Pogbagram

Xiaomi YU7 GT and Xiaomi 17 Max officially released; Lei Jun: I strongly recommend you get a new phone now.

Xiaomi held a new product launch event last night to showcase its comprehensive ecosystem of people, cars, and homes , officially releasing the Xiaomi YU7 GT, YU7 Standard Edition, Xiaomi 17 Max, and several new wearable and IoT products.

Xiaomi YU7 GT:

• Prices start at 389,900 yuan, with the "GT Fully Equipped" version priced at 429,900 yuan;
• Equipped with a V6s Plus + V8s EVO dual-motor system, it boasts a maximum horsepower of 1003PS, a top speed of 300km/h, and a 0-100km/h acceleration time of 2.92s.
• It uses the same track-version braking system as the SU7 Ultra, equipped with carbon ceramic brake discs and Akebono six-piston front and four-piston rear high-performance calipers;
• Using an 897V silicon carbide high-voltage platform and a maximum charging rate of 5.2C, the CLTC has a range of 705km.

At the launch event, Lei Jun stated that the YU7 GT broke the Nürburgring SUV lap record with a time of 7 minutes and 22.755 seconds, 14 seconds faster than the previous record. Xiaomi also released a comparison of its cornering speed with the Cayenne Turbo GT, with a result of 32 wins, 17 losses, and 1 tie.

At the press conference, Lei Jun also compared the sales data of the YU7 and the Tesla Model Y, saying that in the past 10 months, the sales of the YU7 exceeded those of the Model Y in two months, and said that "it is not shameful to lose to the global sales champion".

Related Reading: First Look | Xiaomi YU7 GT Review: How Fast is the Fastest SUV on the Nürburgring?

Xiaomi 17 Max is positioned as a large-screen standard flagship:

• The 12GB + 256GB configuration is priced at 4799 yuan, while the top-of-the-line 16GB + 512GB configuration is priced at 5799 yuan.
• It features a fifth-generation Snapdragon 8 Ultra processor and a 6.9-inch ultra-large flat screen with a peak brightness of 3500 nits;
• It has a built-in 8000mAh Jinshajiang battery and supports 100W wired and 50W wireless charging.
• The body is 77.6mm wide, 8.2mm thick, and weighs 225g;
• The camera features a 200-megapixel Leica main sensor with a 1/1.4-inch sensor size, and is equipped with a large-sensor Leica periscope telephoto lens.

When announcing the price at the launch event, Lei Jun mentioned that given the current surge in memory prices, Xiaomi's pricing was still very competitive. He emphasized that, according to their predictions, memory prices will continue to rise over the next two years, forcing phone prices to follow suit, and phones may become increasingly expensive.

If you're planning to change your phone in the next year, I highly recommend you do it now.

Other new products include:

• The YU7 standard edition is back, priced at 233,500 yuan, with a CLTC range of 643km, and equipped with lidar, 4D millimeter-wave radar and 700TOPS of assisted driving computing power;
• Xiaomi's first ear clip-on earphones are priced at 799 yuan for the first time, and support intelligent translation of 21 languages, independent recording, and Apple Find My.
• The Xiaomi Mi Band 10 Pro will be available for sale starting at 399 yuan, featuring a 1.74-inch screen, dual-light dual-PD sensors, sleep HRV, 150+ sports modes, and up to 21 days of battery life.
• In the area of ​​IoT and automotive peripherals, Xiaomi also released products such as the Mijia Powerful Wind Pro Vertical Air Conditioner, the Central Air Conditioner Powerful Wind Duct Unit, the MiCare Home Appliance Protection Service, the 20kW DC Home Charging Pile, and the In-Vehicle Table.

OpenAI is expected to file its IPO prospectus as early as this week, aiming to become the first company to go public.

According to The Wall Street Journal, OpenAI plans to secretly file its IPO prospectus with the SEC as early as this Friday, aiming for a formal listing in September. Meanwhile, SpaceX filed new listing documents yesterday, and Anthropic disclosed better-than-expected profit prospects to investors. These three star AI companies are all moving towards the public market on the same day.

The competitive pressure from Anthropic should not be underestimated. It expects to achieve its first quarterly operating profit in the quarter ending in June, and according to Ramp's AI Index report, its enterprise paying customer share reached 34.4% in April, surpassing OpenAI's 32.3% for the first time.

Internally, OpenAI is divided on its IPO timeline: CEO Altman is pushing for a September IPO, while CFO Friyal is more cautious. Founding member Kapacsi announced yesterday that he is joining Anthropic, continuing the talent drain issue.

Former Jane Street engineer joins DeepSeek to advance coding agent

According to his LinkedIn profile, former TSY Capital co-founder Cui Tianyi joined DeepSeek in March of this year. Cui previously worked at Jane Street for nine years before co-founding Hong Kong-based quantitative trading firm TSY Capital in 2022.

According to the South China Morning Post, Cui Tianyi will be involved in AI harness-related work after joining the company. This involves packaging the underlying model into an agent product infrastructure that can perform tasks, call tools, and work continuously.

Nubia's president responds to Doubao's restaurant reservation blunder: Doubao phones really can…

Yesterday, Ni Fei, President of ZTE Terminal Business Unit and President of Nubia Technology Co., Ltd., released a video in response to the recent "Doubao AI restaurant reservation blunder".

The incident stemmed from a netizen's experience sharing their experience of using Doubao AI to book a restaurant: the AI ​​gave a virtual response in the conversation, which the user believed and went to the restaurant, only to be refused.

In the video, Ni Fei stated that although the Doubao app may not currently support actual reservation functions, the Nubia M153 equipped with the Doubao mobile assistant can perform such operations. He demonstrated this live in the video: issuing voice commands directly to the phone to filter nearby restaurants, make reservations for 6 PM the following day, and request a group of 5 people for dinner.

Doubao then selected a restaurant, confirmed the time and number of people based on her previous dietary preferences, and completed the reservation. Ni Fei said at the end of the video, "I believe that before long, everyone will feel that using Doubao to reserve a restaurant is just a common thing in life."

Tesla's supervised version of FSD will be launched in China.

This morning, Tesla officially announced the latest progress of its supervised version of FSD on Twitter, mentioning that the supervised version of FSD can be used in China.

According to multiple reports, the supervised version of FSD in Europe is currently only available on a subscription basis, and the one-time purchase option has been completely discontinued globally in February 2026. Lithuanian car owners pay €99 per month for the subscription; if they have already purchased the enhanced Autopilot version, the monthly fee can be as low as €49.

Tesla stated that the subscription price will be increased as the system is gradually upgraded to unsupervised autonomous driving.

According to IT Home, Tesla's recent recruitment page shows that it is recruiting Intelligent Driving Test (real vehicle testing) technicians in nine cities: Beijing, Shanghai, Tianjin, Chongqing, Guangzhou, Shenzhen, Chengdu, Suzhou, and Wuhan.

The job posting is for a research and development position in the Autopilot department, indicating that flexible travel to various locations both domestically and internationally is required to conduct field tests of vehicles on public roads, test tracks, and validation sites.

In addition, in April of this year, Tesla stated that it was working hard to launch intelligent driver assistance systems in the Chinese market as soon as possible.

Antigravity permanently triples Gemini limit

Antigravity team core member Varun Mohan announced yesterday that, in response to community feedback regarding rate limiting issues, the platform will permanently increase the rate limit for all paid Gemini models by 3 times and immediately reset the weekly quota for all users.

In his subsequent response, Mohan clarified that this 3-fold increase was a permanent adjustment, not a temporary measure.

Previously, several developers reported on social media that after the launch of Antigravity 2.0, the rate limit of the Gemini model was significantly tightened, and some users exhausted their weekly quota in about 2 hours.

CapCut announces integration with Gemini, enabling video editing via chat.

CapCut announced yesterday at X that it will partner with Google Gemini, allowing users to access CapCut's video and image editing functions directly within the Gemini app without having to switch to a separate app.

According to reports, once the collaboration is implemented, users will be able to use CapCut's creative editing capabilities in the Gemini chat interface to complete image and video processing, achieving a seamless workflow from creative conception to content production.

BMW Group appoints Fang Yuan as Vice President of Corporate Affairs for BMW China

According to Sina Auto, BMW Group announced today the appointment of Fang Yuan as Vice President of Corporate Affairs for BMW China, and concurrently as Vice President of Public Relations and Corporate Social Responsibility at BMW Brilliance. This appointment will take effect on June 1, 2026.

It is understood that Fang Yuan has over 25 years of experience in public relations and communications, with a long-standing presence in the automotive industry. Since joining BMW Brilliance Automotive Ltd. in 2012, she has held various management positions in areas such as product and technology communications, media relations, social media communications, and crisis management.

The report also pointed out that his rich experience in different business areas, his deep understanding of the company, and his high sense of responsibility will provide a solid foundation for him to fulfill his new responsibilities.

Lagou.com, once a leading online recruitment platform, has entered bankruptcy proceedings.

According to Jiemian News, Beijing Lagou Network Technology Co., Ltd., the operating entity of Lagou.com, an internet vertical recruitment platform, has entered bankruptcy reorganization proceedings.

This time, the company filed for bankruptcy on its own initiative. Jiemian News visited Lagou's official website and found that the site is still accessible, but many job postings and user comments are from six months ago. The 400 customer service hotline goes unanswered, the official Weibo and WeChat accounts have not been updated since March 2025, and the app has been removed from the Apple App Store.

Founded in 2013, Lagou.com initially focused on recruitment for internet companies and professionals. Eight months after its launch, it completed a $5 million Series A funding round, followed by a $25 million Series B funding round in 2014, at which point its valuation reached $150 million. In 2017, 51job made a strategic investment of $120 million, acquiring a 60% stake.

NetEase's net revenue in the first quarter was 30.6 billion yuan, with game revenue reaching 25.7 billion yuan.

NetEase released its Q1 2026 financial report yesterday, reporting net revenue of RMB 30.6 billion, a year-on-year increase of 6.1% and a quarter-on-quarter increase of RMB 27.5 billion; gross profit of RMB 21.2 billion, a year-on-year increase of 14.8%; and net profit attributable to shareholders of the company of RMB 10.7 billion.

• Net revenue from games and related value-added services was RMB 25.7 billion, representing a year-on-year increase of 6.9%, with online game operation revenue accounting for approximately 97.5% of the segment's net revenue.
• Youdao's net revenue was RMB 1.3 billion, up 3.8% year-on-year, but down quarter-on-quarter, mainly due to a decrease in revenue from learning services and smart hardware.
• NetEase Cloud Music's net revenue was 2 billion yuan, a year-on-year increase of 6.6%;
• Net revenue from innovation and other businesses was RMB 1.5 billion, a year-on-year decrease of 4.6%, with the quarter-on-quarter decrease mainly due to reduced revenue from e-commerce and advertising businesses.

Gaming remains the main focus of this quarter's financial report. NetEase stated that existing games such as the Fantasy Westward Journey series, Identity V, Egg Party, Justice Online mobile game, and Sixteen Sounds of Yan Yun maintained activity and revenue performance, while products such as Sixteen Sounds of Yan Yun and Marvel Cinematic Universe are also advancing their globalization.

Lenovo's revenue for fiscal year 2025/26 reached 589.9 billion yuan, with AI revenue doubling year-on-year.

Lenovo Group today announced its fourth quarter and full-year results for fiscal year 2025/26. Full-year revenue reached RMB 589.9 billion, a year-on-year increase of 20.3%; adjusted net profit increased by 42.1% year-on-year, setting a new record. According to HKFRS, fourth-quarter net profit was nearly six times that of the same period last year.

• Fourth-quarter revenue was nearly RMB 150 billion, a year-on-year increase of 27.1%, the highest growth rate in nearly 20 quarters;
• IDG (Intelligent Devices Group) reported revenue of RMB 101.2 billion in the fourth fiscal quarter, a 24% year-on-year increase; its full-year PC market share reached a record high of 24.2%.
• ISG (Infrastructure Solutions Business Group) reported revenue of RMB 39 billion in the fourth fiscal quarter, a year-on-year increase of 37%, achieving a return to profitability as expected, with an operating profit margin hitting a record high; full-year revenue exceeded RMB 136 billion, a year-on-year increase of 32%, achieving full-year profitability;
• SSG (Solutions Services Group) reported revenue of RMB 17.75 billion for the fourth fiscal quarter, marking the 20th consecutive quarter of double-digit growth, with an operating profit margin of 22.4%.
• AI-related revenue grew by 105% year-on-year for the full year; AI revenue accounted for 38% of total revenue in the fourth fiscal quarter.

The most significant structural change this fiscal year was ISG's return to profitability, with AI server revenue increasing by 50% year-on-year and generating an order backlog of over RMB 140 billion. Lenovo completed its acquisition of storage company Infinidat in April, further strengthening its enterprise-level storage capabilities.

CEO Yang Yuanqing stated that the company is confident in its goal of becoming a $100 billion enterprise within the next two years.

Cloudflare laid off 20% of its workforce; CEO says this will become a common phenomenon in the business world in the coming year.

According to The Wall Street Journal, Cloudflare CEO Matthew Prince published a commentary yesterday, revealing the decision-making logic behind the company's layoffs of more than 20%.

He stated that the layoffs were not due to business difficulties, characterizing them as a proactive business restructuring, and cited Peter Drucker's 1954 book "The Practice of Management" as an analytical framework.

Prince categorizes internal roles into three types: "Builders" are responsible for product development, "Salespeople" are responsible for market expansion, and "Measurers" encompass functions such as internal audit, finance, legal affairs, compliance, middle management, and operations.

He believes that builders and salespeople will not be replaced by AI —increased engineer productivity means companies will recruit more engineers, and sales positions rely on interpersonal trust, making them equally difficult to replace. "Measurers," however, are different; AI can continuously perform these tasks at a lower cost and with higher accuracy.

Based on this assessment, Cloudflare's layoffs primarily targeted the measurement team: middle management positions were significantly reduced, operational functions were consolidated into a unified team, and marketing and finance departments were both scaled back.

Prince concludes his article by saying that AI will not eliminate all jobs, but it will change the way every business operates—ultimately allowing humans to focus more on two things: creating and capturing value: building and selling.

The OPPO Find X9 Ultra Hasselblad Imaging Kit in "Polar Glacier" color is now available for purchase at 2499 yuan.

OPPO announced yesterday that the OPPO Find X9 Ultra Hasselblad Professional Imaging Accessory Kit in the "Polar Glacier" color is now officially on sale, priced at 2499 yuan; the "Tundra" color will be available on May 31.

The set includes an OPPO teleconverter hand rest, a professional teleconverter adapter ring and filter adapter ring, a custom professional photography shoulder strap, a teleconverter lens hood, a teleconverter carrying bag, and OPPO Hasselblad custom front and rear lens caps and lens decorative rings.

ThinkPad X13 Gen 7 Overseas Specifications Released

Lenovo has announced the specifications of the ThinkPad X13 Gen 7 laptop in overseas markets. The new machine can be equipped with AMD Gorgon Point and Intel Panther Lake platforms, and weighs about 930 grams.

• The Intel version starts with a Core Ultra 5 325 and can be upgraded to a Core Ultra 7 356H; the AMD version starts with a Ryzen AI 5 Pro 440 and can be upgraded to a Ryzen AI 7 Pro 450.
• Both Intel and AMD versions support 16GB or 32GB of LPDDR5X-8533 memory and up to 1TB of SSD storage; the AMD version offers an optional PCIe 5.0 SSD, while the Intel version supports a PCIe 4.0 SSD.
• 13-inch 1920 × 1200 IPS screen, maximum brightness of 400 nits, 60Hz refresh rate; offers 41Wh and 54.7Wh battery options;
• It features one USB-A 5Gbps port, two Thunderbolt 4 ports, one HDMI port, a 3.5mm headphone jack, and a Kensington lock slot. A smart card reader is optional.

In terms of pricing, the Intel version starts at AU$2,429 in the Australian market, which is approximately RMB 11,794; the AMD version starts at AU$2,220, which is approximately RMB 10,779.

Photoshop 27.7 adds local AI model removal.

According to 9to5Mac, Adobe released Photoshop 27.7 yesterday, which includes a new local AI model in the "Remove" tool, allowing objects to be removed on the device without having to upload edits to the cloud.

After this feature is launched, users can edit downloaded models locally, as well as cancel or remove them, and switch between processing in the cloud and on-device. Adobe help documentation indicates that Photoshop will automatically check hardware compatibility. However, removing models locally using AI has hardware limitations:

• Mac: M1 Pro ARM or equivalent Intel CPU, 8 cores, medium dedicated graphics card, ≥8 GB video memory, 16 GB RAM, macOS 14 or later;
• Windows: 8-core Intel Rocket Lake or equivalent CPU, dedicated graphics card such as NVIDIA RTX 3060+, ≥8 GB video memory, 16 GB RAM;

Tencent's Hy-MT2 translation model, supporting 33 languages, is now open source.

Tencent Hunyuan officially open-sourced the Hy-MT2 series of multilingual translation models yesterday, including three versions: 1.8B, 7B, and 30B-A3B (MoE architecture). It supports mutual translation between 33 languages ​​and 5 minority languages/dialects. The weights and code are now available on Hugging Face, ModelScope, and GitHub.

• Version 1.8B, after 1.25-bit ultra-low precision quantization, has a size of only about 440MB. It can perform local offline inference on mobile phone chips such as Apple, Qualcomm, and MediaTek, with an inference speed 1.5 times faster than the previous generation.
• The 7B and 30B-A3B versions achieved scores of 96.9% and 98.1% respectively on the FLORES-200 benchmark, surpassing open-source models of similar scale, such as DeepSeek-V4-Pro.
• In real-world business scenario testing, the GEMBA score of the 30B-A3B surpassed that of the Gemini 3.1 Pro, and its performance in vertical fields such as finance, politics, and education was among the best of mainstream translation models.

The "Tencent HyTranslate" WeChat mini-program, based on Hy-MT2, has been launched, supporting voice input, customizable translation styles, and offline inference.

xAI quietly released grok-build-0.1, focusing on intelligent agent programming scenarios.

xAI recently launched grok-build-0.1 model page in its official developer documentation, positioning it as a fast code model specifically trained for agentic coding, and it is currently in the early access phase.

According to the official documentation, grok-build-0.1 supports multimodal input of text and images, has a context window of 256k tokens, and natively possesses the ability to call tools, output structured data, and perform reasoning.

In terms of pricing, the base price for input tokens is $1/1M token, which drops to $0.20/1M token for hits after enabling prompt word caching, and the price for output tokens is $2/1M token.

ZCube Inference Cluster Network Architecture Announced

Yesterday, Zhipu announced its new inference cluster networking architecture, ZCube, stating that it has been validated in a production environment and addresses the structural network congestion problem in the separate deployment of large model PD (Prefill-Decode).

• Cluster inference throughput improved by 15%;
• The first token response tail latency (TTFT P99) decreased by 40.6%;
• The hardware cost of switches and optical modules has been reduced by one-third.

Cohere's open-source Command A+ and 218B MoE models are geared towards enterprise agents.

Cohere has officially announced the open-sourcing of Command A+, a MoE model with 218 billion parameters, activating approximately 25 billion parameters per inference, and licensed under the Apache 2.0 license.

The official blog states that Command A+ is designed for enterprise-level agents, complex inference, multimodal document processing, and data sovereignty scenarios. Cohere emphasizes that this model can be deployed on one B200 or two H100s and improves inference speed for text and multimodal inputs through optimizations for the MoE architecture.

Douyin e-commerce launches special campaign to regulate trendy toy blind boxes

Douyin e-commerce issued a statement yesterday, saying that the platform recently discovered that some merchants and influencers were engaging in illegal activities such as making the probability of drawing items opaque when selling and promoting trendy blind boxes, promoting products with uncertainty, inducing users to consume through gambling psychology, and guiding consumers to jump to third-party websites for points redemption.

The platform stated that such behavior harms consumer rights and disrupts the normal business operations of merchants. Douyin e-commerce will further intensify its governance efforts, taking measures such as product bans, business suspensions, deductions of security deposits, and store removal against violating merchants and influencers.

The typical violations listed this time include four categories: gambling-style marketing that allows users to exchange low-priced blind boxes for high-value goods; directing winners to third-party links, WeChat groups, and other channels outside the platform; converting blind box rewards into points and inducing continued participation; and selling goods with opaque probabilities in the form of "group buying," "black box," or "self-made blind boxes."

Douyin e-commerce requires merchants and creators to conduct self-checks to ensure that blind box products are clearly priced, that the drawing rules are clearly publicized, and that the probability information is true and transparent. They are also prohibited from guiding consumers to trade on third-party platforms or setting up gambling-like points or ranking mechanisms.

Global TV brand shipments grew by 3.3% year-on-year in the first quarter.

According to the latest research report from TrendForce, global TV brand shipments reached 47.12 million units in the first quarter of 2026, a year-on-year increase of 3.3%, but a decrease of 12.7% compared to the previous quarter.

The first quarter is typically a slow season for TV sales, and coupled with the reduction in trade-in subsidies, overall demand is conservative. TrendForce believes that the year-on-year growth in shipments this quarter is mainly due to cost pressures: demand for AI servers and high-end computing has driven up the tight supply of DRAM and NAND Flash, and the price of TV memory is expected to rise significantly by the end of 2025. Brand manufacturers are stocking up and pulling in inventory in advance to reduce subsequent cost pressures.

In terms of brand rankings, Samsung maintained its position as the world's number one in shipments in the first quarter, followed closely by TCL. Hisense, LG, and Xiaomi ranked third to fifth, respectively. Benefiting from its expansion in North America and emerging markets, as well as the promotion of Mini LED and large-size products, TCL shipped 7.68 million units in the first quarter, a year-on-year increase of 11.3%, the highest growth rate among the top five brands.

TrendForce predicts that global TV shipments will decline slightly by 1% year-on-year to 194.2 million units in 2026, amid continued increases in memory and material costs. Leading brands will continue to focus on Mini LED, large-screen, and high-end products, while other brands will shift more towards mid-to-large-screen models with FHD resolution.

Woodpecker Repair sparks controversy again after a toilet repair that initially cost 156 yuan was ultimately charged at 1780 yuan.

According to Sina Finance's "BUG" column, the home repair platform Woodpecker has once again sparked controversy due to an incident where a toilet repair quote was quoted at 156 yuan but the final payment was 1780 yuan.

According to reports, Ms. Zhang from Nanchang had a verbal agreement with the repairman for a repair fee of 156 yuan, but after the repairman arrived, she was told that additional charges would be incurred for replacing parts such as the motor, and she ended up paying 1,780 yuan to exchange for a new Jomoo brand toilet; her family later discovered that the same toilet was priced at about 550 yuan on Taobao.

After the incident escalated, Woodpecker's official customer service responded that the technician in question had violated regulations and that the platform would expedite the process. However, when the "BUG" column contacted Woodpecker's 400 customer service hotline, the representative stated that the platform's standard procedure involves technicians scanning a QR code to arrive at the customer's home, conducting the inspection, and then charging according to the price list. They claimed to be unaware of Ms. Zhang's incident.

Woodpecker was exposed on CCTV's 3.15 Gala in 2025 for "repairing problems where none exist, exaggerating minor issues, and overcharging." Data from the Black Cat Complaint Platform shows that Woodpecker Home Repair has accumulated 10,147 complaints, with 226 complaints in the past 30 days; compared to approximately 6,000 complaints when it was exposed on 3.15 in 2025, this represents an increase of about 69% in one year.

The sequel to Once Upon a Time in Hollywood is set to release on November 25th.

According to the blogger "Hollywood Watch," the sequel to Quentin Tarantino's "Once Upon a Time in Hollywood," titled "Clive Booth Continues His Adventures," is scheduled to premiere on Netflix on November 25th. The new film stars Brad Pitt, is directed by David Fincher, and is written by Quentin Tarantino.

This film revolves around Cliff Booth from "Once Upon a Time in Hollywood," serving as an extension of the original character. Brad Pitt previously won an Academy Award for Best Supporting Actor for his role in the film.

Eddy Cue, VP of Entertainment at Apple, won the Cannes Lions International Festival of Creativity's Person of the Year in Entertainment Award.

Apple’s senior vice president of services and health, Eddy Cue, was announced to be the 2026 “Entertainer of the Year” award at the Cannes Lions International Festival of Creativity in June.

The organizers stated that Cue has long been a driving force behind Apple's business in entertainment, media, and creative content. Apple has previously won awards at the Cannes Lions International Festival of Creativity, including the 2025 "Creative Marketer of the Year" award.

Cue is currently in charge of services such as Apple TV and Apple Music.

Volcano Engine launches Volcano Drama Creation 1.0, an AIGC short drama creation platform.

Volcano Engine announced yesterday the official launch of Volcano Drama Creation 1.0, a one-stop AIGC short drama creation platform. The platform is designed for the short drama industry's creation process and is compatible with models such as Seedance and Seedream. Volcano Engine claims it can shorten production cycles by more than 80%.

Volcano Drama Creation breaks down short drama production into stages such as script understanding, storyboard analysis, material generation, and strategic review, and grants users editing access to core stages. The platform has recently added features including prompt editing, an IP asset library, and team collaboration: the former adjusts visual details and style, the IP asset library covers both real and virtual avatars, and the team function supports multi-role collaboration and points allocation.

#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.

If you're planning to change your phone in the next year, I highly recommend doing so now.

This quote comes from Lei Jun's remarks at last night's launch event when he introduced the price of the Xiaomi 17 Max. It sounds like a sales pitch, but it also points out a rather helpless reality—in the next two years, memory costs will continue to rise, and mobile phone prices will most likely only get more and more expensive.

So, if you're looking to get a new phone, how about the brand new Xiaomi 17 Max?

In conclusion: This is an "ordinary" product. The configuration is basically as expected, except that the screen and battery are larger and the imaging configuration has been slightly modified.

This is not to say that it is not a good product. In fact, its experience is quite solid, and it is probably my favorite phone in the entire Xiaomi 17 series.

However, it would be more interesting to discuss why the Xiaomi 17 Max exists in the context of the broader environment than to discuss the product itself.

Large screen and super long battery life

Unlike the Redmi K90 Max, which focuses on performance, the "Max" in Xiaomi 17 Max has multiple meanings.

The most obvious difference is the 6.9-inch screen size, which is larger than the Xiaomi 17. However, it is not a simple proportional enlargement. Whether it is the appearance design or the configuration, its highlights are very prominent: huge battery capacity, 200-megapixel main camera and periscope telephoto lens.

Even though the design languages ​​of the entire Xiaomi 17 family (standard version, Max, Pro, Ultra) are quite different, the Xiaomi 17 Max is still the most out of place—its black square Deco design clearly belongs more with the Xiaomi 13-15.

In recent years, Xiaomi has put a lot of effort into the "feel" of its phones. The Xiaomi 17 Max continues the excellent feel of the Xiaomi 17, with a frosted glass back panel and a metal frame. The texture is quite delicate, the transition is smooth, and the weight distribution is relatively reasonable. It is much easier to hold than the Xiaomi 17 Pro Max with a back screen.

Xiaomi officially stated that the 1.5K screen of the Xiaomi 17 Max uses "Super Pixel" technology, which can display clearer textures and details. However, without a magnifying glass to look at the pixels, the difference is actually hard to perceive. On the contrary, the 6.9-inch narrow bezel screen brings a very obvious improvement in the viewing experience.

With the body size further increased, the Xiaomi 17 Max also managed to fit in a massive 8000mAh battery, making battery life another extremely prominent advantage.

During the week I used the Xiaomi 17 Max, I almost completely said goodbye to "battery anxiety": after two days of normal use, the battery still had about 15% left.

Moreover, the large battery did not bring any noticeable burden to charging. With 100W fast charging fully enabled, the phone can be charged from 12% to full in about 55 minutes, which is a satisfactory speed.

Finally, let's look at the performance. Although it uses the same Qualcomm Snapdragon 8 Elite Gen 5 processor as the Xiaomi 17, the larger body and more space for heat dissipation allow for more flexible performance scheduling. The Geekbench 6 score is slightly higher than the Xiaomi 17, but the difference is very small.

Footage with both selective omissions and juicy details

Besides the large screen and large battery, another highlight of the Xiaomi 17 Max is the adjustment of its imaging configuration, with two key features: a 200-megapixel main camera and a periscope telephoto lens.

Let's talk about the 200-megapixel main camera first. In terms of specific configuration, it has a 1/1.4-inch sensor and a large f/1.68 aperture, which is the same as the main camera of the vivo X300. In fact, its size is smaller than the 1/1.28-inch main camera of the Xiaomi 17.

Based on specifications and user experience, the main camera of the Xiaomi 17 Max and the Xiaomi 17 are evenly matched. In low-light scenes, the detail of the Xiaomi 17 Max's images is significantly weaker than that of the Xiaomi 17, but it is less prone to highlight clipping and overexposure issues like those found in the latter.

The 200-megapixel configuration, which even the Pro version lacks, is more of a sweet spot. The processing time of up to 10 seconds and the file size of up to 60MB make this feature unsuitable for everyday shooting. It is more suitable for use in some complex scenes and where lighting conditions are required.

What significantly impacts the user experience is that when reviewing 200-megapixel photos in the album, there's a wait for a clearer image to load each time you zoom in, as well as a waiting time after each shot—while Vivo, with the same 200-megapixel resolution, can already produce photos instantly and zoom in smoothly.

The periscope telephoto lens performs even better. Unlike the Xiaomi 17 Pro series, the Xiaomi 17 Max features a 3x telephoto lens, making it more suitable for everyday shooting compositions.

Even when magnified to 6x or higher, the Xiaomi 17 Max still performs very well, although the algorithmic effect becomes slightly more pronounced in low light.

The only drawback is that the ultra-wide-angle lens of the Xiaomi 17 Max is still the same "narrow ultra-wide" lens as the Xiaomi 17 series, with a focal length of only 17mm. The resulting images are not much different from those of the main camera, and there will be a noticeable "not wide enough" feeling.

As a user who isn't exactly a photography enthusiast, I prefer to use a telephoto lens for everyday shooting. Therefore, the imaging experience of the Xiaomi 17 Max is more comfortable than the standard version. The main camera also has a unique sweet spot of 200 million pixels, which is more in line with the needs of daily photography and light creative work.

Large-screen phones are back.

In the first half of last year, Xiaomi launched its "half-generation flagship" – the Xiaomi 15s Pro – whose biggest highlight was Xiaomi's first self-developed processor, the Xuanjie O1.

This year, the Xiaomi 17 Max has taken that spot.

If the Xiaomi 15s Pro is a phone that "chasees dreams," its most appealing aspect is that long-lost emotional value and idealistic spirit—

The Xiaomi 17 Max is a more "down-to-earth" phone. It doesn't rely on storytelling or emotional appeal, but wins people over with its powerful and robust configuration.

This shift will be particularly evident in 2026. So far this year, all new smartphones released, except for those from Apple and Huawei, have seen significant price increases due to soaring costs. For example, the Oppo Find X9s Pro, with the same specifications, is 1000 yuan more expensive than last year's Find X8s+.

When products become more expensive, to attract customers, they must have a very eye-catching advantage that makes consumers feel that their money is "worth it".

For example, the vivo Y600 Pro was originally a low-end product line. In previous generations, it tried to emulate the flagship X series in terms of appearance and configuration. However, this year it has switched to features such as a large battery and dust and water resistance, which is a more tangible advantage for consumers in the 2000 yuan price range.

The Xiaomi 17 Max, positioned as a flagship phone, follows a similar strategy: a large screen and a large battery provide a solid foundation for practicality, while sacrificing the size of the main camera sensor in favor of more perceptible "sweet spots" like 200-megapixel resolution and periscope telephoto lens to further support the price.

The Xiaomi 17 Max starts at 4799 yuan, and after the national subsidy, it's 4299 yuan. At this price, you get a Snapdragon flagship chip, an 8000mAh large battery, and a 6.9-inch large screen. Plus, it's a new phone from 2026. All these advantages combined contribute to the Xiaomi 17 Max's product strength.

These very "flashy" upgrades are designed to make you feel that this not-so-cheap phone is a worthwhile purchase.

Interestingly, after three years of sluggish sales, Apple finally discontinued the "large screen standard version" model last year.

The starkly contrasting choices made by Xiaomi and Apple reflect the completely different brand identities and strategic paths the two companies have taken.

Aside from a larger screen and a larger battery, the iPhone Plus doesn't offer any other upgrades and costs 1,000 yuan more, placing it between the standard and Pro versions.

For Apple, the iPhone Plus is more like a price anchor, further highlighting the value of the "Pro" version. They don't need to add more features to it, which would reduce the original high profit margin of the product and may also affect the sales of Pro products.

Xiaomi cannot match Apple's brand power, making it unable to further support a premium price. Coupled with the precedent of the iPhone Plus's lukewarm reception, Xiaomi needs to provide consumers with a more compelling reason to buy through more obvious features.

In a sense, the Xiaomi 17 Max is actually the "Xiaomi 17s", a half-generation updated flagship designed to refresh its presence – and the "large screen" is a very "present" selling point.

#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.

Five or six years ago, when a consumer had a budget of just over 100,000 yuan to buy a pure electric SUV, their only requirement might have been "reliability." As long as it drove well, had an acceptable exterior and interior, and sufficient range, that was enough. But now, with the same budget of over 100,000 yuan, consumers are no longer asking "whether it can be driven" but rather "how easy it is to use."

Between these two issues lies an entire industry evolution cycle.

In the early days of pure electric SUVs, users were most concerned about range, space and price; later, intelligent cockpit, driver assistance and chassis quality began to become the dividing line; today, a more basic question has returned to the forefront: if a pure electric vehicle is to truly enter the daily lives of families and young people, can it be like a gasoline car, with less planning and more spontaneity?

This is the question that the third-generation Yuan PLUS aims to answer. On May 21st, BYD officially launched the third-generation Yuan PLUS, with an official guide price of 119,900 yuan to 149,900 yuan. The new car offers two CLTC range versions with 540km and 630km ranges, all equipped with the second-generation blade battery and flash charging technology, and offers an optional "Eye of the Gods B" – laser-assisted driving version, also known as DiPilot 300.

The Yuan PLUS of the past was a calling card for BYD to open up the global market. Just as the Yuan Dynasty was the most extensive dynasty in Chinese history, the third-generation Yuan PLUS is more like a "value refill" for the mainstream market: it does not make any single breakthrough or extreme cost performance, but tries to provide relatively good performance in high-frequency perception aspects such as energy replenishment, battery, intelligent driving, space, chassis and safety, and then continue to become a blockbuster product.

All models are equipped with BYD's flash charging technology, fully charging in 10 minutes.

Even today, pure electric vehicle owners still experience some range anxiety. However, the anxiety isn't necessarily about the range itself, but rather about refueling. In the era of gasoline-powered cars, users rarely planned specifically "which gas station I'll refuel at." But for a long time, pure electric vehicle owners have needed to calculate routes, waiting times, charging station availability, and even consider temperature, power output, and queue conditions in advance. This psychological cost is a problem that pure electric vehicles must address as they further penetrate the mass market.

One of the most significant changes in the third-generation Yuan PLUS is the standard inclusion of BYD's flash charging technology across the entire range. BYD states that the new car can achieve a 10%-70% SOC charge in 5 minutes, a 10%-97% SOC charge in 9 minutes, and in a -30℃ low-temperature environment, a 20%-97% SOC charge in 12 minutes.

The experience of fully charging in about 10 minutes is quite similar to that of recharging a gasoline car. You can stop at a service area to charge, get out of the car to use the restroom, stretch your legs, take a short walk, and then return to find the car fully charged.

In the past, long range solved the problem of "whether you can get there," while fast charging solved the problem of "how long to wait after you arrive." When a pure electric SUV priced at just over 100,000 yuan offers both a 630km range and minute-level charging, it's no longer just a city commuter tool, but begins to possess stronger cross-city travel capabilities. For young families, this means less mileage calculation for weekend trips, holiday road trips, and impromptu business trips, and more spontaneous travel.

BYD has also improved the charging experience in detail. For example, the zero-gravity flash charging gun adopts a liquid cooling and suspension design to reduce the weight of the charging cable; and the contactless payment reduces fragmented actions such as scanning, waiting and paying.

The adoption of the second-generation blade battery represents another advancement. The third-generation Yuan PLUS is equipped with the second-generation blade battery, with capacities of 57.545kWh and 68.547kWh respectively. The official claim is that the energy density is more than 5% higher than the first generation, while safety, lifespan, and low-temperature fast charging capabilities are further enhanced. For mainstream-priced vehicles, battery technology upgrades are often more important than specifications, as they determine a car's real-world driving experience several years later.

From "good enough" to "good enough", A-segment SUVs can be small but cannot be lacking.

What makes the Yuan PLUS special compared to other electric vehicles is that it's not a niche model. As BYD's first global strategic passenger vehicle, the Yuan PLUS has achieved cumulative global sales of over 1.1 million units in four years since its launch, covering more than 100 countries and regions. Precisely because of its large scale and strong benchmark role, each upgrade is not just a facelift of a single model, but rather a migration of the mainstream A-class pure electric SUV standard.

The third-generation Yuan PLUS measures 4665×1895×1675mm with a wheelbase of 2770mm. Rather than simply making the car bigger, the Yuan PLUS focuses on maximizing space utilization. For example, it offers 39 storage spaces throughout the vehicle, including a phone holder, cup holder, and glasses case in the front, and a map pocket, B-pillar hooks, and seat drawers in the rear. The trunk also features a sunken design and side storage areas. It offers a 180L power front tailgate and a 750L rear trunk, along with a power tailgate, retractable cargo cover, and dual rear storage boxes, aiming to proactively address the fragmented, real, and sometimes awkward driving scenarios common with family users.

At this level, it even comes equipped with an in-car refrigerator that combines heating and cooling, an integrated tissue box, a "magic house" for the passenger side, and a threaded expansion interface for the driver's side. These features don't sound like the big items that traditional car engineers love to emphasize, but they are actually closer to the real lifestyle of young users: some want to display figurines, some want to hang action cameras, some want to prepare baby food for their children, and some just want a pack of tissues to stop rolling around in the car.

A 15.6-inch central control screen, a 12-inch W-HUD, an 8.8-inch LCD instrument panel, 50W wireless phone charging, 256-color music-inspired ambient lighting, a 16-speaker sound system, and a mobile phone NFC car key—these features constitute the basic digital experience. Meanwhile, the ever-changing themes, the My Little Pony co-branded theme, the Cyber ​​Wooden Fish, the in-car game center, and the interconnection of ecological devices give this cabin a stronger emotional value.

This approach is somewhat similar to today's smartphones, where hardware specifications are the basics, but what truly determines user stickiness is often the wallpaper, themes, plugins, ecosystem devices, and those subtle expressions of personalization.

Rear-wheel drive, five-link suspension, and the Yunlian-C – BYD hopes its major features will also be appealing.

In many people's minds, the core task of a 150,000-level SUV is practicality, while handling and chassis are often not the first priority.

The third-generation Yuan PLUS adopts a rear-engine, rear-wheel-drive layout, offering two motor versions: 200kW and 240kW. The highest-powered model accelerates from 0 to 100 km/h in 5.9 seconds. All models feature a front MacPherson strut and rear five-link independent suspension, and are equipped with a 4.7m turning radius. Higher-spec models also feature the Yunlian-C intelligent damping body control system, which can adjust damping in milliseconds based on road conditions and driving status.

Electric vehicles, with their fast motor response and low center of gravity, are inherently prone to creating a nimble feel. However, the true quality of the chassis often depends on the suspension structure, damping control, and vehicle posture management. Rear-wheel drive and a five-link suspension provide a better foundation, and the Yunlian-C attempts to adapt this foundation to more scenarios: smoother when going over speed bumps, more stable during sharp turns, acceleration, and braking, and reduced unnecessary swaying during long journeys.

Safety features are also a key focus of the third-generation Yuan PLUS. The new car is equipped with the iTAC 2.0 intelligent torque control system, which utilizes the motor's response speed and high-precision control to actively manage rear axle drive and feedback torque, reducing the risk of slippage when starting on low-friction surfaces, accelerating through corners, or when releasing the accelerator or applying light braking. It also features the TBC high-speed tire blowout stability system, providing tire blowout stability control within a speed range of 40km/h to 140km/h. Combined with seven airbags, driver fatigue monitoring, roadside assistance call, and a five-way dashcam, this forms a more comprehensive safety system.

In terms of driver assistance, the third-generation Yuan PLUS offers an optional "Eye of the Gods B" – a laser-based driver assistance system, equipped with Eye of the Gods 5.0. The official statement emphasizes that it adds reinforcement learning capabilities on the end-to-end basis, enabling functions such as city navigation, high-speed navigation, and intelligent parking in all scenarios.

To some extent, the evolution of the Yuan PLUS also represents a new stage for Chinese brand pure electric vehicles: in the past, the focus was on making electric vehicles cheaper and easier to drive; now, the goal is to make charging, intelligence, space, and lifestyle more natural. Of course, the Yuan PLUS's journey has never been limited to the domestic market, but to the global market, with a target of hundreds of thousands of units sold annually. Therefore, to a certain extent, the Yuan PLUS has a strong defining power in this mainstream market. The third-generation Yuan PLUS aims to define itself by achieving a balance between range, charging, comfort, safety, quality, and intelligence at a reasonable price, becoming a "hexagonal warrior" in this mainstream vehicle market.

#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.

"If this tweet receives a like, Codex resets the limit."

I've lost count of how many times the usage limit has been reset this year. A couple of days ago, Ultraman posted on X, prompting Tibo, the head of Codex, to reset the usage limit once again.

A meme has been created online, showing that whenever someone tries to walk towards Anthropic or Gemini, Ultraman stands behind them and silently presses the Codex limit reset button, causing the person to turn around and be pulled back to OpenAI.

OpenAI has also gained a large number of new users in the past six months thanks to the popular Codex platform. Foreign media reports indicate that OpenAI's first-quarter revenue reached $5.7 billion, $1 billion higher than Anthropic, with Codex being a major contributing factor.

We previously introduced a beginner's guide to Codex, which included detailed steps from downloading and installing it from the ChatGPT website to connecting the ChatGPT App to your phone for remote control.

Many readers left comments praising Codex's usability; however, they also reported several issues, such as the requirement to link a phone number after downloading Codex. Our testing confirmed that logging out and then back in does indeed require linking a phone number.

At this point, it's recommended to log in to your browser first, i.e., actively open the website https://auth.openai.com/log-in and log in beforehand. Then, when you log in through Codex, the login link will only ask for authorization; there will be no further prompts to bind your phone number.

Different accounts may encounter different situations, probably because OpenAI has invested too much computing power in Codex and doesn't want users to take advantage of it too easily.

Early this morning, Codex released a whole host of new features. Now, simply pressing the Command-Command key on your computer will attach an application window to Codex's conversation thread. Codex will automatically capture a screenshot and text of the window, including content not visible on the screen, as context for the conversation.

Previously, you had to take screenshots manually, but now Codex can not only process screenshots, but also directly read the information of an entire application window.

In addition, the option to operate Codex on your computer within the ChatGPT App, which was updated last time, has also been upgraded. Previously, the option was to keep Codex always on, but now even if the computer screen is locked, ChatGPT can still remotely operate Codex.

The /goal command has now been officially released, moving from the lab version. Previously, when we shared information about multi-agent collaboration, some readers mentioned that the /goal feature is similar to multi-agent collaboration; both manage a task as a project, have a complete goal lifecycle, and iterate through different mechanisms.

The /goal feature first appeared in the Codex CLI at the end of April, and it does indeed help to better handle the increasing number of long tasks.

Unfortunately, whether it's pressing command or continuing remote control after locking the screen, these are updates for the macOS platform. For Windows users, they can only wait for OpenAI to make progress.

One netizen commented, "Mac users always get to enjoy great things, while Windows users can only watch enviously, haha." It has to be said that the value of the Mac mini as an AI PC is still increasing.

Saves a lot of trouble with application snapshots

This feature is called Appshots, and it's easy to enable. Just update Codex, go to the application settings, find "App Snapshots," and you'll find a video tutorial. You can also customize the keyboard shortcut.

However, it should be noted that pressing the command key means pressing the two command keys on the left and right sides of the space bar on the keyboard, not clicking them twice.

Pressing both command keys simultaneously on any screen will cause Codex to automatically capture a screenshot of the page and quickly open Codex, placing the screenshot in the input box. We can then ask questions based on this window snapshot.

However, thanks to Codex's capabilities, this window snapshot is not just an OCR text extraction of an image. Codex can further utilize features such as Computer Use and Chrome automation based on this window.

For example, we press the command key at the beginning of an article in Lark Docs and tell Codex to look at what the window is saying. Codex will then use Google Chrome's tools to automatically browse the webpage for more context.

This is the biggest difference between it and a regular screenshot. In addition to putting the screenshot content into the context, Codex will also automatically send the window information, the application from which it came, and other status information to Codex in sync.

For example, when reading a public account in WeChat, we can press two command buttons to open Appshots. However, there's a small bug: when Codex uses Computer Use to control the WeChat window and swipes up and down on the public account to exit the image preview, it directly logs out of WeChat.

When the official promotional video introduces Appshots, it doesn't simply use it as a screenshot, but combines it with Computer Use and Google Chrome.

It's like directly asking them to modify the contents of our memo.

There's no need to copy any images anymore; just use command + command and tell the image prompt to edit it.

It's this extra step in the application that reduces the burden of using AI and makes the Codex experience even smoother.

A comprehensive user guide for /goal

By typing a forward slash in the dialog box, we can see a shortcut option for "Goals," which allows us to "set goals that Codex will continuously strive to achieve."

The value of a goal lies in its existence as an independent task definition, rather than a regular dialogue prompt. Codex repeatedly uses the goal to determine "what else needs to be done" and "whether it has been completed," automatically advancing round after round until the task is completed, paused, or the token limit is reached.

These two judgments are also the core mechanisms of the objective: "Continue" and "Complete Audit". "Continue" involves automatically injecting prompts after each round, allowing the model to decide the next step. "Complete Audit" requires the model to check against the objective item by item.

The most common pitfall in Goal models is carelessly adding a single line of code. The key principle for writing a good Goal is that Codex must be able to determine whether the task has been completed.

The official documentation also mentions that good goals should include specific results, measurable metrics, or test standards. They provide examples such as migrating a project from one programming language to another.

Migrate this project from JavaScript to TypeScript.

Requirements: Compiles successfully in strict mode; explicit any type is not allowed.

There's an even more direct requirement: "Reduce the interactive time on the homepage to less than 1 second."

These examples all have specific, verifiable standards, and are not just empty words like "optimize" or "improve".

If you haven't come up with specific standards, Codex suggests running `/plan` first. Let Codex discuss it with us to clarify the acceptance criteria before switching back to `/goal` in normal mode.

Here are some useful tips: you can add a sentence at the end of the goal text, such as "Use a token budget of 80,000 tokens for this goal," to set the token budget.

And don't send /goal at the beginning of a session. Instead, give other requirements for the project first, so that it has a certain outline, and then give it the goal.

Even when your screen is locked, Codex can still operate your computer.

In addition to these major updates, Codex Thursday also brought many features that enhanced the user experience.

Locked Computer Use is the most noteworthy feature. Simply put, it allows Codex to continue operating desktop applications and completing tasks in the background even after the Mac screen is locked.

Netizens' comments on this feature are mostly focused on how groundbreaking it is, how futuristic it is, and how scary it is.

If Codex can run Mac applications without active user sessions, this could be a first step toward persistent agent infrastructure.

To use the feature to continue operating after locking the screen, you must manually enable it and enter a password. This is done in Settings, by finding PC control and then enabling lock screen operation.

Normal Computer Use requires the screen to be unlocked for Codex to "see" and interact with the interface. This feature breaks that limitation. We can close or lock our Mac and then remotely initiate a Codex task from our phone, iPad, or another device. It will automatically unlock temporarily, complete the task, and then lock again.

Codex installed an Apple Authorization Plug-in to integrate with the macOS unlocking process. When there is an active Computer Use task, the plugin allows Codex to temporarily unlock the screen; outside of the task window, unlocking permissions are denied.

OpenAI has also implemented several layers of restrictions on this feature to prevent it from becoming a backdoor for other dangerous operations:

• The unlock window is extremely short and is only valid during the current Computer Use operation.
• Covering all displays, the screen content is invisible to physical bystanders during the temporary unlocking period.
• Local input detected, immediate relock – if someone touches the keyboard or mouse, the system automatically pauses and requires manual unlocking.
• This path is only open to Codex; other applications or local processes cannot use it.

Another advanced annotation feature is that when using Codex Vibe Coding to open a webpage through the built-in browser of Codex, it also provides annotation tools that allow direct modification of the webpage content.

In addition to the Codex series of updates, ChatGPT also launched a new feature today: ChatGPT can now create and edit presentations directly in PowerPoint, and can also use GPT Image 2 to generate images for use in PPT.

As Codex becomes more and more user-friendly, the rate at which wallets burn through cash is also accelerating.

Our Pro account's weekly usage limit resets on the 27th, but today (the 22nd) we only have 10% left. We can only silently pray that it resets again.

If this post receives a like, your Codex limit may be reset. .

#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 price range of 1,000 to 1,500 yuan marks a watershed moment in the true wireless earphone market.

At this price point, TWS devices are becoming increasingly difficult for users to accept with significant shortcomings—strong noise cancellation, good sound quality, clear calls, and robust connectivity and interaction are all essential. What's even more challenging is that this market is often dominated by two types of powerful players: traditional acoustic brands and major mobile phone manufacturers—the former possessing acoustic expertise, and the latter having system entry points.

But Anker is different.

It lacks a mobile operating system interface and the inherent prestige of traditional acoustic brands. In the past, when people talked about Anker, they thought more about chargers, power banks, overseas channels, and the ability to create blockbuster products.

As Anker's new foray into the audio-visual field, Anker noise-canceling headphones attempt to answer a question: Without a smartphone interface and without the halo of traditional acoustics, how can Anker sell a flagship pair of headphones priced at around 1,000 yuan?

Flagship headphones must first and foremost be good headphones.

Anker noise-canceling headphones are positioned as flagship products, so first and foremost, they must be good headphones. Specifically, this means looking at three aspects: sound quality, noise cancellation strength, and call clarity.

In terms of sound, the Anker noise-canceling headphones do not follow the "high-resolution, highly transparent" tuning approach of many current TWS headphones. Its default sound is thicker and fuller, with an overall slightly warm tone, reminiscent of the "pop" presets commonly found in early music players.

The advantage of this approach is that the sound is less likely to be dry. The low frequencies have ample depth and quantity, and the drum beats have a certain elasticity and liveliness, while the overall atmosphere is also slightly thickened. It doesn't make the high frequencies too bright, so there is no obvious harshness even after listening for a long time. The details of vocal articulation, exhalation, and breathing are also quite smooth, making for a comfortable listening experience.

However, this tuning also comes at a price.

The low frequencies are quite prominent, and when Dolby audio is enabled, the ambiance is amplified further, sometimes to the point of being overpowering. High-frequency instruments are not harsh, but their extension and sense of airiness are relatively restrained.

For users who prefer a crisp, clear, and well-defined sound, the default sound of Anker noise-canceling headphones might be a bit heavy. If Anker could tone down the bass a little and increase the detail and transparency in the default tuning, the overall balance would be better.

Additionally, the inability to adjust the sound or control Dolby audio when LDAC is enabled is somewhat inconvenient. While Dolby audio may not be essential when listening to music using high-resolution encoding, allowing EQ or sound mode adjustments would at least allow users to fine-tune the output according to their listening preferences, which would be more user-friendly.

In terms of noise cancellation, Anker's noise-canceling headphones have reached the level of top-tier TWS noise-canceling headphones.

In public spaces where train stations, shopping malls, crowds, and ambient noise are mixed together, it can significantly suppress continuous background noise.

In our current experience, its noise cancellation is approaching the perceived level of top-tier products like AirPods Pro and Huawei FreeBuds Pro. Whether it can consistently match or even surpass them in all scenarios requires more systematic testing, but at least it has entered the range where flagship headphones can be seriously compared.

What really surprised me was the phone call.

I was calling a friend from the train station. There were announcements, crowds, the sound of suitcase wheels, and constant ambient noise. It was raining outside the room where he was, so the background wasn't quiet either. But throughout the call, he said my voice was very clear, even easier to hear than the person on the other end of the line.

The call performance of Anker noise-canceling headphones is one of the most important aspects of these headphones to consider.

While call quality and noise cancellation aren't as readily apparent as these factors, they often determine a user's true evaluation of a pair of headphones. When buying headphones, users rarely think about "call noise cancellation" first. But if someone can't hear you clearly, the first reaction will be: these headphones are no good.

Call noise cancellation is indeed difficult to achieve – wind noise, human voices, background noise, and sudden noises all mix together, and headphones must filter out ambient sounds without fragmenting human voices. In the past, those who truly excelled at this were mostly manufacturers like Huawei with long-term accumulation of communication technology.

Anker's most valuable progress this time occurred precisely in this inconspicuous but authentic place.

The flagship headphones of 2026 cannot be just headphones.

If the competition among flagship headphones in the past two years mainly revolved around noise cancellation, sound quality, fit, and spatial audio, then things began to change in 2026.

A flagship pair of headphones may begin to take on a second identity. Specifically for Anker noise-canceling headphones, the charging case is a small AI hardware device.

It's not a device that relies entirely on local computing power for all intelligent processing; AI voice transcription requires uploading to a cloud model for processing, and the payment method primarily involves consuming points. Therefore, more accurately, it's a flagship pair of headphones, but with an added AI capability entry point.

Headphones are portable and naturally suited to environments with sound. In the past, the sound received by headphones was mainly used for playback, noise reduction, and calls; now they are beginning to be used for transcription, translation, and organization.

The Anker noise-canceling headphones support face-to-face translation and offer AI recording functionality. The screen on the charging case allows for direct volume adjustment, noise cancellation mode switching, preset customization, and access to AI features.

On the software side, the earphones and the charging case are two separate Bluetooth products that need to be connected separately. Anker doesn't treat this charging case merely as a battery compartment with a screen, but rather as a near-lightweight terminal.

Of course, the concept of AI headphones is not yet truly mature. Whether users will frequently use face-to-face translation and AI recording, whether using the headphone case for interaction will become a long-term habit, and whether AI can generate stable value in more everyday scenarios all require further observation.

But when you spend over 1,000 yuan on a pair of headphones, in addition to noise cancellation, sound quality, and call quality, you also get a small hardware entry point that can record, translate, and handle some AI interaction—which seems like a pretty good added value.

It is indeed a very clever design.

Anker wants to reshape Anker

It can be said that Anker noise-canceling headphones are an exceptional product, and also a complex one.

It needs a screen, dual Bluetooth connectivity, and must balance noise reduction, sound quality, and call quality, while also integrating AI recording and face-to-face translation functions. Such a product is no longer simply a combination of several mature modules; it requires companies to achieve deeper integration across chips, algorithms, system interaction, and hardware design.

Understanding this product solely from the perspective of headphones might not provide a complete picture. However, looking at it from the perspective of Anker as a company makes things clearer—the core of this product is Anker's self-developed Thus technology. chip.

Anker then said Called a neural network in-memory computing AI audio chip, it aims to solve a small but complex problem:

How can complex AI models be run most efficiently in a tiny, low-power device like headphones that requires real-time feedback?

In the past, most computing devices followed the von Neumann architecture: data resided in memory, and computation occurred in the processor. When AI models were running, the frequent data transfers led to increased energy consumption and latency.

For headphones, the problem is even more severe. They lack large batteries, ample cooling, and sufficient space for hardware, so any latency, power consumption, and heat generation are amplified.

The idea behind in-memory computing is to make computation happen as much as possible where the data resides. Reducing data movement lowers power consumption, improves efficiency, and also allows for the inclusion of more complex neural network models in smaller devices.

Such an architecture is uncommon in the consumer electronics industry. Anker's inclusion of it in a pair of headphones is sending a signal:

Anker is not content with being an integrator of mature solutions; it wants to control the underlying core technologies.

In the past, our understanding of Anker was usually limited to product definition, industrial design, supply chain, overseas channels, and the ability to create blockbuster products. These capabilities fueled Anker's past growth. But today, simply making good products is no longer enough.

For a hardware company of Anker's size, the question it needs to consider is no longer just how to create the next blockbuster product, but how to evolve from "making good products" to "mastering core technologies".

This change is also reflected in the business and brand structure.

Besides its audio-visual business, Anker's main businesses also include energy storage and security. These businesses have higher average order values ​​and value-added potential in overseas markets, and rely more on brand trust, technological stability, and long-term service capabilities. These are not overseas businesses that can be established simply by selling in large volumes at low prices.

At the same time, Anker is also consolidating its brands. Anker Soundcore, which was originally more of an independent sub-brand, is being rebranded as "Anker Audio-Visual". This means that Anker hopes to consolidate its capabilities, which were scattered across different product categories, under the main brand.

Anker noise-canceling headphones are a litmus test.

First and foremost, it's a flagship pair of headphones, so noise cancellation, sound quality, and call quality have to be top-notch. But it's more than just headphones, because it packs AI hardware, software interaction, and a self-developed chip all into this little box that costs just over 1,000 yuan—this level of sophistication and complexity is a challenge, but it's definitely also an opportunity.

Anker doesn't want to be content with the status quo and become a "brand that only sells power banks".

Anker is reinventing Anker, and the Anker noise-canceling headphones—and the Thus embedded within them—are all part of that reinvention. The chip was just a test.

#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.