctness verified in just one person-month of work—within 32 days, nearly one million lines of code were modified to fully implement the Archon engine, enabling it to train MoE models with hundreds of billions of parameters.

GitHub repository: https://github.com/inclusionAI/AReaL

Related paper: https://arxiv.org/abs/2505.24298

Zhiyuan Lingqu OS Open Source Launched

Yesterday, Logic Robotics announced that its Lingqu OS Alpha version has been officially released as open source.

According to reports, Lingqu OS is based on the full-scenario needs of embodied intelligent robots and aims to build an ecosystem architecture that "adapts to embodied hardware in the south and supports intelligent applications in the north".

The official statement indicates that the Lingqu OS Alpha version released this time is based on the full-size Zhiyuan Expedition A2 ontology, which is based on mass production practice. It features an open-source cross-platform embodied software framework, a bipedal motion control framework based on reinforcement learning, and a one-stop toolchain for bipedal motion control simulation training and deployment.

In terms of features, Lingqu OS Alpha version has features such as "unified communication middleware framework", "one-stop toolchain for reinforcement learning simulation, training and deployment" and "end-to-end development and deployment".

Lingqu OS Alpha: https://github.com/Link-U-OS

iQOO's flagship product line director announces resignation

Yesterday, "Golan V," the product line director of iQOO's flagship series, confirmed on Weibo that he would be leaving the iQOO team.

He stated that he joined the company in 2017 and the iQOO founding team in 2018, "thus beginning a great journey of 'Born to be strong and never stop exploring'."

Golan has not yet revealed his future plans.

According to available information, Ge Lan's real name is Jian Zhong, and he previously served as the product director of iQOO's flagship product line for a long time. Before joining the iQOO brand, he was a senior hardware engineer at Samsung Electronics, possessing extremely deep industry background and technical expertise.

Andrew Ng: AGI is still a long way off

In a recent interview on "This Is The World," artificial intelligence scholar Andrew Ng pointed out that AGI (Artificial General Intelligence) has been overhyped and turned into a marketing term, and asserted that the industry will not achieve true AGI by 2026.

In the interview, Andrew Ng criticized some companies for frequently altering and lowering AGI standards for public relations or financing purposes.

He proposed a brand-new "Turing AGI test": only if AI can independently complete economically valuable work tasks over several consecutive days, just like a skilled human remote worker, can it meet the public's reasonable expectations of AGI.

Andrew Ng stated that achieving this goal is still decades away, and with publicly available internet data nearing depletion, the path of simply relying on expanding the scale of parameters to improve intelligence is facing challenges.

Regarding the future direction of technological evolution, Andrew Ng pointed out that the core business value in 2026 and beyond will focus on "intelligent agent workflow".

He believes that compared to blindly pursuing an omniscient and omnipotent monolithic model, giving existing large language models the ability to call tools and guardrails, allowing them to handle vertical tasks such as legal compliance, medical assistance, and customer service step by step, will generate more certain economic benefits.

Samsung unveils concept phone with manually retractable screen

At the recent MWC 2026, Samsung showcased its latest retractable screen concept phone.

According to Android Authority, the concept phone has a 5.1-inch screen with a 16:9 aspect ratio when folded, and expands to 6.7 inches (22:9 aspect ratio) when fully unfolded, similar to the size of current standard candybar phones. It has an FHD+ resolution and a pixel density of 426ppi.

However, the device does not demonstrate motorized extension. Furthermore, the mass production timeline is yet to be determined.

Both Wenjie and Zunjie models have increased in price! First to feature pixel-level LiDAR.

Yesterday, the HarmonyOS Intelligent Mobility Technology Upgrade Launch Conference was officially held.

Related reading: The Shangjie and Zunjie models both see price increases! The Shangjie Z7's interior, featuring a first-ever pixel-level LiDAR, is also revealed.

At the conference, the 2026 Wenjie M9 and Zunjie S800 announced that they will be equipped with a new generation of 896-line dual-optical-path pixel-level LiDAR and comprehensively upgraded active safety technology.

Specifically, the new 896-line lidar can stably identify obstacles higher than 14 centimeters at a distance of 120 meters; at the same time, the radar surface is covered with a specially made tempered glass window.

In terms of pricing, the Zunjie S800, equipped with the new LiDAR, starts at 728,000 yuan, an increase of 20,000 yuan compared to the previous 192-line version; the starting price of the Wenjie M9 also rises to 479,800 yuan, a similar increase of 10,000 yuan.

In addition, the Shangjie Z7 series was also showcased at the launch event. The new car will be available in both sedan and shooting brake versions. The Shangjie Z7 series will feature a dynamic four-dimensional screen and an "inspiration showcase" in front of the passenger seat.

Android version of "PC Mode" is now available.

Recently, Google officially launched a new desktop mode for the Pixel 8 series and subsequent models.

It is understood that mobile phones that support this function can simply connect to the monitor using a USB-C cable to obtain a PC-like interface on the monitor.

It's worth mentioning that this feature not only supports mobile phones, but tablets can also have a similar experience, although tablets can achieve screen extension display.

According to reports, this feature draws inspiration from the design philosophy of Samsung DeX, aiming to provide users with a more efficient multitasking environment. Users can run multiple applications simultaneously, freely adjust the window size, and it supports wired/wireless keyboards and mice, making it convenient for light office work.

Voyah's new FE model will be equipped with a brand-new 896-line quad-laser system.

March 4th marked a significant moment in the field of intelligent driving. HarmonyOS officially announced the release of its new generation 896-line quad-LiDAR system. On the same day, the Voyah FE was unveiled, its silhouette revealing four new LiDAR modules.

Some sources speculate that the Voyah FE may be equipped with the world's highest-specification 896-line lidar, entering a new generation of four-laser radar era, just like the Zunjie.

It has been revealed that the four lidar sensors equipped in the Zunjie are L3-level intelligent assisted driving hardware. Their detection capabilities will see a qualitative leap in ultra-high-definition imaging, ultra-long-range detection, and ultra-precise recognition. They can accurately identify previously invisible gravel and distant obstacles, shortening response time. Even in adverse weather conditions such as rain, fog, and nighttime, perception stability is significantly improved.

Previously, the Voyah FE gained online popularity due to its nicknames "Ferrari" and "Four-Laser Ferrari," and was hailed as an FUV that exceeded expectations, making it the most exciting "rebellious new work" for young people in 2026. According to Voyah's previously revealed new car plan, the Voyah FE will be launched in mid-2026.

Luckin Coffee's major shareholder is reportedly set to acquire Blue Bottle Coffee.

According to LatePost, Centurium Capital has won the bid for Blue Bottle Coffee and is about to complete the transaction with Nestlé.

According to sources close to Luckin Coffee, Centurium has acquired Blue Bottle Coffee's global stores, while Nestlé will retain Blue Bottle's coffee machine and capsule businesses. Another source close to Centurium stated that the deal has been signed but not yet finalized.

The report points out that Centurium Capital, the actual controller of Luckin Coffee, began actively evaluating high-end coffee targets at the end of last year, including not only Blue Bottle, but also Costa, M Stand and % Arabica.

Centurium Capital, founded by Li Hui in 2017, is a private equity fund management company that currently manages over $7 billion in assets. Centurium became an early investor in Luckin Coffee in 2018, and after the Luckin Coffee accounting fraud scandal in 2020, it spearheaded the company's restructuring, taking over the equity of founder Lu Zhengyao and continuing to invest, becoming the actual controller of Luckin Coffee with over 50% of the voting rights. Last May, Li Hui became the chairman of Luckin Coffee.

The report also stated that the acquisition of Blue Bottle will help Luckin Coffee advance its premiumization strategy, and Luckin will also leverage its supply chain capabilities to optimize Blue Bottle's cost structure. Centurium Capital, which owns both Luckin and Blue Bottle through a "controlling investment," has achieved complete coverage in the coffee industry, from mass consumption to high-end boutique coffee.

According to 36Kr, Blue Bottle Coffee is still operating at a loss. As of June 30, 2025, Blue Bottle Coffee's revenue for the past 12 months was approximately $250 million, with the United States contributing approximately $150 million and the Asia-Pacific region contributing approximately $100 million; it is projected to achieve profitability in 2026.

The Sims 4 is set to launch on an official trading market.

Recently, the Sims 4 team officially announced the launch of a new official trading platform, "The Sims 4 Marketplace," which allows creators to earn a share of revenue by selling custom game content.

Creators will reportedly receive 30% of actual sales revenue (e.g., $0.30 for every 100 Moola a player spends). EA stated that the remaining 70% of the revenue will cover platform commissions, server maintenance, and official localization translation costs in 18 languages.

The trading platform will first launch on PC and Mac platforms on March 17, 2026, and will later be expanded to PlayStation and Xbox consoles.

Microsoft Xbox may support AI-assisted gameplay in the future.

According to Neowin, Microsoft has recently been reported to have applied for a new patent for Xbox games, which aims to use a cloud-based AI model to take over control and automatically complete challenging game levels for players.

The patent is titled "State Management of Video Game Assistance Sessions." The patent documents reveal that Microsoft is designing a cloud-based assistance system that would allow players to request a "helper" to temporarily gain full control of their current Xbox session when they encounter a game obstacle.

This mechanism differs from traditional multiplayer games where players summon or cooperate online. Instead, the system directly takes over the underlying physical or logical operations and independently completes the levels.

The document states that although the "assistant" could be another real player, Microsoft explicitly uses an AI model as the key technology for performing such core boosting tasks in its patent.

Timothée Chalamet to hold promotional tour in China; new trailer released for "Marty the King".

According to Sina Movies, a new trailer for the movie "Marty the Great" has been released, and star Timothée Chalamet will also embark on his first trip to China to promote the film.

The film is reportedly an original fictional story about a table tennis player, based on the life of American Marty Reisman. Directed by Josh Safdie, the older of the Safdie brothers, and starring Gwyneth Paltrow, Odessa Adlon, and Tyler, The Creator, it is produced by A24 and will be released on March 20th.

The horror-comedy film "Forbidden Fruit" releases its poster.

According to Hollywood Watch, the horror-comedy film "Forbidden Fruit" has released a poster and will be released in North America on March 27.

The film tells the story of Apple, Cherry, and Fig, employees of the same store, who form a (crossed out) fruit (crossed out) witch group. A new colleague, Pumpkin, questions their sisterhood, forcing the three to confront their inner darkness, which leads to violence.

The film is directed by Meredith Alloway, produced by Diablo Coty, and stars Lily Reinhart, Laura Don, Victoria Pedretti, and Alexandra Shipp.

Reports indicate that Warner Bros. has confirmed the development of a Game of Thrones movie.

According to a recent report by Page Six Hollywood, Warner Bros. has officially launched development on the first theatrical film based on "Game of Thrones," with renowned screenwriter Beau Willemmon having already submitted a prequel script to the company.

The report states that the film project is based on the fantasy novel universe of George R.R. Martin, with the core plot focusing on Aegon I, the founder of the Targaryen dynasty, and mainly telling the story of his war to conquer Westeros.

The project is currently in the early stages of production, and specific production schedules, directors, and cast have not yet been announced.

Sources familiar with the matter revealed that Warner Bros. executives gave a positive review of William Mong's initial script, believing that his experience in writing about political maneuvering was highly compatible with the film's narrative needs.

#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

Just now, at the Apple experience event in Shanghai, iFanr got its hands on the iPhone 17e, which was officially released two days ago.

From an appearance standpoint alone, the iPhone 17e is exactly the same as last year's iPhone 16e: the familiar thick bezel and notch screen, which can be traced back to the iPhone 14 four years ago, and it still does not have a high refresh rate. If you are used to the iPhone 17 Pro, the iPhone 17e will feel a bit laggy.

Actually, the iPhone 17e has another update that is not immediately obvious but very practical: like the iPhone 17, it features a second-generation Ceramic Shield front cover, making it more drop-resistant and scratch-resistant.

Turning to the back, there's still only a single camera, but this year's iPhone 17e features a brand new "light pink" color option, which looks even better than the basic black and white. This pink is indeed very light, often appearing very close to white.

As an entry-level model, the iPhone 17e's product segmentation remains quite precise. Upon first handling, it's immediately apparent that the phone's build quality isn't as refined as flagship models.

The phone's back panel is still made of frosted glass, and there is no smooth transition between it and the metal frame, making it slightly uncomfortable to hold.

The upgrades to the iPhone 17e focus more on the phone's internal components. A MagSafe magnetic coil has finally been added to the bottom of the back panel, allowing for seamless compatibility with existing magnetic accessories.

Thanks to the addition of MagSafe, the iPhone 17e supports 15W wireless charging power, double that of the previous generation.

We also tested the performance of the "downgraded" A19 chip in the iPhone 17e. When playing demanding games like Arknights: The End, the phone started to get hot after about a minute of screen-on time, so it seems to be struggling a bit.

Of course, the on-site experience was brief and for reference only. More detailed performance and gaming capabilities can be compared and tested once we have the phone ourselves. Perhaps by then, game and app developers will have already optimized for this device and its chip.

In terms of photography features, the iPhone 17e, besides lacking an ultra-wide-angle lens, also makes the following compromises:

The "adjustable color" photography style feature, which was already available on the iPhone 16, is not included.

The iPhone 17's new Center Stage front-facing camera feature, which allows you to take high-resolution horizontal selfies when holding the phone vertically, is also absent.

One important detail: the Chinese version of the iPhone 17e supports eSIM this time, but no longer supports physical dual SIM cards. For dual SIM dual standby, one of the SIM cards must be an eSIM.

The most crucial advantage of the iPhone 17e is its price. Apple has been very generous this time, not only maintaining the starting price of 4499 yuan, but also upgrading the storage from 128GB to 256GB – more for the same price. In 2026, when phone prices are expected to rise, the iPhone 17e actually offers better value for money.

What would you like to know about the iPhone 17e? Feel free to leave your questions in the comments section, and we will answer them one by one in our subsequent review.

#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

Two stage formats combine wonder and scientific rigor, taking collective attention beyond the threshold of the "already seen."

There comes a point during a plenary session when the audience stops listening "in pieces" and returns to being a whole. It's an almost physical transition: conversations pause, phone screens dim, attention realigns. At that moment, it's decided whether a convention or corporate meeting will remain an ordered sequence of content or become a shared experience capable of leaving its mark.

In business contexts, however, surprise cannot be a simple stage effect. It must have logic, meaning, and coherence with the company's message. This is why BrainArt® offers two formats specifically designed for the stage, designed to address the technical and narrative core of the plenary session: Live Demo and Scientific Speech . They can work together or separately, modulating the depth of the experience based on the event's tone and communication objectives.

From peak attention to memory: the grammar of wonder in the workplace

The plenary session is the place where direction, rhythm, and collective perception coincide. In distributed moments, corners, networking, and parallel activations, the event is fragmented into individual interpretations. In the auditorium, however, the audience becomes a single organism: and precisely for this reason, the most effective amazement is not the loudest, but the most shared.

In a business context, amazement works when it doesn't "fill" the agenda but rather directs it. When it creates a peak that separates a before and an after, offering the audience proof, an image, or an idea that becomes a reference. This is where BrainArt® formats come in.

The Live Demo brings a real-time performance to the stage: a presenter or volunteer wears an EEG sensor, and brain activity is transformed live into a visual composition projected onto a screen. The format's strength lies in a detail that the audience perceives immediately: they're not watching "prepared" content, but a process happening right now. It's a revelation that takes the form of a rehearsal, and for this very reason, it captures attention without asking for it.

Alongside the demonstration, the Scientific Speech (15–20 minutes) acts as a bridge between wonder and understanding. It's neither an academic lecture nor a generic motivational talk: it's tailor-made edutainment , capable of connecting neuroscience and brand themes in a credible way. If the demo triggers the perceptual event, the speech gives it direction, transforming surprise into meaning and, consequently, more stable memory.

The most interesting effect comes later: when the peak doesn't end in the moment, but leaves an output , a visual trace that summarizes the experience and makes it narratable. This is where amazement stops being a flash and becomes a common language: something people quote, repeat, and share internally. Not to "make noise," but because they have a clear message to take out of the room.

In this way, Live Demo and Speech Scientifico answer an increasingly central question for conventions, meetings, and product launches: how to create a moment that isn't just entertainment, but a device capable of elevating the content. One focuses on the peak and the test, the other on the interpretation and meaning. Together—or each individually—they create what's often missing from corporate events: a moment that doesn't just please, but endures.

The article " When the Plenary Ignites: How to Turn Conventions and Meetings into Memorable Moments" was written on: Tech | CUENEWS .