Pogbagram

Just now, the 2026 Beijing Yizhuang Humanoid Robot Half Marathon kicked off.

The top three teams all used the Honor "Lightning" robot, and their results are as follows:

• First place: The Monkey King Team, score: 00:50:26
• Second place: Thunder and Lightning, score 00:50:56
• Third place: Spark Spreads Like Wildfire Team, score: 00:53:01

The champion and runner-up were only 30 seconds apart, and all three finished in under 53 minutes, significantly improving last year's champion's time of 2 hours and 40 minutes, and also breaking the world record for the half marathon.

More than 300 robots, 26 mainstream brands, contestants from 13 provinces and municipalities, and overseas laboratories from Germany, France, and Brazil competed in a highly cyberpunk manner on a 21.0975-kilometer track.

The scene was packed with people, and there were so many (robots) everywhere you looked.

I thought it would be a tough test of technical skills, but after only 5 minutes of watching the live stream, I seamlessly switched to watching a variety show.

Let's take a look at the most memorable moments from today's race that are definitely worth capturing and saving. The audience was thoroughly enjoying it, and even human runners were actively cheering for the robot from the starting line, creating a strangely exciting scene.

The first to appear were almost all highly anticipated seeded players. The Jueying Chitu team from Beijing Glory fired the first shot, and their aircraft was the "Lightning," which has been extremely popular this year.

According to this year's competition rules, the participating robots are divided into two modes: autonomous navigation and remote control operation. The scores of the remote control group will be multiplied by a weighting factor of 1.2, and various penalties during the competition will also be added. The robot that crosses the finish line first may not be the final champion.

The starting procedure uses a streamlined, single-launch approach, with one robot released every 30 seconds. Throughout the run, the robot stays on the right, with the left side reserved for overtaking and obstacle avoidance. Following vehicles must maintain a safe distance of at least 20 meters from the robot.

The robots ran at breakneck speed, with the initial speed of the robots at the start consistently between 6 and 8 m/s. Not long after the race began, the robots that started later overtook the leading group. Upon closer inspection, some robots had ice packs strapped to their backs for cooling.

Then came the first memorable moment of the day: a robot running suddenly stopped, seemingly trying to get into a car. Another robot veered off course and crashed into the roadside fence, delivering a performance worthy of an Oscar for its audacity in faking an accident.

The robots are highly sensitive to obstacles in their path while running, and will brake suddenly and fall if anything goes wrong. Therefore, the organizing committee requires the robots to start at intervals to avoid such chain collisions.

The robots don't consume energy gels, but there are energy replenishment stations along the track for battery swapping and emergency response. Some robots lost equipment during the replenishment process without even realizing it.

If speed isn't enough, make up for it with style. Add to that the hair that flutters in the wind, and the humanoid robot has today's OOTD (Outfit of the Day).

The track environment is more complex than last year. The race is 21.0975 kilometers long and introduces the ecological section of Nanhaizi Park for the first time. The track integrates more than 10 types of terrain, including flat ground, slopes, curves, and narrow sections, with 12 left-turn lanes and 10 right-turn lanes, including curves of nearly 90 degrees, which will greatly test the robot's path planning and dynamic balance capabilities.

So falling down in the middle of the run is basically a daily occurrence.

Another memorable scene on the track: a human runner and a robot raced side-by-side, only for the robot to accelerate and overtake the human runner. The moment the image froze was comparable to a world-famous painting.

What was the human contestant thinking the moment he looked at the robot?

The humanoid robots that started in the middle and later stages of the race were a completely different story. They swayed slowly, as if they were under the influence of alcohol, focusing on simply finishing the race healthily. The most heart-wrenching scene came just before the finish line. The robot, which had maintained its pace the whole way, suddenly collapsed to the ground with the finish line within reach.

An emergency rescue operation was launched, and the stretcher bearer had already rushed onto the track. With the help of engineers, it shakily got up again and crossed the finish line. Thus, the first robot to cross the finish line was born.

Because of the staggered start, the robots in front had already finished their runs, while the teams behind hadn't even started yet. A tiny robot with a cute nose appeared in the middle, holding a baby bottle, creating a cute track. Seriously, will a robot that short actually reach the finish line?

As the weather got hotter and hotter, the supply station, in addition to swapping batteries, also took on the task of physical cooling. Then, the top-tier robot made its appearance: a chicken-shaped robot from the Greater Bay Area also joined the operation.

Wow, it even stopped running and ate on its own initiative, quite the posture, I suspect it triggered a human-friendly protocol.

There's also a robot that suddenly stops swinging its arm halfway through its run, hovering with one arm in mid-air, resembling a martial arts hero like Yang Guo fighting alone against the world.

After crossing the finish line, one robot, perhaps overly excited, dashed into the adjacent green belt and was eventually carried out by paramedics. Another competitor performed a serpentine sprint before the finish line, a classic moment that was nowhere to be seen.

There were also heartwarming moments after the race. After working so hard, the engineers took a beautiful photo with their robot.

By the way, this is what today's finisher's medal looks like.

The design is inspired by a metallic mecha style, with a forged texture throughout, sharp lines, and an intricate structure. Even more ingenious is the central unfoldable section; when pulled open, the entire medal transforms into a three-dimensional miniature humanoid robot.

The finisher's trophy looks like this.

Before the competition began, the most frequently asked question from netizens was: Why do robots have to look like humans?

The reason humanoid robots insist on standing upright on two legs is that all the infrastructure of human society is designed according to the human form. If a humanoid robot is to enter a factory or a home, a humanoid structure adapted to the physical world would actually be more suitable.

I understand the principle, but after seeing today's track, I have a bold idea: to make the robot run faster, why not just give it two wheels?

Yes, it's this one.

If we take it a step further, with four wheels and a streamlined body, the speed would definitely increase to another level. See, it's already become a car. So, forget it; we have to walk the path on our own two legs.

Today, those steel figures that stumble, stage accidents, and crash headfirst into the green belt on the track represent the most clumsy form of humanoid robots, and may be the last time they still make a fool of themselves.

As for the pointlessness of competitions like humanoid robot running, our film has already provided the answer: robots that can run are useless. Robots that can do kung fu might also be of little use. But when a humanoid robot that can do kung fu runs up to you at 8 m/s to challenge you to a duel, it becomes very useful.

#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.

The ability to generate an interactive 3D world by taking a series of photos is nothing new. The real challenge is fitting this vast world into the average person's mobile browser.

Just now, World Labs, the AI ​​world model company under Fei-Fei Li, released and open-sourced a new achievement: Spark 2.0.

This dynamic 3D Gaussian point cloud (3DGS) rendering engine, designed specifically for web applications, is gradually making it a reality to smoothly run ultra-large 3D scenes with hundreds of millions of particles in browsers on any device.

Why is it so difficult to fit a 3D world with hundreds of millions of particles into a mobile phone?

You may have heard of "3D Gaussian Splatting," or 3DGS for short. In short, it's a technology that transforms real-world scenes into interactive 3D content without the need for modeling; it can be generated simply by taking a series of photos.

Unlike traditional 3D modeling which uses triangular facets, 3DGS uses millions of translucent colored ellipsoids, each called a "splat".

Each splat is not a simple point, but an ellipsoid with a complete "personality". It records its position in space, the length of the radii of the three axes, the orientation angle, the RGB color value, and the transparency.

The most crucial property is transparency. It determines the weight of the splat's influence on its surroundings when stacked. If you plot the spatial density of a single splat, you'll get a Gaussian curve: the center is the densest, gradually blurring outwards, with the edges naturally merging into the background.

It is this method of layering "soft boundaries" that allows millions of splats to be stacked together to present the texture of brick walls, the translucency of leaves, and the reflection of glass, rather than the plastic texture of a bunch of hard-edged triangles.

The results are excellent, and the amount of information is substantial. A high-quality 3DGS scan scene often contains tens of millions of splats, easily exceeding 1 GB in file size.

This presents a thorny problem: the upper limit for smooth rendering on a regular mobile phone is about 1 million to 5 million splats, which is an order of magnitude lower than the 40 million splats of high-quality scans.

Existing renderers also cannot correctly render multiple scanned objects in the same scene. They can only render one at a time, or the objects are out of order and "stuck" on each other's surfaces, making it look like a mess.

Thus, Spark was born. According to the official blog, Spark was initially an internal tool used by World Labs. World Labs needed to display 3DGS-generated worlds on a webpage, but existing renderers had inherent limitations: some could only render single objects, some relied on WebGPU (which many devices did not yet support), and some did not support dynamic animations.

After comparing several options, they decided to build one themselves.

Their chosen technology platform is THREE.js, the most popular 3D framework on the web, running on WebGL2 and covering almost all modern devices. The core rendering logic proceeds in three steps: first, a global splat list is generated across objects on the GPU; then, the splats are uniformly sorted from farthest to nearest; and finally, all splats are rendered at once.

"Global sorting" may sound unremarkable, but it's actually the key to allowing multiple 3DGS objects to coexist in the same scene without clipping into each other. Spark also provides a GPU processing pipeline on top of this, allowing users to perform custom operations on each splat, such as recoloring, adjusting opacity, and dynamic animations, using GLSL or implementing it with a node graph like Blender.

Version 1.0 solved the problem of multi-object rendering, but scenarios with 40 million splats remained an insurmountable hurdle. This led to Spark 2.0.

Make the device always render only the amount of information that is "necessary".

Spark 2.0 is built on a combination of three technologies: Level of Detail (LoD), progressive streaming, and virtual memory management. While each of these technologies has precedents, their combined strength enables the smooth rendering of billions of splatter instances in mobile browsers.

1. Continuous LoD Tree: Using resources wisely.

Level of Detail (LoD) is already a well-established concept in the gaming industry. It uses thousands of triangles for nearby trees and only a few dozen for distant trees, allocating resources as needed to save computational power. Unreal Engine's Nanite system follows the same approach, linking triangle detail to view distance and automatically scaling.

Spark 2.0 moved the same logic to Splat, making it even more thorough.

Switching between several discrete versions can easily cause visual "jumps". Spark's approach is to build a complete "continuous LoD tree". Each internal node is an approximate version of its child splats, which converges upwards layer by layer and finally reaches the root node, which is the coarsest single splat in the entire scene.

During rendering, the system dynamically draws a line on the tree based on the current viewpoint, capturing the lower-level details in the area closer to the viewpoint and the higher-level coarse-grained details in the distance.

The entire process is constrained by a fixed splat budget: approximately 500,000 for mobile and 2.5 million for desktop. The total number of splats in the scene doesn't matter; as long as the actual number of GPUs deployed remains stable within the budget, the frame rate will naturally remain stable.

On top of that, Spark introduces "foveated rendering," which allocates more of the rendering budget to the direction you are looking, automatically narrowing the details of edges and areas behind you. This effect is especially noticeable on VR devices, where eye tracking is usually required to achieve it. Spark approximates this with a fixed cone-shaped region, and it works just as well.

2. Brand new .RAD format: "Streaming" like watching short videos.

The rendering efficiency issue has been resolved, but the transmission efficiency issue remains equally challenging. There are two existing 3DGS file formats: .PLY and .SPZ. The former is uncompressed, and a 10M splat can reach a whopping 2.3 GB. Although it allows for simultaneous downloading and display, the file size is simply unbearable.

The latter uses columnar storage and Gzip compression, compressing the same amount of data to 200-250 MB. The cost is that the entire file must be downloaded before it can be displayed, because the attributes of each splat are scattered throughout the file, and the complete content cannot be pieced together if any part is missing.

To achieve the best of both worlds, Spark 2.0 introduced a new format, .RAD (RADiance fields). It splits the splat data into independent blocks of 64K splats each, compresses them separately, and records the byte offset of each block in the file header, allowing random access to any block.

The first block is always the coarsest 64K splats of the entire scene; once downloaded, the scene outline is immediately visible. Afterward, the system determines which areas require the most refinement based on the viewpoint, prioritizing the fetching of those corresponding data blocks, gradually revealing details from a blurry image. Three parallel Web Worker threads synchronously fetch and decode in the background, ensuring that details follow you wherever you go.

3. GPU Virtual Memory: Storing unlimited space within limited video memory.

Streaming solves the bandwidth problem, but the hard limit on GPU memory remains a tough nut to crack. Mobile browsers have strict constraints on video memory, making it impossible to fit a scene with 40 million splatters.

Spark 2.0 borrows from the operating system's virtual memory mechanism to address this issue.

The system allocates a fixed memory pool on the GPU, with a maximum of 16 million splats, and uses a page table to record which .RAD data blocks are currently residing on the GPU. When a specific region needs to be rendered…

When the memory is full, the corresponding block is loaded into the memory. When the memory is full, the old blocks are swapped out in the order of least used blocks.

Thanks to this mechanism, 3DGS scenes from different sources can share the same memory pool. Theoretically, as long as the network speed is sufficient, countless independent scanned scenes can be seamlessly stitched together to form an infinitely large world.

A single link delivers to the world

Following the release of Spark 2.0, Fei-Fei Li immediately stated publicly, "Spark 2.0 can now smoothly play over 100 million splat objects on any device. I am very honored to contribute to the open-source ecosystem of web-based 3DGS rendering."

She didn't emphasize "what she accomplished," but rather focused on "what she contributed to the open-source community." This statement is intriguing. 3DGS rendering is a rapidly evolving field; a single company cannot drive the entire ecosystem forward. Open source is the right approach to accelerate this process.

Judging from existing case studies, developers are indeed using Spark to explore various avenues. Webby Award winner James C. Kane independently developed a multiplayer spaceship shooting game called Starspeed.

The entire game scene is built from over 100 million splats, and comes with 10 synthesized wave-style original soundtracks, all of which are streamed in .RAD format through the browser. The stunning sci-fi environment can be played directly in a webpage.

In the art direction, there's Hugues Bruyère's *Dormant Memories*. He's the co-founder of the interactive experience studio Dpt. This series juxtaposes 3D scans of real locations with imagined spaces to create explorable, interactive environments. The boundary between reality and fiction becomes blurred in the splat graininess, which is surprisingly fitting.

Ryu Fujiwara from Hololive's Spatial Information Technology Department used Spark to render multiple large-scale real-world capture scenes, with a maximum of 40 million splats per scene, which ran smoothly on smartphones, Quest, and Vision Pro.

These diverse attempts by developers fully demonstrate Spark 2.0's powerful capabilities across different devices and application scenarios. But this is only half the story.

For a company dedicated to building an "AI world model," Fei-Fei Li's team's ambitions extend far beyond simply providing an open-source rendering tool. If Spark 2.0 solved the final delivery challenge of "how to watch smoothly," then their real trump card is how to create this 3D content with low barriers to entry.

Spark and World Labs’ creation platform Marble are deeply integrated. You can generate a 3D world in Marble using a line of text or an image, combine multiple worlds into a larger scene using Marble Studio, and then render it into a shareable web experience using Spark. This process is already working.

Technological advancements often prioritize "unlimited improvement," but sometimes "good enough" might be the most suitable approach. Spark 2.0 tells just such a story.

The idea of ​​having a device render only "just enough" splats at any given time is not new, but implementing it in web rendering while maintaining compatibility with mobile browsers and Quest presents real engineering challenges at every step.

When AI can generate infinitely large 3D worlds, what medium will we use to deliver them to every ordinary person? Now, World Labs' latest answer is a web link.

Here is the blog address:

https://www.worldlabs.ai/blog/spark-2.0

#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.

In the past few years, China's new energy vehicles have achieved remarkable progress in comfort, intelligence, and technology, making outstanding achievements that have attracted worldwide attention.

However, when it comes to the ultimate high-performance flagship mid-to-large SUV, the first thing that comes to mind is often the classic icons of the gasoline-powered car era—BMW M, Mercedes-AMG, and Porsche.

As we all know, high performance has always been the "universal language" of ultra-luxury car brands around the world, showcasing a car company's car manufacturing expertise, technological accumulation, and brand culture.

Therefore, as the wheels of history turn into the new energy era, the Chinese automotive industry also needs a "performance icon" that can represent the world's top standards. Thus, standing at the historical juncture of Geely Group's 40th anniversary and nearly 30 years of car manufacturing, the Jike 8X was born.

According to Jike, high-performance luxury SUVs in the past often came with compromises: powerful performance but difficulty in achieving ultimate handling and comfort; strong off-road capability but cumbersome when navigating the city; and a rigid body but lagging behind in intelligent interaction.

However, the birth of the Jike 8X is to challenge the technological limits of "wanting it all, wanting it all, and wanting it all," achieving the strongest performance in the fields of electric drive, chassis, safety, and intelligence.

The Jike 8X was launched in four versions, with a starting price of 356,800 yuan.

Performance flagship

The Jike 8X is equipped with the SEP (Super Electric Hybrid) system, which adopts a 900V hybrid high-voltage architecture.

Under this architecture, the ultimate Dawn Edition has a combined maximum power of 1030kW, more than 1400 horsepower, and is matched with a 2.0T hybrid-specific engine with a maximum power of 205kW, and a P1 generator with a peak power of 145kW.

The JK8X Wraith Edition, equipped with a three-motor system, accelerates from 0 to 100 km/h in just 2.96 seconds, making it one of the very few mid-to-large-sized hybrid SUVs to break into the "2-second club" for 0-100 km/h acceleration. The regular dual-motor version also achieves 0-100 km/h acceleration in 3.7 seconds.

Of course, in addition to explosive power, energy replenishment and sustained performance are also important indicators for plug-in hybrid vehicles.

The Jike 8X is equipped with a large-capacity battery pack that supports 6C supercharging. Thanks to its 900V architecture and discharge capability, the vehicle can still maintain high power output even when the battery is depleted. According to Jike, even when the battery is depleted, the Jike 8X's 0-100 km/h acceleration performance is still superior to traditional gasoline-powered performance benchmarks such as the BMW X5M and Porsche Cayenne.

In aggressive driving scenarios, the XGIMI 8X uses four motors (front and rear) to work together to distribute the battery power supply pressure, and with the generator providing real-time energy replenishment, the power performance under high-intensity output is more stable and continuous.

Depending on the configuration, the JK8X offers two battery options: 55.1kWh and 70kWh, with pure electric ranges of 256km, 257km, and 328km respectively. The new car also supports 900V 6C super-fast charging, which can charge from 20% to 80% in just about 9 minutes at room temperature.

Despite its curb weight of nearly 3 tons, the XGIMI 8X can still achieve a fuel consumption of approximately 7L/100km under WLTC conditions when the battery is depleted, with a combined range of over 1200km.

Meanwhile, the Jike 8X adopts a closed dual-chamber air suspension and a dual-valve CCD electromagnetic damping system, which are rare in its class, and introduces an active anti-roll bar previously only seen in high-end performance cars costing millions. Thanks to this mechanical combination, the body roll angle can be suppressed to within 2 degrees when cornering at high speeds of 80km/h.

Combining the vast AI digital chassis and the Dinghai intelligent hub, the Jike 8X can utilize the vehicle's computing resources to connect the previously independent powertrain, suspension system, braking, steering, and driver assistance systems at the underlying level. This enables real-time perception and precise execution. In the event of extremely dangerous emergencies such as a high-speed tire blowout, the system can instantly increase the stiffness of the air springs by more than 30% and quickly lower the vehicle's center of gravity to maximize the vehicle's handling stability and occupant safety.

Furthermore, this chassis system not only serves on-road driving but also caters to the needs of unpaved road scenarios. For complex off-road conditions such as sand, grass, rocks, and snow, the system can adaptively switch modes and distribute power, enabling this flagship SUV, positioned for on-road performance, to also handle various all-terrain challenges.

Security Flagship

In the high-end mid-to-large SUV segment, safety performance has always been a crucial cornerstone for measuring a product's core competitiveness. Leveraging Geely Group's extensive automotive manufacturing experience, the Jike 8X features high-level redundancy designs in both passive bodywork and active safety features.

In terms of passive body structure, the XGIMI 8X adopts an industry-rare "integrated double-door ring" architecture. This design effectively reduces structural weaknesses caused by traditional splicing processes, allowing the A-pillar, B-pillar, C-pillar, and the lower door sill beam to better cooperate in bearing forces and disperse collision energy.

Meanwhile, the vehicle features a pioneering sandwich design in the critical A/B pillar area, with a hot air expansion tube embedded inside as the core support, boasting a tensile strength of up to 2000MPa. Combined with JK's self-developed one-piece die-cast rear body technology, the vehicle's torsional stiffness reaches 42,400 N·m/deg, providing the passenger compartment with extremely high physical compressive strength.

The JK8X also integrates chassis hardware with passive safety. When the system senses an unavoidable side collision, its active anti-roll bar can urgently raise the body on the impact side within 0.7 seconds, forcing the vehicle to withstand the impact force with the strongest door sill beams.

To address the core issue of battery safety in new energy vehicles, the JK8X has constructed a three-dimensional protection system known as "985," consisting of 9 layers on the sides, 8 layers on the bottom, and 5 layers on the top. To mitigate the risk of chassis bottoming out during daily driving, a multi-functional composite high-strength plate is installed at the bottom of the battery pack. This plate not only disperses the impact of sharp external objects but also acts as an internal buffer, maximizing the physical safety of the battery cells.

In addition, the entire Xkk8X series comes standard with advanced GAEB (Intelligent Automatic Emergency Braking) and GAES (Intelligent Automatic Emergency Steering) systems, which still have excellent recognition and response capabilities even at high speeds of up to 130km/h.

In addition, the vehicle is equipped with an SOS emergency stop assist function. By monitoring the driver's vital signs and fatigue status in real time, the system will activate the driver's senses through multiple levels of sound and light when an abnormality is detected. If the driver remains unresponsive, the vehicle will automatically take over and find an opportunity to safely pull over to the side of the road.

Smart flagship

At the level of electronic and electrical architecture, the Jike 8X attempts to improve the overall vehicle's intelligence ceiling by reconstructing the underlying logic.

The vehicle is the first to be equipped with the WAM World Behavior Model, which breaks down the barriers between various electronic modules in traditional automobiles. It integrates and connects the underlying data of the driver assistance domain, intelligent cockpit domain, chassis domain, and power domain, thereby forming a central computing hub with global scheduling capabilities.

Based on this cross-domain integration architecture, Jike 8X has launched an intelligent agent called "Super EVA" in the human-computer interaction layer.

In terms of functionality, Super EVA attempts to break away from the passive execution mode of traditional in-vehicle voice assistants that follow a "question and answer" approach. With the support of large-scale model algorithms, it can combine vehicle environmental perception and driving decisions to achieve advanced interaction with a certain level of scene understanding and proactive collaboration.

Meanwhile, the Jike 8X relies on the Smart AI Agent architecture WAM model to make extensive optimizations in path planning and scene perception, and has implemented advanced functions such as gesture summoning and AI destination navigation, which can improve the efficiency and convenience of passage in complex scenarios.

To meet the needs of different levels of intelligent driving, the Jike 8X offers two hardware versions: H7 and H9.

The H7 is equipped with a single NVIDIA Thor chip and more than 30 high-precision sensing elements on the exterior of the vehicle. The H9 is upgraded to dual NVIDIA Thor chips, with a combined computing power of up to 1400 TOPS, and is equipped with up to 5 LiDARs, forming a "triple 360-degree" global environmental perception system.

Elegant flagship

In the engineering and development of traditional automobiles, there is often a trade-off between mechanical performance and ride comfort: in pursuit of ultimate power and handling, large high-performance powertrains and complex suspension structures can easily encroach on cabin space.

Faced with this industry pain point, the JK8X did not choose to simply increase the size of the vehicle body, but instead tried to rely on the underlying integration capabilities of the SEA S architecture to release interior space through efficient system integration and layout optimization.

Official data shows that the effective interior area of ​​the Jike 8X cabin reaches 5.93 square meters, with an average floor area of ​​over 1.18 square meters per person.

The second row of the vehicle is equipped with front and rear sliding rails with a length of 150mm, which allows users to flexibly adjust the rear legroom and cargo capacity according to actual travel scenarios.

Thanks to this, even an adult who is 1.8 meters tall can still have more than 200 millimeters of legroom when sitting in the back row.

In terms of storage capacity, the Jike 8X's trunk has a standard volume of 1133 liters, which is among the leading levels in its class. Its interior depth is even deep enough to fit a standard stroller without folding it.

In terms of ergonomics and comfort features, the Xkk8X has been designed with differentiated features to meet the needs of occupants in different positions.

To match its high-performance driving attributes, the driver's seat incorporates dynamic side wing support. When the vehicle is performing actions such as fast cornering, lane changing, or rapid acceleration, the seat side wings can respond and tighten rapidly within 0.1 seconds, providing the driver with more stable lateral body support.

In the rear, the Jike 8X introduces the "Eames"-style reclining chairs previously featured in the Jike 009. These chairs boast a class-leading maximum backrest tilt of 137 degrees, complemented by extra-large thigh rests and footrests that can fold down 90 degrees, enhancing comfort for long journeys. Furthermore, all seats come standard with a 22-point zone massage system.

The rear ceiling features a 17-inch OLED true-color ceiling-mounted screen, paired with a Dolby Atmos sound system to create a multi-dimensional in-car theater. When playing media content, this system can work in real-time with the car's audio system, seat vibration, air conditioning fan speed, ambient lighting, and fragrance system to provide a more immersive audio-visual experience.

To enhance the convenience and sense of occasion in daily use, the Jike 8X is equipped with electric foot pedals and electric doors, and supports the convenient function of "automatic door closing when the driver presses the brake pedal".

Meanwhile, the vehicle's exterior lighting system incorporates intelligent interactive light signals. In addition to providing a welcome light carpet and adaptive lighting for curves at night, the system can also actively project corresponding warning lights in specific scenarios such as "yielding to pedestrians at zebra crossings".

Looking at the current Chinese new energy luxury SUV market, product lines often inevitably fall into a homogenized competition focused on "family comfort".

In this environment, the Jike 8X chose a different path, launching a combination of "ultimate driving control + flagship luxury".

This positioning, which emphasizes personal driving pleasure and hardcore mechanical quality, is extremely rare in the current market segment. There are very few competing products on the market that can directly rival it in terms of size, performance, and luxury.

Meanwhile, the division of labor between the Jike 8X and Jike 9X is also very clear: the 9X is larger and more comfort-oriented, while the 8X is more sporty and compact, which can theoretically cover different target user groups.

In China's new energy vehicle market, where competition is intensifying and product definitions are becoming increasingly similar, choosing a differentiated path from "purely household" requires not only brand courage but also groundbreaking product strength as a fundamental support.

With a starting price of 356,800 yuan, and its robust configuration in areas such as the three-electric system, mechanical chassis, and advanced intelligent cockpit, the JK8X has demonstrated a high level of hardware sincerity and "cost-effectiveness".

#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.

News worth watching even on the weekend

DeepSeek reportedly launches its first funding round, with a valuation of at least $10 billion.

According to The Information, DeepSeek is seeking its first external funding round, aiming to raise at least $300 million at a valuation exceeding $10 billion.

Sources familiar with the matter revealed that DeepSeek is in contact with investors to bolster its funding reserves. It is understood that DeepSeek was previously exclusively funded by the hedge fund Magic Square Quant, and has repeatedly rejected investment offers from top venture capital firms and tech giants.

The report points out that founder and CEO Liang Wenfeng's change in funding strategy is mainly aimed at coping with the high cost of model development by increasing investment in computing power and improving compensation to cope with fierce competition for talent and technology.

According to reports, since the release of the R1 model in early 2025, DeepSeek has not launched a new generation of models and has recently faced the loss of key researchers. Luo Fuli, a core contributor to the V3 model, has joined Xiaomi, while another core researcher, Guo Daya, has moved to ByteDance with a high salary.

In addition, the V4 model, originally scheduled for release in February this year, has been delayed several times due to engineering and technical challenges such as out-of-the-box adaptation of Huawei chips.

Related reading: On the eve of the V4 release, DeepSeek was reported to have secured its first round of funding, with a valuation exceeding $10 billion.

Sora's creator officially announced his resignation.

Just now, Bill Peebles, the creator of Sora, officially announced his departure; on the same day, Kevin Weil, the former chief product officer of OpenAI, also posted: Today is my last day at OpenAI.

Related reading: Breaking news! Sora's creator announces his resignation! OpenAI is in complete chaos.

In terms of core project development, Peebles previously led the development of Sora, a high-fidelity 1080p multi-lens video generation model; while the "OpenAI for Science" project team that Weil was in charge of before his departure has been split up.

Amid executive personnel changes, OpenAI is undergoing intensive internal restructuring. According to foreign media reports, the Prism team, which focuses on scientific workbench products, has been merged into the Codex AI programming product division.

The company's current strategic focus has been narrowed to enterprise customers and programming business, and it plans to integrate several independent applications, including Prism and Atlas, into a single platform to improve the commercialization efficiency for professional workflows.

It is understood that prior to this large-scale business contraction, OpenAI had announced the shutdown of the Sora standalone application a month ago, citing "lack of a sustainable business model" as the official reason.

Under the continued pressure from competitor Anthropic in the enterprise market, OpenAI is accelerating the divestiture of high-investment, low-conversion-rate exploratory projects to address increasingly pressing revenue and commercialization metrics.

China's mobile phone market in the first quarter: Huawei and Apple saw growth, while Xiaomi experienced the largest decline.

Yesterday, market research firm Counterpoint Research released its latest research report:

In the first quarter of 2026, shipments of smartphones in China declined by 4% year-on-year. Huawei regained the top spot with a 20% market share, while Apple became the fastest-growing leading brand with a 20% year-on-year increase.

Specifically:

• Huawei achieved a 2% year-on-year growth in the first quarter, with a market share of 20%, marking a new high for single-quarter market share since the fourth quarter of 2020. The report points out that the stable supply of the Mate 80 and Enjoy 90 series, as well as the cost buffer provided by the local supply chain, were key factors supporting its rise to the top.
• Apple ranked second, with its shipments increasing by 20% year-on-year in the first quarter, driven by the continued strong sales of the iPhone 17 series and subsidies and promotions. It was the fastest growing brand among the top six manufacturers.
• Xiaomi's shipments declined by 35% year-on-year in the first quarter, ranking fifth, due to the underperformance of its core models and a conservative pricing strategy.

Counterpoint senior analyst Ivan Lam pointed out that soaring component costs have weakened the pull effect of Lunar New Year promotions and pushed up retail prices, but the high-end market remains resilient, driven by imaging, foldable screens and AI agent functions.

It is worth noting that analysts predict that high storage costs will persist throughout 2026, leading to a "double whammy" of shrinking shipments and thinning profit margins for terminal manufacturers.

OpenAI invests a whopping 20 billion yuan in Nvidia as a "backup plan".

According to The Information, OpenAI has agreed to pay over $20 billion (approximately RMB 136.4 billion) to AI chipmaker Cerebras over the next three years for computing power services in exchange for a stake in the latter.

According to multiple sources familiar with the matter, OpenAI's commitment of $20 billion in this agreement is double the amount previously rumored.

In exchange, OpenAI will acquire a minority stake in Cerebras through warrants; if OpenAI's final expenditure reaches $30 billion, its stake could be as high as 10%.

In addition, OpenAI has agreed to provide Cerebras with approximately $1 billion in dedicated funding to develop and operate data centers that run OpenAI's artificial intelligence products.

In February of this year, OpenAI released GPT-5.3-Codex-Spark, its first model designed specifically for real-time programming. This model is also the first result of OpenAI's collaboration with Cerebras.

It is understood that the GPT-5.3-Codex-Spark provided by OpenAI runs on Cerebras' Wafer Scale Engine 3.

It's worth mentioning that the WSE-3 has 4 trillion transistors and 900,000 AI cores, which is 19 times more transistors and 28 times more computing power than the NVIDIA B200. In addition, thanks to the fact that the chip is made directly from the whole wafer, the WSE-3 has 206 times faster memory bandwidth than NVIDIA's latest generation NVLink.

Lei Jun: Xiaomi will not produce cars priced below 100,000 yuan in the next few years.

At 6:30 a.m. yesterday, Xiaomi founder Lei Jun launched a live broadcast as scheduled to test the long-distance battery life of the new generation Xiaomi SU7 from Beijing to Shanghai, charging only once during the journey.

At 21:40 last night, Lei Jun and his team successfully arrived at the Xiaomi 5G Future Center construction site in Jinqiao, Shanghai. The entire journey took 15 hours, with an actual driving distance of 1,313 kilometers. Upon arrival, the vehicle still had 19 kilometers of range remaining.

During the live broadcast, Lei Jun responded to several hot topics. Regarding netizens' concerns about whether Xiaomi would launch a car priced under 100,000 yuan, he stated: "To achieve true intelligence in electric vehicles, it's difficult to control costs below 100,000 yuan. A highly intelligent car will be very difficult to produce for 100,000 yuan within the next 10 years."

He also stated that Xiaomi's goal is to become one of the world's top five car manufacturers, producing cars comparable to Porsche and Tesla, and driving China's automotive industry towards high-end development.

During the live stream, Lei Jun also addressed the long-standing label of "marketing guru" attached to him. He stated that this label is "a conspiracy" intended to make people believe that Xiaomi's cars rely on marketing rather than product quality. "They are praising me on the surface, but actually discrediting me."

According to 36Kr Auto, Xiaomi Group announced via internal email on the same day:

Hu Zhengnan has been appointed Vice President of Xiaomi Group and CTO of the Automotive Division; Song Gang has been appointed Vice President and Chief of Staff of Xiaomi Group's Automotive Division. This marks the first time Xiaomi Auto has established a CTO position since its founding in 2021.

It is understood that both newly appointed executives have over twenty years of experience in the automotive industry, with backgrounds spanning both traditional automakers and the new energy vehicle sector.

• Hu Zhengnan entered the automotive industry in 1997 and has nearly 30 years of experience. He is one of the most representative vehicle engineering experts in the generation of China's automotive industry that has risen independently.
• Song Gang joined Tesla's Shanghai Gigafactory in 2018 and completed the mass production and delivery of Model 3 in the same year the factory was built (2019).

The report also pointed out that, to date, Xiaomi's automotive R&D staff number close to 10,000.

iOS 27 AI Feature Preview: Improved Visual Intelligence

According to MacRumors, Apple will introduce four new Apple Intelligence features in iOS 27.

According to the backend code confirmed by MacRumors, this update mainly covers three core modules: visual intelligence, wallet, and browser.

• Visual intelligence: Supports scanning food nutrition labels and directly syncing them to the health app; automatically identifies and extracts phone numbers and addresses from printed materials and enters them into the address book.
• Apple Wallet: The app will leverage AI vision technology to allow users to generate digital vouchers directly by scanning physical event tickets or membership cards.
• Safari browser: Introduces an AI-powered intelligent categorization feature that automatically names "tab groups" based on the current webpage content.

It is reported that Apple will officially release iOS 27 at its Worldwide Developers Conference (WWDC26) in June this year.

Peng Zhihui: Zhiyuan Robotics is already self-sufficient and currently has no shortage of funds.

According to IT Home, at the APC2026 Zhiyuan Partner Conference held yesterday, Peng Zhihui, co-founder, president, and CTO of Zhiyuan Robotics, pointed out in a media briefing:

In 2025, the company's revenue exceeded 1.05 billion yuan, and this year it may see several times more growth. We are currently commercializing very quickly and are able to generate our own revenue. Frankly speaking, the company is not short of money at present.

Peng Zhihui said that Zhiyuan's overall operation rhythm and development plan are doing very well, and it knows which direction it should go.

It was also revealed that Logic will spin off some of its key business units for financing , and its valuation has exceeded $3 billion in a very short time, with a growth rate even faster than when Logic started.

In addition, Peng Zhihui also said at the meeting that Unitree is a remarkable company. "It has been established for ten years and has done a very good job in the field of physical products. Its product scale and efficiency are very high, and it is very worthy of our learning ."

Regarding the Tesla Optimus, Peng Zhihui said that judging from their promotional videos, the effect is very good, but since it has not yet been mass-produced, it is not appropriate to make an evaluation at this time .

It points out that overseas countries have an advantage in original innovation, but China may be faster in terms of engineering implementation and commercialization.

Last year, the automotive industry's profit margin was only 4.1%, with three companies selling cars at a loss, while Seres made nearly 10,000 yuan in profit per vehicle.

According to CBN, the domestic automobile industry saw record-breaking production and sales last year, but fierce price competition continued to squeeze automakers' profit margins.

The report states that after compiling data on the price and profit per vehicle of 12 major listed automakers, it was found that the profitability of each automaker varied significantly, with none of the companies achieving a profit per vehicle exceeding 10,000 yuan.

Seres ranked first in terms of average price and profit per vehicle: the average price per vehicle was approximately 304,800 yuan, and the profit per vehicle was 9,936 yuan; while Chery and BYD ranked second and third in terms of profit per vehicle, at 7,228 yuan (average price per vehicle 103,500 yuan) and 6,398 yuan (average price per vehicle 140,900 yuan) respectively.

Among the emerging electric vehicle manufacturers, only Li Auto achieved an average vehicle price of over 250,000 yuan and a profit of 5,857 yuan per vehicle. NIO and XPeng both incurred losses: NIO's average vehicle price was over 250,000 yuan, with a loss of approximately 38,000 yuan per vehicle (based on Non-GAAP); XPeng's average vehicle price was approximately 159,200 yuan, with an adjusted loss of 1,071 yuan per vehicle.

From an industry-wide perspective, data from the China Passenger Car Association shows that last year, my country's automobile production and sales increased by 10.4% and 9.4% year-on-year, respectively, with total revenue reaching 11.18 trillion yuan, costs reaching 9.85 trillion yuan, and profits of only 461 billion yuan, with a profit margin of 4.1%, which is at a historical low and lower than the national industrial average profit margin of 5.31%.

Honda suspends some gasoline vehicle production plants in China

According to Reuters, Honda plans to close at least one of its joint venture gasoline vehicle factories in China this year in response to fierce competition in China's new energy vehicle market and declining sales in China.

Sources familiar with the matter revealed that Honda will officially cease production at one of its joint venture gasoline vehicle plants with GAC Group in June this year, and is evaluating suspending operations at another joint venture plant with Dongfeng Motor Group next year.

Currently, Honda's two joint ventures in China each own two internal combustion engine plants with an annual production capacity of 240,000 vehicles and one smaller electric vehicle plant.

The report points out that if each of the two joint ventures closes one gasoline-powered vehicle factory, Honda's annual gasoline-powered vehicle production capacity in China will be halved from 960,000 vehicles to 480,000 vehicles, and its total annual production capacity will also be reduced sharply from 1.2 million vehicles to about 720,000 vehicles.

Leapmotor Senior Vice President: We definitely won't lose money building cars.

According to IT Home, after the D19 launch event on the evening of April 16, Cao Li, Senior Vice President of Leapmotor, discussed the pricing of the Leapmotor D19 in a media interview.

Cao Li stated that after the release of the D19, Leapmotor's "core" remains unchanged in everyone's eyes—always focusing on making solid products. Whether it's the A series priced below 100,000 yuan or the luxury D series models priced close to 300,000 yuan, Leapmotor adheres to cost-based pricing, allowing customers to purchase products with better quality and higher configurations at the same price.

He also expressed his hope that everyone would see that Leapmotor can not only produce mass-market cars with monthly sales of tens of thousands of units, but also produce top-of-the-line models that can compete with luxury cars costing millions of yuan, and he hopes that this luxury car can sell tens of thousands of units per month.

Regarding the pricing of the D19, he emphasized that Leapmotor will definitely not lose money in manufacturing cars and will maintain a reasonable gross profit margin. The reason why it can maintain an affordable price with high configuration is due to efforts in all aspects, and the core lies in cost control in many areas.

In addition, Zhu Jiang, the founder of Leapmotor, also mentioned at the press conference that the technical foundation supporting the starting price of RMB 219,800 for the (D19) comes from Leapmotor's long-term self-developed system across all domains.

According to currently available information, Leapmotor has independently developed and manufactured 65% of the core components of its vehicles. Around its vehicle assembly plant, Leapmotor has built 17 component factories, producing everything from battery packs and electric drive systems to cockpit domain controllers, lighting components, and thermal management modules.

Geely recalls multiple Polestar 4 vehicles due to potential thermal runaway risk.

According to information from the State Administration for Market Regulation, Zhejiang Geely Automobile Co., Ltd. has filed a recall plan with the State Administration for Market Regulation in accordance with the requirements of the "Regulations on the Recall of Defective Automobile Products" and the "Implementation Measures for the Regulations on the Recall of Defective Automobile Products".

Recall number S2026M0045V: Starting from May 16, 2026, a total of 1,473 domestically produced Polestar 4 pure electric vehicles equipped with 86kWh power batteries, manufactured between November 16, 2023 and May 24, 2024, will be recalled.

It is understood that some vehicles within the scope of this recall may experience an abnormal increase in the internal resistance of the high-voltage power battery due to manufacturing inconsistencies in the components. This could lead to a decline in the performance of some power batteries and, in extreme cases, thermal runaway of the power battery, posing a safety hazard.

The recall involves replacing all battery modules in the affected vehicles free of charge to eliminate safety hazards. Polestar advises customers not to charge the high-voltage battery to more than 70% before replacing the battery module to reduce the risk of thermal runaway.

Seres joins BMW and Mercedes-Benz's supercharging joint venture in China

Yesterday, E-Charge, the supercharging joint venture established by BMW and Mercedes-Benz in China, announced that Seres has officially joined as a new investor, holding an equal stake with the existing shareholders. The three shareholders each hold 33.3% of the joint venture's shares.

Through this equity investment, Wenjie will participate in and support the construction of Yianqi's luxury supercharging infrastructure.

At the cooperation level, the three parties will jointly explore development opportunities for YiAnQi, focusing on expanding network coverage, key regional layout, and service innovation. In addition to enjoying YiAnQi's high-quality basic services, BMW, Wenjie, and Mercedes-Benz users will also receive exclusive brand benefits such as scheduled charging and priority power allocation.

London AI Conference: Software Engineers Are Becoming "Mid-Level Managers in AI"

According to Business Insider, engineers and industry professionals from companies such as Google, Anthropic, and OpenAI gathered at the AI ​​Engineers Europe conference in London recently to discuss the management and deployment of AI agents.

At the start of the conference, OpenAI engineer Ryan Lopopolo pointed out that the nature of software engineers' work will have fundamentally changed by the end of 2025.

With the rapid evolution of AI tools, the core responsibility of engineers has shifted from writing code themselves to "guiding and unblocking" AI agents—telling them where to go, what to look at, and correcting errors when they occur.

David Soria Parra of Anthropic stated on stage, "Last year was a year of exploration, and this year is the year to truly put these agents into production."

He predicts that the application scenarios of AI agents will soon expand from the field of programming to a wider range of knowledge-based work such as financial analysis and marketing, and the era of "general-purpose intelligent agents" is coming.

At the conference, "guardrails" and "context engineering" were frequently mentioned terms. The latter refers to using carefully designed prompts and context to enable intelligent agents to perform better with less computational power. Attendees generally agreed that 2026 will be a crucial year for AI agents to move from experimentation to large-scale deployment.

However, Mario Zechner, founder of Coding Agent Pi, offered a more cautious perspective. He pointed out that the training data for AI agents comes from the internet, which is rife with low-quality code.

He advised engineers to "use it with caution and not let the agent make decisions for you," emphasizing that "all the decisions it makes are learned from the Internet."

OnePlus Ace 6 Ultra to be unveiled next week

Yesterday, OnePlus announced that it will officially launch the Ace 6 Ultra next week, along with two accessories: a game controller and an external cooler.

According to reports, the game controller, once installed, can provide a handheld console-like control experience for the device, and is equipped with trigger buttons operated by the index finger; when the external heat sink is connected, performance limitations can be removed, and the device can enter "Berserk Mode".

In terms of hardware specifications, the OnePlus Ace 6 Ultra is equipped with a MediaTek Dimensity 9500 SoC, a 6.78-inch 1.5K + 165Hz high refresh rate flat screen; the front camera is 16MP, and the rear camera uses a dual-camera combination of a 50MP main camera + an 8MP secondary camera.

Acer unveils several new robot products

On April 17, the AI ​​Platform Partner Conference (APC 2026) was officially held, and four new ontology products and six AI models were launched at the same time. The AIMA (AI Machine Architecture) full-stack ecosystem technology system was also unveiled for the first time.

Robot body products:

• Expedition A3 – the industry's most capable full-size humanoid robot this year: the highest thrust-to-weight ratio in the industry, achieving greater mobility and flexibility in full-size than half-size; dual battery packs support hot-swapping, allowing for 8-10 hours of continuous operation without battery replacement; the industry's first full-size humanoid robot natively supporting interactive intelligent deployment.
• Lingxi X3: The first new-modal humanoid interactive service terminal that combines intelligence, anthropomorphism, fun, security and ease of use.
• The Genius G2 series features: Air for extreme lightweight design, matching human joint movement and supporting human-machine integrated data acquisition mode; and Max for greater load control.
• The Kootop D2 series quadruped robots are equipped with adaptive dynamic balancing, full-scene environment understanding, autonomous task planning, real-time anomaly handling, and natural language interaction capabilities.

Six models of three types of intelligence:

• Motion Intelligence: Two major base models will be launched: a full-body motion control base model that supports sensor-control fusion (achieving adaptive motion control through environmental perception), and a generative motion control base model (generating arbitrary movements in real time through multimodal interaction without pre-programming).
• Interactive Intelligence: Building upon the already widely deployed WITA model, the industry's first end-to-end embodied multimodal interactive model, WITA Omni 1.0, will be released in the third quarter. This model retains information such as dialogue emotion, context, tone, and environment, enabling natural and human-like emotional interaction and supporting mid-conversation interruption and correction.
• Task Intelligence: In the third quarter, the GO-3 model will be launched, which integrates the ViLLA architecture and the world model architecture. It has planning and deduction capabilities as well as reasoning and execution capabilities for complex tasks. The data scale is tens to hundreds of times that of GO-2, and we strive for significant breakthroughs in success rate, generalization and long-term task processing.

Bose SoundLink series launches new "Sunset Peach" colorway.

Bose recently launched a new color for 2026, "Sunset Peach," designed specifically for the Bose SoundLink Micro II, Bose SoundLink Flex II, and Bose SoundLink Plus Bluetooth speakers.

This "Sunset Peach" collection captures the bright and cheerful atmosphere unique to the hazy warm season. This soft coral orange hue blends creativity and vitality, inspired by fresh peaches and the sensory warmth they evoke. It aims to bring an accessible touch of spring vitality to consumers who pursue individual expression and quality of life.

Three new Bluetooth speakers in the "Sunset Peach" color are available for purchase starting today at Bose official stores, official online platforms, and authorized dealer stores.

Canva AI 2.0 Released: Generate Complete Designs in One Sentence

Yesterday, Canva officially released the Canva AI 2.0 update, which the company calls "the biggest transformation since bringing design from complex desktop software to the browser."

After the Canva AI 2.0 update, users can access the platform's full suite of tools through a unified dialogue interface, describe their needs in natural language, and automatically generate complete, "readily usable" content.

Canva AI 2.0 also introduces a persistent memory mechanism that learns users' creative habits over time, automatically maintaining brand style and visual consistency. The newly added "object-based intelligent editing" feature allows for precise adjustments to specific elements in the design via text prompts without affecting other parts.

In addition, Canva Code adds support for HTML import, and the platform has also launched a unified connector interface that can be integrated with third-party tools such as Slack, Gmail, Google Drive, and Calendar.

Canva AI 2.0 will be available to Pro (US$120 per year) and higher plan users, while free users can use basic AI features.

Zhipu AutoClaw: Enabling Self-Evolution

Yesterday, AutoClaw, a subsidiary of Zhipu, officially launched its self-evolution mechanism and Skill Store, aiming to solve the pain point of "forgetfulness" in AI agents.

Under the new mechanism, after each round of dialogue, AutoClaw will automatically scan the dialogue content, identify the user's corrections, preferences, or experience with task pitfalls, and present them to the user for approval in the form of an "evolution request" card. After user confirmation, the relevant content will be written into the model's memory and become a persistent capability.

The Skill Store and Zhipu's self-developed GLM Office Skills have been launched simultaneously, covering five modules: PPT, DOCX, XLSX, PDF, and Charts. After integration, users can generate various types of files, including PPT presentations and lecture notes, with a single command.

A Silicon Valley startup is working on a hat that lets you type with your "thoughts".

According to WIRED, Silicon Valley startup Sabi recently announced that it is developing a non-invasive brain-computer interface (BCI) device that resembles a beanie, aiming to allow ordinary users to type on a screen simply by "thinking". The company's initial target typing speed is about 30 words per minute, and it expects to continue to improve as users spend more time using the device.

This device is based on electroencephalography (EEG) technology. It collects electrical signals from the brain through metal sensors attached to the scalp, thereby decoding the user's "inner language." The cap contains 70,000 to 100,000 miniature sensors to compensate for signal attenuation caused by the skull as a barrier in wearable devices.

Sabi CEO Rahul Chhabra stated that the first product is expected to be officially released by the end of this year, and a baseball cap version is also under development simultaneously.

The official team behind "Honor of Kings World" apologized and announced several optimization plans.

Recently, the development team of "Honor of Kings World" released its first-week report, outlining the progress of optimizations for several issues raised by players.

In response to numerous user experience issues, such as the inability to skip cutscenes and a slow start, the development team apologized to players and stated that they have fully implemented a "rapid response iteration" mechanism, directly translating player feedback into patch schedules.

Regarding the highly requested minimap feature, the development team stated that the minimap was not designed previously due to considerations of guiding players to adapt to the "above-ground + underground" three-dimensional map structure, but admitted that this design was "wishful thinking." The minimap is now part of the development plan and is expected to be released in version S1.

Regarding the issue of the "phantom" social function interfering with the sense of immersion, the development team stated that they will rethink the design logic and promised that "social interaction in 'Honor of Kings World' will not be forced," and players can switch freely at any time.

Luckin Coffee will launch bottled coffee, priced at 6-7 yuan.

According to LatePost, Luckin Coffee will launch ready-to-drink bottled products at the end of April, initially offering three products: Classic Americano, Grapefruit Americano, and Coconut Latte, priced at 6-7 yuan per bottle.

Sources close to the distribution channels confirmed that a batch of distributors signed contracts in early April and will begin receiving and distributing goods to retailers within the month. Luckin Coffee has not restricted sales channels; convenience stores, snack hypermarkets, and mom-and-pop shops are all permitted to sell. Luckin Coffee officials stated that their primary focus is on establishing a presence in modern channels such as convenience stores, supermarkets, and vending machines.

On the production side, Luckin Coffee does not have its own production lines. Its ready-to-drink products are manufactured by Huizhou Tongshi Enterprise Co., Ltd., which belongs to Uni-President Group and has long provided OEM services for brands such as Red Bull, Wanglaoji, and Nongfu Spring. The client is Luckin Coffee Technology (Hainan) Co., Ltd.

Seven leading food delivery platforms were fined 3.597 billion yuan for "ghost delivery" practices.

According to Xinhua News Agency, the State Administration for Market Regulation yesterday issued administrative penalty decisions against seven e-commerce platforms—Pinduoduo, Meituan, JD.com, Ele.me (now Taobao Flash Sale), Douyin, Taobao, and Tmall—for their "ghost delivery" scandal. The decisions ordered the seven platforms to correct their illegal activities and imposed fines and confiscations totaling 3.597 billion yuan.

An investigation revealed that the aforementioned seven e-commerce platforms engaged in three main violations:

• The company failed to strictly review and verify the licenses of food business operators entering the network and failed to fulfill its legal obligation to conduct qualification reviews.
• Signed cooperation agreements with order-switching platforms, knowingly or should have known that the order-switching behavior infringed upon consumers' legitimate rights and interests, but failed to take necessary measures;
• The platform's legal representative and food safety director failed to fully perform their food safety management responsibilities.

All seven platforms have been ordered to suspend adding new cake shops for periods ranging from three to nine months. Following the launch of the investigation, the State Administration for Market Regulation immediately instructed the platforms to rectify the situation. Currently, all seven e-commerce platforms have removed the unverified "ghost shops" and ceased their cooperation with relevant order-switching platforms regarding food and beverage order transfers.

"Front Teeth" is scheduled for release on May 1st.

The realistic film "Front Tooth" officially announced its release date yesterday, and will be released nationwide on May 1st.

The film is written and directed by emerging director Li Xin, and stars Zhang Yu and Chen Haoyu. The story begins with a minor accident in life: the male protagonist, Li Weiyang (played by Zhang Yu), is involved in a collision while riding a motorcycle, resulting in his girlfriend, Shen Qing (played by Chen Haoyu), losing two front teeth.

Despite being short of money, Li Weiyang insisted on getting the best front teeth for his girlfriend, which forced him to deal with the perpetrator and the insurance company, completely disrupting his originally frugal but peaceful life.

It's the weekend!

One Fun Thing | Gaode Maps' robot dog video revealed: It can help buy milk tea

According to Sina Tech, several netizens recently posted videos showing a Gaode Maps robot dog crossing the street and buying milk tea near a Gaode Maps building. The videos show the Gaode Maps logo printed on the robot dog.

According to the report, a representative from AutoNavi recently stated that AutoNavi has made in-depth investments in the field of embodied intelligence and is actively exploring hardware product forms such as quadruped robots and humanoid robots. The first quadruped robot is expected to be released soon.

According to videos taken by netizens, the Gaode robot dog waits for the green light when crossing the road and will actively avoid pedestrians when it encounters them.

In addition, in a milk tea shop, the Gaode robot dog was able to enter the shop on its own and tell the staff which number of milk tea it wanted. After the staff handed the milk tea to the robot dog, it was able to leave the shop on its own and walk back to Gaode headquarters.

Public information shows that AutoNavi established the Embodied Business Department in January this year. It has released multiple Embodied models and achieved state-of-the-art (SOTA) status in more than ten authoritative evaluations.

What to watch this weekend? | *The Peanuts Movie*

"The Peanuts Movie" is adapted from the classic comic strip "Peanuts" created by Charles M. Schultz in 1950.

The film unfolds with a dual narrative structure: the main storyline follows the consistently unlucky but kind-hearted Charlie Brown (voiced by Noah Schnapp) as he falls for the "red-haired girl" (voiced by Francesca Capaldi) and searches for his self-worth amidst repeated setbacks;

The subplot features Snoopy (voiced by Bill Mortens) as an ace pilot, engaging in aerial battles with his nemesis, the "Red Baron," in a fantasy world, and encountering his dream girl, "Fifi" (voiced by Kristen Kenonise).

In terms of visual presentation, the film uses CG technology to highly reproduce the iconic simple lines, exaggerated body movements and two-dimensional style of Schultz's comics, which has been praised by the audience as a "perfect recreation".

A Guide to Buying Books Without Reading | "The True Self": When AI Resurrects the Dead, How Can Humanity Confront Its True Self?

"The True Heart" is a novel by Japanese author Keiichiro Hirano, set in the near future with highly advanced technology.

Six months after his mother's accidental death, the protagonist, Sakuya Ishikawa, uses VF (Virtual Figure) technology to bring his mother back to life in a virtual reality image, attempting to explore the true reason why she longed to "die freely" before her death.

However, does the mother "resurrected" through data processing truly possess a "heart"? As Saku piece together an image of his mother from the people she encountered during her lifetime, one drastically different from his memories, the novel's questions deepen:

When "reunion" becomes achievable, will people get closer to the truth about their loved ones, or will they sink deeper into denial of loss?

The novel incorporates real-world issues such as the gap between rich and poor, loneliness, meaninglessness, and the pain of losing loved ones into its narrative. In an era where technology can constantly postpone farewells, it re-examines the meaning of loss, love, and life.

Game Recommendation | Diablo IV to be Free to Claim

Yesterday, the official Chinese server announcement for Diablo IV revealed that the game will be available for free for a limited time, with the new expansion pack launching simultaneously on April 28th.

• The base game is free to claim for a limited time and will be permanently added to your library: New players can visit the official page from 0:00 on April 28th to 23:59 on May 27th to claim the base game of Diablo IV for free (originally priced at 128 yuan, now 0 yuan), and it will be permanently added to their library after claiming.
• Exclusive 2400 Platinum Coins Reward for Veteran Players: All players who owned any version of Diablo IV in the Chinese server before April 28th can receive 2400 Platinum Coins for free from 00:00 on April 28th to 23:59 on May 27th.
• All players, new and old, will receive a brand-new Mythic skin upon reaching the maximum level. After the release of the expansion pack "Lord of Hatred", they will receive a brand-new limited Mythic skin.

#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.

The Leapmotor D19 officially launched tonight. When the final price appeared on the screen, this long-awaited large SUV finally unleashed its most lethal move.

The starting price is 219,800 yuan, while the top-of-the-line three-motor performance version is priced at 269,800 yuan.

I don't know if this price intimidated the competitors, but it certainly intimidated me.

In the current automotive market, large SUVs of similar size and configuration generally have a starting price of over 250,000 yuan. Consumers have long been accustomed to paying an extra premium for the market position of traditional luxury brands or leading emerging car manufacturers.

Leapmotor skipped the trial phase and brought a flagship model with impressive capabilities into the competitive 250,000 yuan price range.

At the press conference, Zhu Jiangming, founder of Leapmotor, set a tone for the D19:

The D19 is a top-of-the-line, well-rounded vehicle with no weaknesses. We use the best core components and the best suppliers.

For a full-size SUV, filling the configuration list may not be difficult; the difficulty lies in offering such a restrained price after filling it.

Overwhelm your opponents with parameters

Knowing the starting price of 219,800 yuan and then reading the configuration list of the D19 will create a strong sense of incongruity.

Intelligent cockpit and driver assistance are the most challenging and expensive parts of new car manufacturing. In this area, Leapmotor has not made any compromises in terms of tiered configuration reduction, and the Qualcomm 8797 chip is the basic standard configuration for the entire series.

If you have a greater desire for computing power, the top-of-the-line model will also offer two 8797 Extreme Edition chips, with a total computing power of an astonishing 1280 TOPS.

Matching the computing power is an ample memory pool, with the cockpit having 48GB of RAM and 128GB of storage space, and the driver assistance system being equipped with a separate hardware combination of 64GB and 128GB.

The LiDAR is also standard across the entire range, using the Hesai AT128. When the D19 is delivered, it supports 44 advanced driver assistance functions, including automatic U-turns on narrow roads without markings, 120-meter reversing with tracking, and memory parking across floors.

The "Super Zero" voice assistant, which integrates the Qianwen big data model, also conveniently solves some everyday travel pain points. For example, when the parking space is too narrow to get into the car, you can stand outside the car and use gestures to bring it out.

The powertrain is another highlight of the D19. These days, it's common to equip range-extended vehicles with large batteries, and Leapmotor has made the D19 the range-extended model with the longest pure electric range.

The 80.3 kWh CTC battery pack, coupled with an 800V high-voltage platform, gives this range-extended vehicle a pure electric range of 500 kilometers, and a 15-minute charge can replenish half the battery. It can be driven purely on electric power for daily use, and fuel is used for long distances – a very powerful combination.

The pure electric version is the first to use CATL's 115 kWh super hybrid battery cells, combined with a full-stack 1000V high-voltage architecture. The CLTC has a pure electric range of 720 kilometers, and a 15-minute charge can add 350 kilometers of range.

The 269,800 yuan tri-motor version naturally focuses more on performance, with a combined power of 540kW (about 724 horsepower), a 0-100 km/h acceleration time of only 3.94 seconds, and the ability to perform a compass turn with a minimum radius of 3.6 meters.

Meanwhile, the 50,500 Nm torsional stiffness provides an extremely high ceiling for chassis tuning. The front all-aluminum double wishbone suspension paired with the rear five-link suspension, along with the dual-chamber closed air suspension and CDC continuously variable damping shock absorbers, constitute the hardware foundation of the D19.

The LMC 2.0 fusion control system is responsible for directing this hardware. It can support stable control even with a dual-wheel tire blowout at speeds up to 120 km/h, and can also adjust the suspension stiffness by anticipating road undulations using cameras. Addressing the issue of motion sickness in large vehicles, the system's built-in anti-motion sickness algorithm improves the relief effect by 25.2%.

At the launch event, Cao Li, Vice President of Leapmotor Technology, used the analogy of a kitchen oxygen chamber and a double bed to summarize the lifestyle attributes of this car.

The D19 boasts an interior room utilization rate of up to 88%, offering 6-seater and 7-seater versions. The 7-seater model features a one-touch fold-down rear seat, instantly creating a 1.94-meter flat bed.

The 176-liter electric front trunk comes with a 220V power supply, which can be connected to an extension kit to create an outdoor cooking station; the center armrest box inside the vehicle houses an 8.1-liter automotive-grade refrigerator with heating and cooling, supporting a wide temperature range from -6 degrees Celsius to 50 degrees Celsius.

The most interesting feature is the automotive-grade "Forest Oxygen Chamber," which provides 8 liters of oxygen-rich air per minute and supports both diffusion and nasal inhalation modes. This oxygen supply system is perfect for those who enjoy challenging high-altitude road trips.

In areas with high-frequency contact between passengers and the vehicle, the D19 uses semi-aniline top-grain leather, and the second row features a 120° zero-gravity seat on the right side. With the double-layered soundproof glass and completely flat side windows closed, the 23-speaker Dolby Atmos system kicks in. From auditory to tactile, this car strives to create its own unique sense of luxury.

A high-level game that had to be played

At the press conference, Zhu Jiangming said that the technological foundation supporting the starting price of 219,800 yuan comes from Leapmotor's long-term self-developed system across all domains.

According to publicly available information, 65% of the core components of Leapmotor's vehicles are self-developed and self-manufactured. Around the vehicle assembly plant, Leapmotor has built 17 component factories nearby. From battery packs and electric drive systems to cockpit domain controllers, and then to lighting components and thermal management modules, almost all of these major components are processed on Leapmotor's own production lines.

The highly vertically integrated manufacturing model brings obvious business advantages. From parts rolling off the production line to assembly into the complete vehicle, the intermediate packaging, transportation, and inventory processes are significantly reduced. The profit margins that external suppliers would have needed to obtain are all converted into internal cost reductions by Leapmotor.

By controlling the underlying supply chain, Leapmotor has the confidence to absorb the hardware costs of higher-end models in the 200,000 yuan price range and has complete control over the pricing of the entire vehicle.

This is something Leapmotor has no choice but to do at present.

Based on the 2025 full-year financial report released last month, Leapmotor's overall market size has expanded to a rather astonishing scale. The company delivered nearly 600,000 new vehicles throughout the year, recording a net profit of 540 million yuan and achieving a turnaround from loss to profit. However, when broken down by individual vehicles, Leapmotor's current net profit per vehicle is only around 900 yuan.

In fact, Leapmotor's current sales base heavily relies on its B-series and C-series models, which target the lower-tier market. These high-volume products keep Leapmotor's overall average price per vehicle at just over 100,000 yuan. The 14.5% gross profit margin reported in the financial report represents the maximum profit margin achieved within the existing price range through its fully self-developed R&D model.

In the capital-intensive industry of car manufacturing, such meager profit per vehicle lacks the resilience to withstand drastic market fluctuations. While maintaining a low-price, high-volume strategy can indeed stabilize the basic sales volume, it cannot support the company's investment in advanced technologies for the future.

The procurement of dual Qualcomm 8797 chips, the algorithm evolution of the advanced VLA intelligent driving model, and the research and development of edge AI all require hundreds of millions of yuan in funding. Vehicles priced below 150,000 yuan can hardly sustain these cutting-edge and capital-intensive R&D projects. Leapmotor urgently needs to find a new engine for profit growth.

The D19 is tasked with this crucial mission. Leapmotor urgently needs to leverage this new flagship to raise the price point of its products to the 200,000 to 300,000 yuan range. Only by fully opening up profit margins within this new price range can the company maintain its long-term cash flow for technological research and development.

However, facing the high-end market of around 300,000, it is difficult to break through by simply relying on impressive parameters and solid financial calculations. Leapmotor also needs to establish a convincing value narrative.

Fei Xiang's appearance at the press conference perfectly echoed the brand's rebranding goals. This national-level artist didn't deliberately recite dry specifications; instead, he dedicated a significant portion of his presentation to the actual in-car experience. On stage, he meticulously described the fresh air provided by the automotive-grade oxygen chamber and the quality of music playback through the Dolby Atmos sound system.

Leapmotor hopes to cultivate a high-end image that emphasizes quality of life and riding experience through this approach.

To steal orders in the fiercely competitive large SUV market, Leapmotor must provide consumers with a rational reason to buy. It needs to convince its audience to accept a new consumer discourse centered on technology and practicality, guiding them away from the obsession with brand premiums and towards purchasing decisions driven by core technology and practical experience.

With an impressive configuration and a clear profit target, the D19 stormed into the 250,000 yuan market segment. The final result of this tough battle determined the D19's sales volume and also dictated the upward trajectory of Leapmotor's overall business model.

#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.