Carl Pei just confirmed it on X: Nothing’s first AI-native product is coming in December. He called it “a tool, but one built with the personality and care that have come to define a Nothing product.” And the key detail is that it lacks a screen. Pei was pretty clear about that part: this thing “couldn’t have been an app.”

That’s pretty much all we know right now. No form factor, no features, no price, no exact date. Just a vague teaser image with the Nothing logo and a black rounded corner, and Pei’s insistence that this is something different.

This isn’t a total surprise, though. A little over a year ago, Pei talked about Nothing’s plans for AI-native devices and their own AI operating system. What’s interesting is the deliberate decision to skip a display entirely. That already puts it in a different category from phones or most of the AI gadgets we’ve seen so far.

Pei also used the post to take a jab at the rest of the industry. He thinks everyone’s still chasing some kind of “iPhone moment” with pendants, pins, glasses, and companions, and that hunting for “the iPhone of AI today is a dead end.”

There have been earlier rumors about Nothing looking into AI smart glasses (including a possible camera-free version) and shifting more toward AI accessories in general. Nothing hasn’t confirmed whether those reports are connected to this December device, though.

Screen-free AI hardware isn’t brand new territory. We’ve already seen attempts like the Humane AI Pin and Friend, and OpenAI is reportedly working with Jony Ive on some kind of AI smart speaker. Nothing’s version could just be another data point in that still-forming category.

For now, the product is mostly defined by what it doesn’t have, a screen, and by Pei’s claim that it’s an actual tool, not just software pretending to be hardware.

(Source: Nothing)

Read more

• Nothing Ear (3) Pro teasers begin

For the latest stories, visit our News section.

New: Gizmochina is now on WhatsApp. Follow our WhatsApp Channel for the day’s top stories.

You can also join our Telegram community or get Gizmochina Daily, our morning newsletter with our take on the day’s news.

The post Nothing teases a screenless AI-native device ahead of December launch appeared first on Gizmochina.

AGIBOT has introduced Worken, a new framework designed to measure how much useful work robots actually complete in real-world environments. The framework aims to provide a different way of evaluating progress in the robotics industry. Instead of focusing on how many robots are manufactured or deployed, Worken looks at their actual productivity, autonomous working time, and the quality of their output.

What Is AGIBOT’s Worken Framework?

The name Worken combines the words Work and Token. The proposed framework focuses on measuring Effective Intelligent Work Hours, giving businesses a way to understand how effectively Physical AI systems perform assigned tasks.

AGIBOT believes that counting deployed robots alone does not provide a complete picture of their usefulness. For example, two factories may operate the same number of robots but achieve very different production results. Robots that work independently for longer periods and consistently complete tasks with minimal human assistance deliver greater productivity.

The same idea can also apply to logistics and service robots. Simply moving around a warehouse or interacting with customers is not enough. These machines must successfully complete their assigned tasks.

How Does Worken Calculate Robot Productivity?

AGIBOT’s proposed formula combines three factors: the number of active robots, their autonomous working hours and their output efficiency ratio.

The first factor counts robots that are actively performing useful tasks rather than machines that have only been manufactured, delivered or installed. The second measures how long robots work independently, excluding waiting periods and continuous human control. The third evaluates completed work against an agreed productivity benchmark.

For example, a manufacturing facility with 10 active robots, six autonomous working hours per robot and an output efficiency ratio of 0.8 would generate 48 Worken.

This approach aims to distinguish between machines that remain powered on and those that consistently deliver acceptable results.

Potential Applications in Manufacturing and Other Industries

Manufacturing could be one of the first areas where Worken becomes useful because production tasks often have clearly defined targets. Businesses could use the framework to monitor robotic productivity, compare similar operations and identify where machines lose productive working time.

It could also help companies evaluate software updates by checking whether robots complete more tasks independently or require less human intervention.

AGIBOT’s existing industrial applications, including material-handling operations at Longcheer, provide potential real-world environments for exploring the framework.

The concept could eventually extend to service robots, where productivity might be measured through completed guidance tasks, customer assistance and successful visitor interactions.

Worken Is Still an Evolving Framework

AGIBOT describes Worken as a developing measurement framework rather than an established universal standard. Its practical usefulness will depend on reliable operational data, clearly defined tasks and suitable performance benchmarks.

It is also not intended to replace other important measures such as safety, reliability, operating costs and output quality. Furthermore, Worken does not represent a universal monetary value for robotic labour.

If developed successfully, the framework could help businesses make better decisions about robotic investments and understand how software and AI improvements increase productivity over time.

Ultimately, AGIBOT’s approach highlights a broader shift in robotics: measuring progress not just by how many robots enter the workplace, but by how much useful work they actually deliver.

Read More:

• AGIBOT Wins 46 Medals at World Humanoid Robot Games, Takes 18 Gold
• Humanoid Robots Are Scaling Fast, And AGIBOT from China Is Leading the Charge
• Chinese Robot Company AGIBOT Took 39% of the Global Humanoid Market in 2025

The post Are Robots Really Productive? AGIBOT’s New Worken Framework Has an Answer appeared first on Gizmochina.

Soulog has launched the Sense, a compact AI-powered wearSoulog has launched Sense, a compact AI voice recorder that works with the Soulog app to capture conversations, organize personal context, and help users create useful outputs.

The device is now available through Soulog’s official store and on Amazon.

A Different Kind of Recorder

Sense captures audio, while the Soulog app organizes recordings and related materials into context users can revisit and build on.

Three-step loop

That idea is embodied in a three-step loop Soulog calls Capture → Connect → Create.

Capture is built around low-friction input. A physical flip switch starts recording instantly, users can add photos, notes, and files in the Soulog app, and a one-press highlight marks key moments mid-conversation. There’s also a Rewind feature, which saves up to 30 minutes of earlier audio. It requires Sense to be on, advance activation, and configured active hours; it works only within those hours.

Users can review transcripts, Recap, and to-dos, and explore context through Map & Timeline. A Space brings conversations and supporting files together. Transcription is free and unlimited; some AI features require a paid subscription.

In Soulog, users can develop editable Artifacts, including proposals, briefs, and reports. Compatible AI tools can access Soulog context through MCP.

Hardware That Stays Out of the Way

Sense measures approximately 4.0 × 2.5 × 0.12 inches, weighs 1.6 ounces, and includes 8 GB of storage. It uses two MEMS microphones and a VPU, with a stated pickup range of up to 16 feet.

Sense provides up to 50 hours of continuous recording from a full charge. Runtime depends on usage, settings, and operating conditions. Charging is handled over standard USB-C — no proprietary dock required. USB-C charging for 10 minutes supports around eight hours of typical workday use, with results varying by usage and conditions. That last point is a quiet but meaningful differentiator against competing AI recorders that lock users into bespoke charging accessories.

Privacy Focus

According to Soulog, audio is encrypted on the device, during transfer, and in cloud storage. Cloud sync requires user activation. Soulog states that user content is neither sold nor used for model training.

Soulog is targeting three broad groups: urban professionals such as lawyers, consultants, product managers, and journalists who drown in fragmented information; creative professionals who need to connect scattered ideas; and students and researchers turning lectures into papers. If you’re one of them, visit Soulog’s official website or Amazon to upgrade the way you live and work.

The post Soulog Sense Launches as an AI Voice Recorder That Actually Remembers Your Conversations appeared first on Gizmochina.

HOHOTIME is launching its All-in-One AI Smart Training Station on Kickstarter, offering a new approach to strength training and home fitness. Designed with four independently adjustable resistance arms, AI-powered coaching and a library of 860 exercises, the machine aims to combine multiple workout options in a single setup. 

This new design can replace more than 60 traditional fitness machines, making it an option for users who want a versatile training system without filling their workout space with separate equipment.

Four-Arm Design for Full-Body Training

The main feature of the HOHOTIME Smart Training Station is its four-arm resistance system. Each arm can be adjusted independently, with a movement adjustment range of 150 degrees. This design allows users to set up different movement paths for exercises such as presses, pulls, squats and rotational movements.

The system supports 95% of gym exercises and can accommodate different body types, heights and training positions. Each arm offers up to 400 lb of resistance, while the high-strength nylon cables are rated for loads of up to 800 lb. 

The machine has a metal-bodied structure, aluminum-alloy pulley housing and a metallic finish. Built-in wheels are included to make moving the equipment easier. It is available in Black & Golden and White Radiance colour options.

AI Coaching and 860 Guided Exercises

HOHOTIME combines its resistance system with AI-powered workout assistance. Its motion-tracking system uses 42-point skeletal tracking to monitor body movements and provide feedback during exercise sessions. The AI Coach is designed to guide users through workouts, help improve exercise technique and track performance.

The system can monitor workout data such as power, repetitions and movement quality. It also records training sessions to help users follow their progress over time.

The built-in exercise library contains 860 exercises, with no additional exercise-library fee advertised. Training options shown in the campaign include strength workouts, HIIT, boxing, stretching, meditation, recovery and golf-related exercises.

Users can interact with the machine through its touchscreen, physical buttons and rotary controls, or voice control. 

Five Accessories and HYROX Training Support

The training station supports five specialised accessories: Grip-Free Gloves, a T-Bar, a Millimeter-Adjustable Support Belt, a Scapular Stabilization Bench and Ski Handles. These attachments are designed to expand the range of exercises and provide additional training positions.

HOHOTIME also comes with selected HYROX training capabilities. The machine supports six core HYROX training movements, with AI tracking and performance monitoring to help users work on endurance, pushing power, lower-body strength and full-body conditioning. Please note that it is intended for home preparation rather than a confirmed replacement for every official HYROX race station.

The system can also connect with compatible body composition analyzers and heart-rate wristbands, bringing additional fitness information into the training experience.

Kickstarter Price, Warranty and Availability

HOHOTIME is scheduled to launch its Kickstarter campaign on October 22, 2026, with a Super Early Bird starting price of $2,799.

The purchase benefits include a two-year warranty and a 30-day free trial. Buyers should check the final Kickstarter reward tiers for package contents, shipping charges, taxes, return conditions and warranty exclusions before pledging.

With its four-arm resistance system, AI coaching and broad exercise library, HOHOTIME aims to offer a flexible alternative to traditional home gym equipment. 

The post HOHOTIME AI Smart Training Station Brings Four-Arm Resistance Training to Home Gyms appeared first on Gizmochina.

The 46th Hong Kong Electronics Fair (Autumn Edition) will take place from October 13 to 16, 2026, at the Hong Kong Convention and Exhibition Centre (HKCEC). Organised by the Hong Kong Trade Development Council (HKTDC), the event will showcase developments in artificial intelligence (AI), robotics, smart wellness, entertainment technology and consumer electronics.

The fair will run alongside the 29th electronicAsia, organised by HKTDC and MMI Asia Pte Ltd. Together, the two events are expected to bring more than 3,200 exhibitors from 15 countries and regions to Hong Kong. An online exhibition will also run from October 6 to 23 through the Click2Match platform.

AI, Robotics and Smart Wellness Take Centre Stage

The Autumn Electronics Fair will focus on three major themes: AI & Robotics, Smart Wellness and NEXTEntertainment. A new Future Industries Zone will also highlight technologies aimed at emerging industries.

The RoboPark area will feature more than 120 robots, covering applications in commercial services, industrial operations, logistics, entertainment, healthcare and rehabilitation. Visitors can also explore humanoid robots and AI-powered wearable robotic systems. Activities will include robot football, robot dance battles, a RoboDog Garden and a station where robots serve coffee and popcorn.

Companies expected to showcase their technology include AgiBot, UBTECH, Booster Robotics, VIGX Technology and Shenzhen Honor Smart Technology Development. UBTECH plans to demonstrate its Walker C1 humanoid robot, while VIGX will present a pocket-sized AI exoskeleton robot. Other participating robotics companies include AI² Robotics, Digit Robotics, EngineAI, DexForce and Deep Robotics.

The Smart Wellness theme will cover digital health, wearable devices, personal care, gerontechnology and healthy living. NEXTEntertainment will focus on extended reality (XR), virtual reality (VR), augmented reality (AR), smart gaming and AI-powered entertainment.

New Future Industries Zone to Feature More Than 100 Exhibitors

Supported by the Innovation, Technology and Industry Bureau of the Hong Kong SAR Government, the Future Industries Zone is being developed jointly by HKTDC and the Hong Kong Electronic Industries Association (HKEIA). It is expected to feature more than 100 exhibitors, including Lenovo, AV Concept and GP Energy Tech.

The zone will cover five areas: Future Computing, Future Manufacturing, Future Marine & Quantum Technologies, Future Microelectronics and Future Wellness.

Planned activities include industry forums, startup pitching sessions, talent development discussions and business-matching opportunities. The programme will also introduce the Top 100 Future Industries Pioneer List, intended to identify innovative products and solutions with commercial potential.

Speakers expected to participate include Jeffrey Lam, Vincent Ma, Clara Chan and Allen Yeung, representing organisations involved in industrial development, innovation and technology, investment and the information technology industry.

Technology Exhibits Extend Beyond Robotics

Other areas of the fair will showcase technologies for businesses and consumers. Tech Hall will feature AI data solutions, the Internet of Things (IoT) and transport monitoring technologies.

The Adventure Hub will include an Immersive Experience Zone featuring brainwave gaming sensors, allowing users to control games through concentration rather than conventional physical controls. Manna Pacific (HK) Limited will demonstrate an AI-powered indoor fitness screen that uses real-time motion tracking to provide visual guidance and support workout posture and coordination.

The wider exhibition will feature more than 20 themed zones. Its Hall of Fame will present more than 560 established brands, while first-time exhibitors will include HiFuture, Aoni and ZENS. RISE Avenue will focus on emerging brands, gaming equipment and home entertainment products.

The Startup Zone is expected to host more than 120 emerging companies and startups. Participating groups will include IOTHK and JimHang Academy, with activities designed to connect technology developers with potential business partners and investors.

Industry Forums to Discuss AI and Business Expansion

The Symposium on Innovation and Technology, organised by HKTDC and the Hong Kong Electronics and Technologies Association, will focus on “AI-Driven Security & The Next-Gen Infrastructure.” Discussions will cover AI, intelligent systems, infrastructure security, operational efficiency and business competitiveness.

Expected speakers include Alexander Gerfer, CTO of the Würth Elektronik Group; Dr Charles Cheung, Adjunct Associate Professor in the Department of Mathematics at Hong Kong Baptist University; and Silvia Lam Ihensekhien, identified in the event announcement as Chief Security Advisor at Microsoft.

The Hong Kong Electronics Forum will examine how AI, sustainability and technological innovation are shaping the electronics industry. Startup-focused activities will include enterprise-investor matching, mentorship and the Hong Kong Value Creation for Technology: Pitching Competition.

A separate GoGlobal Connect Zone will help mainland Chinese companies explore international expansion through Hong Kong. Sessions will address overseas market entry, financing, risk management, smart manufacturing, sustainability, certification and international brand development. Planned discussions will involve representatives from HKECIC, TechNode, Hang Seng Bank and OCBC Bank (Hong Kong), among others.

electronicAsia to Focus on Electronic Components

Running alongside the Autumn Electronics Fair, electronicAsia will focus on semiconductors, sensors, electronic components and manufacturing technologies. Exhibitors named in the announcement include Easttop Display and Microtech Technology, although specific product launches have not been detailed.

The two exhibitions will use HKTDC’s EXHIBITION+ model, combining the physical events with online business connections through Click2Match until October 23. A complimentary shuttle service between HKCEC and AsiaWorld-Expo will also support visitors attending the wider October exhibition programme.

Hong Kong Electronics Fair 2026: What to Expect

The Autumn Electronics Fair is designed to connect technology companies with buyers, investors, talent and international markets. Its programme combines consumer electronics, robotics demonstrations, emerging technology showcases and business discussions, while the Future Industries and GoGlobal Connect zones place additional emphasis on industrial innovation and overseas expansion.

With AI, robotics, smart wellness and future computing among its central themes, the event will offer visitors an overview of technologies and business opportunities across the electronics sector.

The post Hong Kong Electronics Fair (Autumn Edition) 2026 to Showcase AI, Robotics, and Future Technologies appeared first on Gizmochina.

OpenAI CEO Sam Altman is getting fresh pushback after laying out a pretty blunt take on AI’s trade-offs. In short, he thinks the world should be willing to live with certain harms if it means unlocking the bigger benefits of the technology, and giving people more control over it.

He made the comments on the first episode of Politico’s “Decoded” podcast while talking about where OpenAI and rival Anthropic differ on safety and regulation. “We believe that the world should accept some bad things happening for the benefits of this technology and people having the agency,” Altman said.

He’s in favor of a lighter regulatory approach, arguing that as society gets used to AI, some ugly stuff such as hacks, scams, and other misuse is pretty much inevitable.

Altman is clear that he still thinks the good will massively outweigh the bad. “I wouldn’t take a trade of saying, ‘We’ll make sure there’s no major hacks, there’s no misuse of this technology, there’s zero scams, there’s zero all the other bad things that will happen,’” he explained, “because I think people will do tremendously—orders of magnitude more—good stuff than bad stuff.”

He drew a line between these more limited risks and the truly catastrophic ones, like a serious loss of human control over AI systems. Those, he said, deserve a lot more caution.

The remarks land in the middle of ongoing arguments about AI’s real-world costs: job losses, the environmental hit from data centers, copyright fights, cybercrime, and the huge demand on electricity and water. Critics didn’t hold back.

New York University professor emeritus Gary Marcus accused Altman of “saying the quiet part out loud: suck it up so you can make us rich and powerful.” Online, a lot of people seemed frustrated with the idea that the benefits will just automatically dominate the drawbacks.

The whole exchange highlights a deeper split in the industry: one side wants to minimize near-term harms with stricter controls, while the other is willing to accept some friction as the cost of moving fast and keeping users in the driver’s seat. Whether Altman’s math holds up is still very much up for debate.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

The post Sam Altman believes “the world should accept some bad things happening” for the benefits of AI appeared first on Gizmochina.

AI writing tools have become common, but many are still focused mainly on generating and editing text. Revise takes a different approach by combining AI assistance with a full-featured document editor. The new web-based platform is designed to work with real documents, including contracts, reports, manuscripts, and other files that require more than basic text formatting.

A Document Editor Built for Real Files

Tools such as Type.ai and Lex Editor are mainly designed for writers creating articles, blog posts, and other online content. They offer features such as headings, bold and italic text, alignment, and images, but can be limited when working with traditional Word documents.

Features such as page layouts, tables, indentation, headers, and other Word-specific elements can be difficult to manage in these editors. Some tools may also not preserve every DOCX feature when a document is exported. Revise is designed to address these limitations by focusing on document editing rather than just text generation.

AI Editing With DOCX Support

Revise allows users to import existing DOCX files and edit them with AI without having to start from a blank page. The platform combines text editing, formatting, and document layout controls in one interface.

It also aims to preserve important Word features, including tracked changes, comments, section layouts, and watermarks. This makes the platform more suitable for users who regularly work with complex documents and need to maintain their existing structure while making AI-assisted changes.

Built-In AI Agent for Document Work

A key feature of Revise is its built-in AI agent, which is designed specifically for document editing. The agent can view the document in the same context as the user and make suggested changes directly to its content and structure.

The AI can work with text, lists, tables, formatting, page layouts, and document structure. Its changes are presented as tracked edits, allowing users to review them before accepting or rejecting them. This gives users more control instead of allowing AI changes to be applied automatically.

AI-Powered Review and Comments

Revise also includes tools for reviewing documents. Users can highlight a section and ask the AI to comment on it through a discussion thread. The agent can also review an entire document and identify possible issues.

Rather than automatically changing everything it finds, Revise can flag problems for the user to review manually. This approach can be useful for documents where accuracy and editorial control are important.

Templates and More Control

For users who do not want to create a document from scratch, Revise includes a template directory with ready-made documents. This provides another starting point while keeping the editing and AI features within the same platform.

Overall, Revise aims to bridge the gap between AI writing tools and traditional document editors. By combining AI-assisted editing, DOCX compatibility, document review, formatting, and layout controls, it is positioned as a tool for more complex document workflows where simple text editing is not enough.

The post Forget Basic Writing Tools: Revise Can Edit Your Word Documents With AI appeared first on Gizmochina.

TECNO is exploring how mobile AI can be tailored to fit specific professional workflows through a new collaboration with Digital Trends Media Group. The company has demonstrated four customized AI workflows built around the needs of technology journalists and digital content creators, using its experimental EllaClaw mobile AI agent. Announced on September 23 in Hong Kong, the showcase highlights TECNO’s efforts to make AI more practical for real-world tasks instead of relying only on a fixed set of features. 

TECNO EllaClaw Focuses on Customizable Mobile AI

EllaClaw is described by TECNO as an exploratory mobile Agentic AI system that can adapt to different professions, industries, and user requirements. Instead of working as an autopilot that takes over tasks, TECNO positions it as a digital co-pilot that helps users complete repetitive work while keeping important decisions under human control.

The Digital Trends collaboration was created to explore how this approach could work in professional content creation. The resulting workflows cover several stages of the process, from researching a developing story to preparing content for publication and social media.

Four Workflows for Content Creators

The first workflow, News Background Research, allows users to share a developing story with EllaClaw. The agent can identify the topic, search selected trusted sources, and organize the findings into a structured brief. The research can include historical context, key developments, stakeholder perspectives, and source references. The resulting brief can be saved to the phone’s Notes app, while an audio version allows users to listen to the research.

The second workflow, Research-to-Script, takes the gathered information and turns it into structured storyboards. It can provide multilingual script suggestions, supporting creative assets and content-planning assistance. This is designed to connect the stages of idea generation, research, storyboarding and production without requiring users to move between several different apps.

Personalized Writing and Social Media Checks

EllaClaw also includes a Personal Writing Style workflow. Users can provide writing samples, which the AI analyzes to create a reusable style profile. The profile can then be applied when generating content for different platforms. TECNO says the profile can continue to evolve through new writing samples, feedback and ongoing interaction.

The fourth workflow, Social Media Risk-Word Check, reviews content before publication. It checks copy against platform-specific publishing considerations and helps identify potentially risky wording. The goal is to reduce publishing risks while keeping the original message effective.

Human Control Remains Central

TECNO says user control, transparency and privacy are important parts of the EllaClaw experience. Users remain responsible for reviewing, refining and approving AI-generated content before publication. This keeps creative decisions and editorial judgment with the person using the system.

Although the demonstration focused on technology journalists and content creators, TECNO says the same concept could be adapted for other professions, industries and personal workflows. The showcase therefore presents EllaClaw as an example of how mobile Agentic AI could become more personalized and practical, while still keeping human creativity and decision-making at the center.

Read More:

• TECNO POVA 8 Pro 5G Tonino Lamborghini Limited Edition Blends Italian Design With Smart Technology
• TECNO Showcases New MEGABOOK Laptops and Smartphones at IFA ShowStoppers 2026
• TECNO CAMON Slim 5G Quick Review: This Is the CAMON Slim in Its Final Form!

The post TECNO EllaClaw AI Agent Can Research News, Write Scripts and Check Social Posts appeared first on Gizmochina.

Buying a smartphone in 2026 often involves watching several reviews before making a decision. Buyers may move between first-impression videos, camera comparisons, gaming tests, battery reviews, and long-term coverage. The challenge is that useful information can be spread across hours of video, making it difficult to remember specific claims or compare different reviewers. 

Lynote is an AI-powered note-taking and transcription tool designed to solve this problem by turning long YouTube reviews and other source material into searchable, structured notes, making it easier to find, organize, and compare information. 

Why Smartphone Research Has Become Complicated

A single phone can generate a large amount of review content. One reviewer may focus on cameras, another on gaming performance, while others examine battery life, charging or software. Follow-up videos can also appear after software updates, adding new information to the original reviews.

Reviewers do not always agree, either. One may report better low-light photography, while another notices shutter lag. A third may focus on sustained gaming performance or chipset throttling. Battery endurance and real-world software performance can also produce different observations depending on how a device is tested.

Finding these differences can require repeatedly reopening several videos. An important comment may be buried 30 or 40 minutes into a review, forcing viewers to scrub through long sections just to confirm a particular claim.

From YouTube Videos to Searchable Text

Lynote’s approach begins with transcription. Its YouTube transcript generator can convert videos into full, timestamped text, allowing users to locate specific parts of a review without relying entirely on memory.

The workflow is intended to work with multiple videos, which can be useful when researching several reviews of the same smartphone. A buyer could gather a full review, camera test, gaming analysis, battery test, and long-term review and turn their spoken information into text.

This does not eliminate the original videos. Instead, transcripts provide another way to access information when users need to find a particular statement or return to a specific section.

AI Adds Structure to the Information

A transcript alone does not completely solve the problem. Long transcripts can still contain thousands of words, requiring users to search through different documents individually.

Lynote’s AI note-generation workflow adds an organizational layer. It can work with recordings, documents and pasted YouTube links, turning source material into structured notes. The aim is to make information easier to skim, search and compare.

For smartphone research, notes can help organize observations around areas such as performance, cameras, gaming, battery life, charging and software. Keeping notes associated with individual reviewers also makes it easier to identify where sources agree and where their experiences differ.

Creating One Research Trail From Several Reviews

The process can be viewed as a simple research chain: collect relevant reviews, generate transcripts, create structured notes, and compare the information by topic.

For example, a buyer researching two phones could examine what different reviewers said about sustained performance, camera behavior, battery endurance, and software. If reviewers reach similar conclusions, those observations are easier to identify. If they disagree, the buyer can return to the original video to examine the testing or context behind the claim.

The result is closer to a personal research document than a collection of separate videos. Users can search the notes for a specific issue and return to the original source when they need additional context.

The Research Continues After Launch

Smartphone coverage is also becoming a continuing process rather than a single launch-day event. A device can receive early hands-on coverage, initial reviews, deeper retail-unit testing, camera comparisons, gaming tests, software-update coverage, long-term reviews, and even teardowns.

That means the same research method can remain useful after a phone has been purchased. Owners may later search for information about firmware updates, camera algorithm changes, battery health, charging behavior, or software bugs. Some of this information can appear first in videos rather than traditional written articles.

Multilingual Reviews Create Another Challenge

Research can become even more fragmented for Chinese-market smartphones. Detailed coverage may appear on Mandarin-language channels, while global versions are reviewed separately in English, Hindi, Indonesian, and other languages.

Lynote’s transcript workflow supports video sources across languages. This could help users organize information from different regional sources while researching the same device. Such comparisons can be important because Chinese and global versions may differ in hardware, software, connectivity, or availability.

AI Notes Still Have Limits

AI-generated notes are not a replacement for watching important reviews. Some aspects of smartphone testing are difficult to capture fully in text. Camera color science, for example, is partly subjective, while gaming performance and camera samples are often better understood by seeing the original footage.

A structured note can tell a buyer what a reviewer said, but the original video may still be necessary to evaluate visual samples, demonstrations, and the reviewer’s overall experience. The workflow therefore works more as a research and organization layer than a substitute for the original content.

AI Note-Taking Moves Into Consumer Research

AI note-taking is also expanding beyond meetings, lectures and workplace discussions. Data projects the AI note-taking category to cross $740 million in 2026. The broader note-taking app market is projected to grow from approximately $11 billion in 2025 to around $13.3 billion in 2026, representing a projected 20.6% compound annual growth rate.

Consumer research is one area where that growth could have practical applications, particularly when information is scattered across lengthy videos and multiple sources.

For smartphone buyers, the main purpose of this workflow is to reduce the effort involved in finding and remembering information. Instead of keeping track of which reviewer discussed throttling, night photography, battery behavior, or software problems, users can organize those observations into searchable notes.

The basic process remains straightforward: video to transcript, transcript to structured notes, notes to comparison, and back to the original video when verification is needed. As smartphone coverage becomes more detailed, multilingual and spread across longer periods, tools such as Lynote are targeting the growing challenge of organizing all that information without removing the original reviews from the process. For more coverage of AI tools and how they’re changing tech research and writing, see Kemotech. 

The post Tired of Watching 5 Phone Reviews? Turn Them Into Searchable Notes Using the AI Tool Lynote appeared first on Gizmochina.

Huawei has unveiled the Atlas 960E SuperPoD, a new AI computing system designed to handle increasingly large artificial intelligence models. The company announced the system on September 17 at HUAWEI CONNECT 2026 in Shanghai.

Huawei says the Atlas 960E is the industry’s first SuperPoD based on Near-Packaged Optics (NPO). The system supports training and inference for AI models approaching 10 trillion parameters.

Built for Large-Scale AI

The Atlas 960E combines Huawei’s UnifiedBus technology, also known as Lingqu, with the company’s Hi-ONE optical system. The architecture uses unified memory addressing, allowing computing resources to work more like a single system rather than separate machines.

A single SuperPoD can support up to 4,096 Ascend AI processors and deliver up to 8 EFLOPS of FP8 performance and 16 EFLOPS of FP4 performance. Huawei says the system uses a fully liquid-cooled design and an orthogonal architecture.

Hi-ONE Cuts Optical Hardware

A key part of the system is Huawei’s Hi-ONE NPO optical engine. Huawei says it offers 7.2 terabits per second of transmission capacity and includes an integrated light source.

The Atlas 960E uses about 5,500 Hi-ONE units, replacing roughly 48,000 traditional 800G optical modules. Huawei claims the design can reduce power consumption by more than 550 kilowatts, double fault-free operating time and provide 99.8% system availability.

Scaling to 1 Million AI Processors

Huawei is also extending the SuperPoD approach to general-purpose computing. Its upgraded TaiShan 950 SuperPoD supports up to 4,096 nodes and a 256 TB unified memory pool, while the OceanStor M900 provides petabyte-scale key-value caching for AI workloads.

Huawei also plans to connect multiple Atlas 960E SuperPoDs into larger AI clusters using UnifiedBus and RoCE networking. The company says its architecture can connect up to 512,000 AI processors, while a multi-rail topology can scale an Ascend SuperCluster to 1 million processors.

The Atlas 960E is scheduled for Q3 2027. Huawei is also preparing the Ascend 960 DT for Q1 2027, the Ascend 960 PR for Q3 2027, followed by the Ascend 970 in 2028 and Ascend 980 in 2029.

Huawei Atlas 960E SuperPoD: Key Specifications

Read More:

• TECNO POVA 8 Pro 5G Tonino Lamborghini Limited Edition Blends Italian Design With Smart Technology
• Lenovo Legion C700 Now Available at Giztop: Specs and Price

(via)

The post Huawei Unveils World’s First NPO-Based AI SuperPoD Built for Massive AI Models appeared first on Gizmochina.

Sam Altman isn’t pretending that AI will be perfectly safe forever. Speaking at Salesforce’s Dreamforce conference, the OpenAI CEO said he believes a certain number of accidents with emerging technology are simply unavoidable. What matters most, he argued, is how the industry responds when they happen.

“I do think that some degree of accidents are unavoidable with new technology,” Altman said. “What I really care about is that we have got a great culture of accident reporting and learning.” He pointed to the Federal Aviation Administration and the National Transportation Safety Board as models. Those agencies dig deep into every incident, and that culture of learning is a big reason modern air travel is so safe.

His comments come at a time when people are watching leading AI labs more closely than ever, especially as systems get closer to greater autonomy. Recent remarks from Jacob Coxon, a researcher who recently left Anthropic after previously working at OpenAI, added fuel to the debate by warning that companies are racing toward self-improving superintelligence.

Altman didn’t dismiss the worries. He said public concerns about AI getting out of control or power becoming too concentrated in a few labs are completely understandable. “The world is right to be afraid of this,” he said.

However, instead of slowing everything down or stopping, Altman argued that companies need to keep work on alignment, safety, and monitoring well ahead of the advances in model capabilities. If they can’t maintain that balance, he said, they should be ready to slow or even pause research. The point isn’t to shrug and accept disasters. It’s to be honest when things go wrong, review them quickly, and keep improving so the risks don’t spiral.

These comments fit into a larger ongoing conversation about how fast frontier AI should move. Some industry voices have been calling for more caution, while US policy under the current administration has focused on staying ahead technologically.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

The post Sam Altman says some AI accidents cannot be avoided appeared first on Gizmochina.

OpenAI is making another change to its model lineup. GPT-5.5 is being phased out of several of its main products, which will affect both regular users and developers who’ve gotten used to working with it.

According to the company, GPT-5.5 will be retired from ChatGPT, ChatGPT Work, and Codex on October 14, 2026, across all plans. For most everyday ChatGPT users, the switch should feel pretty seamless: conversations will just move over to whatever model OpenAI sets as the new default. The bigger impact is for people who’ve built workflows around GPT-5.5, especially developers using it through Codex.

If you’re a Codex user signed in with a ChatGPT account, you’ll want to migrate before the cutoff. Official guidance points to switching to GPT-5.6 Sol (or GPT-6 Astra in some references). That means updating workspace defaults, saved model settings, managed configurations, custom agents, scheduled tasks, and any scripts that specifically call for GPT-5.5. The good news is this retirement doesn’t affect the OpenAI API itself. Anything authenticated with an API key should keep working as usual.

This fits the pattern OpenAI has been following lately of retiring older models once newer ones are ready. GPT-5.4 class models were already pulled from Codex at the end of August 2026. GPT-5.5 itself only launched on April 23, 2026, so it’s had a relatively short run in the main product surfaces. Developers who hardcoded the model name into apps or automation tools will need to make the most updates to avoid any disruption once it’s no longer available in those interfaces.

OpenAI is framing the change as part of keeping things current. If you’re still relying on GPT-5.5 in the affected products, you’ve got until October 14 to make the switch. API users get to keep access for now, which should help some production setups stay uninterrupted while the consumer and ChatGPT-authenticated Codex experiences move on.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: @ChatGPT on X)

The post OpenAI retiring GPT-5.5 on October 14: you may need to update your workflow appeared first on Gizmochina.

ALLTIME has officially unveiled oLand, a unique desktop companion that combines artificial intelligence, holographic technology, and a miniature living ecosystem into one interactive experience. Marketed as the world’s first AI holographic companion, oLand is designed to make workspaces feel calmer and more personal by blending nature with AI. The product is now available on Kickstarter, where early supporters can access discounted pricing and exclusive rewards.

A New Approach to AI Companionship

Founded in Hangzhou, China, and backed by MiraclePlus (formerly YC China), ALLTIME aims to create a new form of AI companionship rather than another chatbot. The company’s vision is to offer users a virtual companion that provides emotional support and everyday interaction without the responsibilities of caring for a real pet.

Instead of demanding attention, oLand is designed to quietly fit into daily life, creating a relaxing desktop environment that feels more alive.

A Living Holographic Desktop World

At the heart of oLand is a miniature ecosystem that users can customise with live plants, Marimo moss, figurines, crystals, and other decorative items. Above this physical landscape, holographic AI characters appear, creating the illusion that they live inside the tiny world.

This combination of real and virtual elements allows users to build a personalised desktop display that evolves throughout the day.

The flagship companion is MOLI, inspired by the popular Marimo moss ball. With more than 100,000 fans since its original debut in 2014, MOLI features its own personality, emotional responsiveness, and long-term memory. It remembers previous conversations and gradually adapts to each user over time.

Interactive AI Pets with Unique Personalities

oLand supports multiple AI companions, each with its own personality and special abilities. MOLI, for example, shares playful fortunes and thoughtful daily messages designed to encourage mindfulness and brighten everyday routines.

Users can switch between different AI pets, including characters like Flick, using NFC-enabled interactive pods. These handheld pods also function as remote controllers and squeezable stress-relief toys, allowing users to interact with their virtual companions through simple physical actions.

Designed to Help Reduce Everyday Stress

One of oLand’s standout features is its focus on relaxation. Squeezing the Interactive Pod activates Bubble Pop Mode, a holographic mini-game designed to provide a quick 30-second mental reset between meetings or during busy workdays.

The device also includes a one-minute Meditation Mode featuring guided breathing, calming audio, gentle lighting effects, and participation from the AI companion to help users relax and refocus.

ALLTIME says the AI remains quietly present throughout the day, offering subtle interactions without becoming distracting.

AI That Learns and Adapts

oLand supports natural voice conversations, allowing users to speak with their AI companion while benefiting from long-term memory and customisable voices. According to ALLTIME, conversations are protected with strong privacy measures.

Using built-in millimetre-wave radar, the device detects when users approach or leave their desk and responds with contextual animations. It can also recognise extended periods of sitting and gently encourage short breaks. Users remain in full control and can ignore or disable proactive interactions whenever they choose.

The AI is designed to evolve over time by learning from conversations and everyday behaviour, creating a more personalised companionship experience.

A World That Reflects Real Life

The holographic environment automatically synchronises with local time, weather, seasons, and the day-night cycle, making the miniature world feel more dynamic. Users can also upload favourite videos through the Memory Projection feature, allowing treasured moments to appear as floating holograms above the desktop ecosystem.

Explaining the vision behind oLand, ALLTIME’s founder said: “We explore ways to bring nature back into everyday spaces, not as decoration or something that requires constant care, but as a quiet, functional presence that fits naturally into daily life.”

Kickstarter Launch

oLand is now available through Kickstarter. Early supporters can receive up to 33% off Early Bird pricing, exclusive oLand Blind Box collectibles, behind-the-scenes development updates, and opportunities to help shape future product features.

With oLand, ALLTIME is introducing a fresh take on AI companionship by combining holographic technology, a living desktop ecosystem, and emotionally aware AI into a single product.

The post ALLTIME Launches oLand, World’s First AI Holographic Companion appeared first on Gizmochina.

OpenAI is moving forward on two fronts at once. Recent developments point to both immediate improvements in its AI models and more details about a distinctive physical product that could change how people interact with ChatGPT.

A Bloomberg report has given a clearer look at OpenAI’s first hardware device, described as the “physical manifestation of ChatGPT.” The product is expected to take a doughnut-like shape roughly the size of a hockey puck, made from high-quality metal and notably without a display.

It’s battery-powered for portability, so it could sit on a nightstand, kitchen counter, or desk, or be carried around the house. It’s said to include a camera system and sensors for visual input, along with speaker grilles, microphones, and lights that show when it’s listening.

Moving parts may also give the device a bit of personality by signaling responses. Pricing is projected in the $300 to $400 range, higher than typical smart speakers, with a potential launch in 2027. OpenAI has indicated this could be the first of several devices.

On the software side, OpenAI has rolled out some meaningful updates to GPT-5.6 that users can take advantage of right now. GPT-5.6 Sol now features dynamic response lengths that adapt to how complex a query is, roughly 68% fewer factual errors, and a new reasoning slider for Plus and Pro subscribers that works similarly to effort settings in other models.

Meanwhile, GPT-5.6 Luna has become the default for Free and Go users, offering unlimited text chats and a Think button for deeper reasoning. There are also additional safeguards for users under 18 that block romantic roleplay, age-restricted content, and self-harm prompts.

These parallel moves highlight OpenAI’s dual approach: improving accessible AI tools while preparing premium hardware that puts its models more deeply into everyday life. The doughnut-shaped speaker, if it actually arrives, would move ChatGPT out of apps and browsers and into a tangible companion, while the model updates deliver practical gains in accuracy and usability right away. Together, they show a company expanding beyond pure software toward a more integrated ecosystem.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source | Image)

The post OpenAI’s first hardware is taking shape as a doughnut-shaped speaker amid improvements in GPT-5.6 Sol appeared first on Gizmochina.

Shared Claude conversation links started showing up in public Google search results, putting discussions online that a lot of users probably assumed would only be visible to people who had the direct link.

It first came to light in a Reddit post on r/ClaudeAI. Someone shared a screenshot of Google results for the search “site:claude.ai/share,” and it turned up a bunch of clickable links to actual Claude conversation pages. Claude’s sharing feature lets you create public links so others can see the chat, handy for things like brainstorming or design work. The idea is that anyone with the URL can open it.

People familiar with the feature say the shared pages were supposed to have a “noindex” tag. That’s a standard web instruction that lets search engines look at a page but tells them not to list it in public results. Google and Bing normally respect it. In this case, some of the shared pages apparently didn’t have the proper noindex marking when they were crawled, so they got indexed and became findable through search. Even with a noindex tag, the pages would still work if you had the direct URL. The real problem was that those URLs became discoverable through Google.

The conversations that got exposed covered a pretty wide range of sensitive stuff: legal strategy talks, technical troubleshooting, personal chats, and other private or confidential material. Once they were indexed, anyone who ran the right search could see them.

Fortunately, the related chat logs have been removed from Google’s index. Searching “site:claude.ai/share” no longer brings up those links. If you’ve shared any Claude conversations, it’s a good idea to go check and manage them under Privacy > Shared Chats in the settings.

This kind of thing keeps coming up as more people use AI tools for work and personal stuff. Features that are meant to make collaboration easy can accidentally open the door to wider exposure when the technical safeguards (like proper indexing controls) aren’t fully in place or applied consistently. Other AI platforms have had similar slip-ups before.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Reddit | ITHome)

The post Claude AI’s chat sharing links got indexed by Google, exposing personal chats appeared first on Gizmochina.

China’s AI scene just took another big step forward with the launch of Moonshot AI’s latest model, Kimi K3.

The Chinese AI company has released what it claims is the largest AI model to date with open weights, packing 2.8 trillion parameters.

Moonshot is calling it frontier-level, and they plan to release the full model weights soon, meaning developers and enthusiasts will eventually be able to download, run, and fine-tune it locally.

For now, you can already try K3 through the Kimi app on iOS and Android, the kimi.com website, or the Kimi Work desktop app. No developer account is needed to get started on the free tier (though there are usage limits).

K3 uses a Mixture-of-Experts (MoE) architecture, activating 16 out of 896 experts per request. It comes with some serious specs: a 1-million-token context window, native support for images (and reportedly video), and clever tricks like Kimi Delta Attention for handling long conversations efficiently.

Early benchmarks show it’s competitive. It trails some of the absolute top Western models like Claude Fable 5 and GPT-5.6 in certain areas, but beats or matches others in tasks like coding, agentic workflows, and knowledge work. Independent tests from Artificial Analysis put it in a strong position overall.

API pricing starts at $3 per million input tokens and $15 per million output tokens (with caching discounts). Moonshot says the full open weights will drop on July 27. That’s a huge deal for anyone interested in self-hosting or customizing the model. The company admits K3 still has some catching up to do in overall user experience, but it’s a big leap for China’s AI ecosystem.

If you’re curious about high-end AI development outside the usual Western players, Kimi K3 is worth checking out, especially once the weights are released later this month. It could spark a lot more experimentation and competition in the open-source community.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Kimi)

The post Kimi K3: Moonshot AI unleashes 2.8 trillion parameter model for free appeared first on Gizmochina.

Artificial intelligence on smartphones is entering a new phase. According to Counterpoint Research, the industry is moving beyond AI assistants that simply answer questions and toward “Agent Phones” that can understand a user’s intent and complete tasks automatically. Instead of opening apps one by one, future smartphones are expected to coordinate multiple apps and services in the background, making everyday tasks faster and easier. This shift could redefine how people interact with their devices and how smartphone brands compete in the years ahead.

From AI Assistants to AI Agents

Today’s AI assistants are mainly designed to respond to questions, generate text, or provide recommendations. Counterpoint believes the next step is far more practical. Agent Phones are designed to understand what users want to achieve and then carry out the necessary actions automatically.

Rather than asking users to switch between several apps, these AI agents can manage complete workflows on their behalf. This changes the smartphone from a device that launches apps into an intelligent operating system layer that coordinates apps and services behind the scenes. As a result, future competition may depend less on hardware specifications and more on how reliably AI can complete real-world tasks.

Open Frameworks Are Accelerating the Shift

One of the biggest drivers behind this transition is the emergence of open “Claw” frameworks. These frameworks provide a shared execution layer that allows AI agents to understand user intent, execute multi-step tasks, and work across different applications.

Counterpoint highlights OpenClaw as a leading open-source framework in this space. Because these frameworks are openly available, smartphone manufacturers no longer need to build complete AI agent systems from scratch. This reduces development costs and engineering complexity, making advanced AI features more accessible across different smartphone price segments.

Agent-to-Agent Communication Improves Automation

Another important development is Agent-to-Agent (A2A) communication. Instead of relying on a single AI assistant, multiple AI agents can communicate with one another, share information, delegate responsibilities, and coordinate workflows.

This approach helps remove traditional app boundaries, allowing different services to work together more smoothly. The result is a more seamless experience where users spend less time manually managing apps and more time letting the phone handle routine tasks.

Smartphone Competition Is Changing

Counterpoint believes the smartphone industry is entering a new stage where AI execution matters more than standalone AI features or benchmark scores. Future devices are expected to compete on factors such as task execution quality, contextual understanding, ecosystem integration, intelligent automation, and overall reliability.

The research also outlines two different paths toward agentification. Internet companies are likely to build AI agents around their existing app ecosystems, while smartphone manufacturers are expected to integrate system-native AI agents directly into the operating system for deeper control over device functions and cross-app interactions.

TECNO EllaClaw 

Counterpoint points to TECNO’s EllaClaw as an example of how this new approach can work in practice. Built using the Ella AI framework, the open-source OpenClaw platform, and Agent-to-Agent architecture, EllaClaw transforms a traditional AI assistant into an AI agent capable of completing tasks across multiple applications.

Its features include cross-app task execution and a one-tap phone caretaker that helps address battery drain, excessive data usage, device maintenance, and routine optimization. TECNO also focuses on solving practical problems for users, particularly in emerging markets, instead of concentrating only on AI model size or benchmark performance.

What’s Next?

Counterpoint believes the future of smartphones lies in practical AI that can complete tasks instead of simply responding to prompts. As open frameworks such as OpenClaw become more widely adopted, agentic AI is expected to expand beyond premium flagship devices and reach mainstream smartphones. In the coming years, the most successful AI phones may not be those with the largest AI models, but those that deliver the most useful, reliable, and intelligent everyday experiences.

Read More:

• TECNO Wants AI to Represent Everyone Fairly—Here’s How ‘100 Portraits of Becoming’ Works
• TECNO Unveils Upgraded EllaClaw AI with Cross-App Automation and 40+ Smart Skills

The post AI Phone Is Evolving Into the Agent Phone: How Smartphones Are Learning to Get Things Done appeared first on Gizmochina.

As AI agents get smarter and start handling more complex tasks on our phones, one big challenge is keeping users in the loop without forcing them to constantly switch apps. Google is tackling this with a new feature called Android Halo.

In a recent video on Google’s official developer YouTube channel, Android President Sameer Samat gave more details about it (around the 30-minute mark). First previewed at Google I/O in May, Android Halo is designed to make communication between users and their AI assistants much more seamless and transparent.

A dedicated spot in the status bar

The idea is simple but clever: Android Halo creates a persistent space in the phone’s status bar for your chosen AI agent, whether that’s Gemini or a compatible third-party one. When the agent is working on something, you’ll see a subtle indicator there. It can show progress updates, ask you quick follow-up questions, or display final results without you needing to open the app again.

Samat explained that while AI agents are getting good at running multiple tasks on their own, they still need human input at key moments. Halo acts as a dedicated channel for three main things: asking clarifying questions, sharing updates, and showing completed work.

“This is a dedicated location on the status bar where the user-selected agent can report updates to the user and receive the user’s latest instructions for the current task queue,” he said. He called it an interesting new direction for how we’ll interact with intelligent systems in the future.

By keeping everything visible and easy to access, Halo brings more transparency to background AI activity. It makes these smart agents feel less like mysterious black boxes and more like helpful partners, which should help build user trust.

The feature is expected to arrive later in 2026 with Android 17. It fits nicely into Google’s broader Gemini plans and could work alongside other new AI capabilities. While we’re still waiting for full details on rollout and supported devices, the goal is clear: make AI on Android feel more natural and integrated into daily use.

This move shows where the industry is headed, toward more proactive, agent-style AI. Features like Halo could become essential as these systems take on bigger responsibilities. It’ll be interesting to see how it evolves.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Google for Developers on YouTube)

The post Google explains Android Halo: a status bar communication layer for AI agents appeared first on Gizmochina.

Browser extensions are supposed to make life easier, but some can come with serious privacy risks. Microsoft’s security researchers recently discovered a sneaky Chrome extension pretending to be the popular AI search engine Perplexity AI. Instead of helping users, it was secretly monitoring search activity and redirecting it through attacker-controlled servers.

The extension, called “Search for perplexity ai”, looked pretty harmless on the Chrome Web Store. But behind the scenes, it was doing some shady stuff. Microsoft Threat Intelligence broke down the details in a blog post on June 30, and Google quickly pulled it from the store. Still, if you installed it, it’s not gone until you remove it yourself.

The extension asked for broad permissions that went way beyond what a normal search helper needs. It used “chrome_settings_overrides” to take over your default search engine, so every time you typed something in the address bar, it first sent the query to a fake domain, “perplexity-ai.online”, before eventually passing it on to real search results.

It also used “declarativeNetRequest” permissions to monitor and modify network traffic. This gave it the ability to log your searches and even see real-time typing suggestions. There’s no sign it stole passwords or login info, but it could easily build up a detailed picture of your interests and online habits.

For reference, the real Perplexity AI lives at perplexity.ai. Always double-check the domain and the developer before installing anything, especially AI-related tools that are trending right now.

If you have the extension, remove it right away. Here’s the quick way:

1. Open Chrome and go to “chrome://extensions/”
2. Turn on Developer mode (top right corner)
3. Find the extension with the ID “flkebkiofojicogddingbdmcmkpbplcd” and click Remove
4. While you’re there, check your default search engine settings and reset them if needed

This whole episode is a good reminder that the Chrome Web Store isn’t perfect. Even with review processes, scammers are getting better at impersonating popular services. Microsoft suggests regularly reviewing your extensions, being picky about permissions, and sticking to well-known developers when possible.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source)

The post Sneaky Chrome extension discovered secretly monitoring search activity appeared first on Gizmochina.

Robot lawn mowers are becoming more popular, but many models still feel too expensive or too complicated for average homeowners. Some require buried boundary wires, difficult setup, oversized hardware, or advanced RTK systems that small garden owners simply do not need.

A Simple Robot Mower Designed for Everyday Use

The RoboUP Raccoon 2 SE takes a much simpler and more practical approach. Designed specifically for lawns up to 600㎡, the Raccoon 2 SE focuses on convenience, beginner-friendly operation, and stress-free lawn care. Instead of overengineering the experience, RoboUP delivers a mower that works straight out of the box and feels easy to use from day one.

For first-time robot mower buyers, busy families, seniors, and homeowners tired of pushing heavy lawn mowers every weekend, the Raccoon 2 SE feels like a smart and practical upgrade.

Mow Straight Out of the Box

One of the biggest advantages of the Raccoon 2 SE is its simplicity. Many robotic lawn mowers require complicated installation before mowing can even begin. RoboUP removes most of that frustration with a simple one-press operation.

With a single press on the control panel, the robot mower automatically maps the lawn and starts mowing immediately. There is no need for buried wires, RTK systems, or complicated setup procedures. For simple lawns, users do not even need to download an app to get started.

That “just press play” experience makes the mower especially appealing for beginners who want automated lawn care without technical complexity.

Highly Efficient Daily Lawn Care

Although the Raccoon 2 SE focuses heavily on ease of use, it still includes impressive performance features. The mower uses full-map parallel mowing instead of random mowing patterns, helping it deliver cleaner and more organized lawn coverage. According to RoboUP, this system is up to 1.6× more efficient than random mowing.

The mower also features a 5Ah battery capable of delivering up to 150 minutes of runtime while supporting fast charging in around 70 minutes. Once the battery gets low, the mower automatically returns to the charging dock, recharges itself, and resumes mowing automatically.

Its 20 cm cutting blade and powerful blade rotation help maintain smooth and even trimming results, allowing homeowners to enjoy a consistently neat lawn with very little manual effort.

Flagship Features and Real-World Versatility

RoboUP also includes several flagship-style smart features in the Raccoon 2 SE. The mower uses an AI camera system combined with bumper-based obstacle avoidance to navigate around objects more effectively.

The mower is designed for real-world gardens rather than perfectly flat lawns. It can handle slopes up to 36%, narrow passages as tight as 80 cm, potholes up to 5 cm deep, and steps up to 4 cm high. That versatility makes it much more practical for everyday home gardens.

Another major advantage is its quiet operation. With noise levels at or below 56 dBA, the mower runs quieter than many traditional lawn mowers, making daily lawn maintenance far less disruptive.

The mower also includes useful features like rain protection and automatic recharge functionality, helping it operate more independently with less supervision.

Prime Day Offer Makes It an Even Better Value

This Prime Day, the RoboUP Raccoon 2 SE is available at one of its lowest prices ever, making wire-free robotic mowing more affordable than ever.

RRP: $699

Prime Day Price: $521

Save: $178 (25% OFF)

For homeowners looking to upgrade from traditional mowing or purchase their first robotic mower, Prime Day offers an excellent opportunity to enjoy hassle-free lawn care at a significantly reduced price. With no boundary wires, no RTK station, and a simple drop-and-mow experience, the Raccoon 2 SE delivers exceptional value for small and medium-sized lawns.

Strong Early Market Validation

The Raccoon 2 SE officially launched on March 18 and has already received very positive market feedback. RoboUP is currently cooperating with major retail partners, including Leroy Merlin, FnacDarty, OBI, Bauhaus, and Hornbach.

Before launch, the mower was validated by more than 20 professional media outlets and tested by over 80 beta testers in 2025. RoboUP also gained strong early support through its Kickstarter debut in late 2025, attracting more than 600 backers.

RoboUP Raccoon 2 SE succeeds because it focuses on what small garden owners actually need. It combines beginner-friendly operation, efficient mowing, useful flagship features, and affordable pricing into a package that feels practical and easy to use.

For homeowners looking for their first robot lawn mower under €500, the Raccoon 2 SE feels like one of the smartest and most convenient options currently available.

Availability & Prime Day Deals

Raccoon 2 SE
RRP $699 → Prime Day Price $521 (Save $178, 25% Off)

• Buy on Amazon
• Buy on the RoboUP Official Store

T1200 Pro
RRP $1,549 → Prime Day Price $799 (Save $750, 48% Off)

• Buy on Amazon
• Buy on the RoboUP Official Store

Available now on Amazon and the RoboUP Official Store. Prime Day offers are available for a limited time.

The post RoboUP Raccoon 2 SE Deal: AI Obstacle Avoidance and Easy One-Press Mowing appeared first on Gizmochina.

TECNO has unveiled the next stage of development for EllaClaw, its beta-stage mobile AI agent designed to make smartphones more intelligent and proactive. Announced on June 24, 2026, EllaClaw goes beyond traditional voice assistants by completing everyday tasks across apps, optimizing device performance, and providing personalized assistance. The company says its goal is to make advanced AI practical, accessible, and useful for everyday users, especially in emerging markets.

EllaClaw Brings Practical Agentic AI to Smartphones

Unlike conventional AI assistants that mainly respond to commands, EllaClaw is a cloud-based AI agent that can understand requests, plan actions, and complete tasks on a user’s behalf. It works proactively in the background to simplify complex workflows while keeping users informed about every action it performs.

TECNO says EllaClaw is designed to automate real-world tasks instead of simply answering questions, making smartphones more capable of handling everyday activities with minimal user effort.

Cross-App Intelligence Simplifies Everyday Tasks

One of EllaClaw’s biggest upgrades is its cross-app intelligence. Instead of relying on direct integrations, the AI understands graphical user interfaces (GUI), allowing it to navigate supported third-party apps much like a human user.

With user permission, EllaClaw can interact with shopping, transportation, food delivery, and smart home apps. It performs actions visibly on the screen so users can follow every step, providing greater transparency and confidence in AI-driven automation.

AI Optimizes Phone Performance with 40+ Smart Skills

EllaClaw also works as an intelligent device management assistant by monitoring and optimizing battery life, mobile data, storage, RAM, and CPU performance.

TECNO has introduced more than 40 Smart Skills to improve everyday smartphone usage. Smart CleanUp Boost frees RAM, optimizes CPU resources, and reduces system lag for smoother performance. Smart Power Drain Check detects apps that consume excessive battery and optimizes background activity to extend battery life.

The AI also includes Instant Cool-down Relief, which lowers device temperature during heavy workloads by managing background processes, while Smart Data Guardian tracks mobile data usage, learns personal usage habits, and helps users avoid excessive data consumption. TECNO says this feature is particularly useful in regions where mobile data plans remain limited.

Personalized Assistance for Daily Life

EllaClaw becomes more personalized through persistent memory, allowing it to learn user habits and preferences over time.

Its Morning Briefing combines calendar events, travel itineraries, weather updates, and curated news into a single overview. The Trip Prep Assistant helps organize travel schedules, book rides, set departure alarms, and prepare actionable travel plans before users leave.

TECNO has also added relationship-focused features that remind users to contact family members or partners, notify them about weather changes that could affect plans, and provide contextual reminders for daily activities.

Shopping, Ride Booking, and Smart Home Control

EllaClaw extends its capabilities through several cross-app AI experiences. Users can book transportation using natural language with One-Sentence Ride Hailing, while Shopping Buddy searches for products and compares options across shopping platforms such as Lazada.

The AI can also monitor connected smart home devices, provide status updates, and support smart home routines, allowing users to manage multiple connected devices through a single intelligent assistant.

Privacy and User Control Come First

TECNO says EllaClaw follows a confirmation-first privacy model. The AI asks for user approval before making important system changes or accessing supported applications, ensuring users remain in full control of their devices.

The assistant also performs actions visibly instead of silently in the background, allowing users to see exactly what it is doing. Supported apps require secure opt-in permissions before EllaClaw can access them.

Availability

TECNO says EllaClaw is currently undergoing internal testing and is available only as a closed beta. The company describes it as an exploratory AI concept and plans to announce additional features and broader availability in future updates.

With cross-app automation, system-level device optimization, more than 40 Smart Skills, and personalized assistance, TECNO is positioning EllaClaw as a practical mobile AI agent rather than a traditional chatbot. By combining proactive task automation with transparent, confirmation-first privacy controls, the company aims to make advanced AI more useful for everyday smartphone users while ensuring they remain in complete control of their devices.

The post TECNO Unveils Upgraded EllaClaw AI with Cross-App Automation and 40+ Smart Skills appeared first on Gizmochina.

CHUWI has officially launched pre-orders for its new UniBook AI Laptop, introducing an affordable Windows notebook built around Intel’s latest Wildcat Lake platform. The laptop combines a lightweight design, AI-focused hardware, a color-accurate display, and a generous selection of ports, making it suitable for students, professionals, freelancers, and everyday users. 

With an early bird price of $449, CHUWI is positioning the UniBook AI as an accessible option for users looking to experience on-device AI features without spending flagship laptop prices.

Early Bird Pricing and Launch Campaign

The CHUWI UniBook AI carries a regular retail price of $499, but customers who pre-order the laptop can purchase it for $449, saving $50 during the launch period. The offer is part of CHUWI’s “AI for Everyone, The Tournament for All” World Cup promotional campaign.

The company is also extending discounts of up to $50 on several other products, including the CoreBook Air 226V, AuBox X Copilot+ PCs, Hi10 Max 2-in-1 Windows Tablet, and CoreBook Air 1KG, giving buyers additional options if they are looking for other Windows devices.

Intel Core 3 Wildcat Lake Processor Takes Center Stage

Powering the UniBook AI is Intel’s latest Core 3 Processor 304, based on the company’s advanced 18A process technology. The new processor is designed to deliver efficient everyday computing while introducing dedicated hardware for AI workloads.

The laptop uses Intel Graphics for visual performance and is aimed at handling common productivity tasks such as web browsing, office work, online learning, media consumption, and light creative applications. The new architecture also focuses on better power efficiency, helping extend battery life without sacrificing performance for everyday users.

Dedicated AI Hardware for Faster Local Processing

One of the biggest highlights of the UniBook AI is its built-in AI acceleration. The integrated graphics processor delivers up to 9 TOPS (Int8) of AI performance, while the dedicated Neural Processing Unit (NPU) provides up to 15 TOPS (Int8) for AI-specific workloads.

Instead of sending every AI request to cloud servers, the laptop can process many tasks locally. This improves response times while keeping more personal data on the device. CHUWI says the system supports AI writing assistance, real-time transcription, AI image generation, and AI-enhanced multitasking, allowing users to benefit from modern AI features during everyday work.

14-Inch WUXGA Display Built for Productivity

The UniBook AI features a 14-inch IPS display with a WUXGA (1920 × 1200) resolution. Unlike standard Full HD displays, the taller 16:10 aspect ratio provides additional vertical workspace, allowing users to view more content while editing documents, browsing websites, or working with spreadsheets.

The display also covers 100% of the sRGB color gamut, making it suitable for users who require accurate colors for photo editing, graphic design, or content creation. With 300 nits of brightness and an 88% screen-to-body ratio, the display is designed to deliver a comfortable viewing experience for both work and entertainment.

Lightweight Design Made for Daily Travel

CHUWI has designed the UniBook AI with portability in mind. The laptop features a lightweight metal chassis that weighs approximately 1.2kg, while its slim 16.4mm profile makes it easy to slip into a backpack or laptop bag.

The laptop measures 314.5 × 221 × 16.4mm, making it compact enough for classrooms, offices, cafés, and business travel. A 180-degree lay-flat hinge also allows the display to open completely flat, making it easier to share content with colleagues or classmates during meetings and presentations.

Comfortable Keyboard and Privacy Features

The laptop includes a backlit keyboard, allowing users to work more comfortably in dimly lit environments. A precision touchpad provides smoother navigation and supports Windows gesture controls for improved productivity.

CHUWI has also included an HD webcam for video calls along with a physical privacy shutter. This simple feature allows users to physically block the camera whenever it is not being used, offering an extra layer of privacy during remote work and online meetings.

Memory, Storage, and Battery Life

The UniBook AI comes equipped with 8GB of LPDDR5 memory running at 6400 MT/s, providing fast memory speeds for everyday multitasking. Storage is handled by a 256GB PCIe 3.0 SSD, and users can expand the storage if they need additional space for files and applications.

Power is supplied by a 53.38Wh battery, which CHUWI claims can deliver up to 13 hours of local 1080p video playback under suitable conditions. The laptop also supports 65W USB-C Power Delivery, allowing users to recharge using modern USB-C chargers instead of carrying a dedicated power adapter.

Extensive Port Selection Sets It Apart

One area where the UniBook AI stands out is connectivity. Many modern thin-and-light laptops have reduced the number of physical ports, often requiring users to purchase USB hubs or docking stations. CHUWI takes a different approach by including a wide selection of built-in ports.

The laptop offers two full-featured USB-C ports, three USB-A ports (including two 5Gbps ports), HDMI, Gigabit Ethernet, a TF card slot, and a 3.5mm headphone jack. This allows users to connect external storage, wired internet, projectors, monitors, cameras, keyboards, and headphones without relying on adapters.

Modern Wireless Connectivity and Multi-Display Support

For wireless connectivity, the UniBook AI supports Wi-Fi 6, providing faster wireless speeds and improved network stability on compatible routers. Bluetooth 5.2 enables reliable connections with wireless headphones, mice, keyboards, and other accessories.

The laptop can also drive up to two external 4K displays at 60Hz, giving users additional screen space for multitasking, presentations, video editing, financial work, or office productivity.

Designed for Everyday Productivity

With its Intel Core 3 Wildcat Lake processor, dedicated AI hardware, lightweight metal design, color-accurate display, long battery life, and generous port selection, the CHUWI UniBook AI is designed to meet the needs of a wide range of users. Combined with Windows 11 Pro and an attractive early bird price of $449, the laptop offers an affordable entry point for those looking to experience the latest AI-powered PC features without paying premium flagship prices.

Key Specs:

The post CHUWI’s Affordable AI Laptop Launches with Intel Core 3, 13-Hour Battery, and Dual 4K Support appeared first on Gizmochina.

The smartphone world is changing fast. Generative AI features are quickly moving from fancy extras on premium phones to something people expect on pretty much everything. According to Counterpoint Research, phones with GenAI capabilities are on track to make up nearly half of all global shipments next year, even while the overall market is heading for its biggest drop in years.

Counterpoint’s latest forecast shows GenAI-equipped smartphones jumping from 36% of shipments in 2025 to 45% in 2026, and then up to 52% in 2027. It’s becoming a standard feature faster than many expected. Apple has already unveiled several GenAI features that are expected to arrive with the iPhone 18 series this September. Android OEMs, meanwhile, are moving beyond basic AI tools and focusing on more advanced “Agentic AI” experiences for their flagship devices.

But here’s the twist: this AI boom is happening while the broader smartphone market is struggling. Global shipments are expected to fall 13.9% in 2026, down to just 1.08 billion units, reportedly the lowest level since 2013. The main reason? A serious memory chip shortage. Manufacturers are shifting production toward high-profit AI server memory (like DRAM and HBM), which has caused prices for the memory used in phones (LPDDR) to shoot up.

This is hitting cheaper phones the hardest. Some budget and entry-level models might even disappear from shelves. Higher-end GenAI phones can absorb the increased costs more easily, so we’re seeing even stronger premiumization, with people either buying flagship phones or looking for refurbished ones.

This situation clearly benefits big players like Samsung and Apple, who already dominate the high-end smartphone market. Samsung is in an especially strong position because it also produces a large share of the advanced memory chips powering AI servers and data centers. Meanwhile, brands that rely on selling lots of affordable phones, especially in emerging markets, are feeling the squeeze.

On the bright side, as GenAI features get smarter (on-device image generation, intelligent assistants, advanced photo editing, etc.), they’re giving people real reasons to spend more. Once the memory supply situation eases, probably late 2027, the industry could come out stronger with more capable phones overall.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Counterpoint)

The post Over 50% of smartphones in 2027 will include advanced AI appeared first on Gizmochina.

In a pretty big surprise for the AI world, China’s Zhipu AI’s latest model, GLM-5.2, just grabbed the top spot in one of the most respected creative coding benchmarks out there.

On June 19, 2026, Design Arena announced on X that GLM-5.2 has taken the #1 position in their single-round HTML web design leaderboard (non-agent category). The platform, basically the world’s first big crowdsourced benchmark where real people blindly vote on AI-generated designs, says it beat out Claude Fable 5, along with the Opus 4.6 and 4.7 versions. That’s a solid five-place jump from its predecessor, GLM-5.1, landing it with an Elo score around 1360 in the Code Categories arena.

What makes this win stand out is how the model actually designs. It’s apparently excellent at clean layouts, pulling in images from CDNs, typography, visual hierarchy, and those nice subtle animations that make a site feel alive. It also plays really nicely with popular libraries like Chart.js and Three.js. According to the results, it boosted its win rate by about 6 percentage points this time around.

One interesting detail: GLM-5.2 leans heavily on Tailwind CSS (showing up in 91% of its designs) and Font Awesome (51%). By comparison, Fable 5 only used Tailwind in about 57% of its sessions, which might explain some of the practical gaps people noticed.

Beyond the design quality, the pricing is a breath of fresh air. GLM-5.2’s API runs around $1.40 per million input tokens and $4.40 for output, way more affordable than Fable 5’s $10/$50. Plus, it’s an open-weights model under the MIT license with a solid 1M-token context window, which means teams can run it locally and tackle bigger, longer projects without being locked into a proprietary service.

Design Arena’s approach gives these results real weight: millions of votes from actual creators focused on real aesthetics and usability, not just synthetic tests. This isn’t just a win for web design; it puts GLM-5.2 among the strongest open models across broader coding benchmarks too.

For the AI community, this feels like another sign that Chinese labs are moving fast in specialized creative areas. While Claude Fable 5 is a strong player, GLM-5.2’s mix of strong performance, low cost, and openness could make it a go-to choice for a lot of web development teams going forward.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Design Arena)

The post China’s GLM-5.2 beats Claude Fable 5 in web design, claims No. 1 spot appeared first on Gizmochina.

OpenAI may be getting ready to launch its next-generation AI model, GPT-5.6, as early as next week, according to reports circulating among testers and AI enthusiasts. While the company has not officially confirmed the model, growing speculation suggests that GPT-5.6 could bring major improvements in reasoning, coding, and long-context understanding.

The rumored lineup is expected to include GPT-5.6 Mini, GPT-5.6 Standard, and GPT-5.6 Pro. Several OpenAI Pro subscribers have claimed they already received access to a newer model through GPT-5.5 Pro, fueling speculation that OpenAI is quietly testing GPT-5.6 before a public launch.

Users Claim ChatGPT Suddenly Feels Smarter

The latest wave of rumors began when many ChatGPT users reported noticeable improvements in the platform’s performance. Testers on X claimed the AI appeared more capable in reasoning, coding, and creative tasks.

Developer Anshu Chimala shared side-by-side comparisons of AI-generated landing pages and said the outputs were significantly better than previous versions. Other early testers reported that the model required fewer corrections and delivered stronger first-attempt results.

Several users also suggested that the AI now spends more time thinking before responding, leading to higher-quality outputs despite slower generation speeds.

Improved Reasoning and Problem-Solving

One of the biggest rumored upgrades is a major improvement in reasoning capabilities. Early feedback suggests GPT-5.6 can better understand complex instructions, plan multi-step tasks, and solve difficult problems with fewer mistakes.

Testers reported that projects which previously required multiple revisions and 20 to 40 minutes of work can now be completed more efficiently. While some front-end web development limitations reportedly remain, users describe the model as showing deeper understanding and more reliable decision-making.

According to leaked information, OpenAI may have increased an internal reasoning parameter known as “Juice Value” from 768 to 960, potentially contributing to stronger performance.

Larger Context Window Could Reach 1.5 Million Tokens

Another major rumor centers on a substantial context window upgrade. Reports suggest GPT-5.6 could expand the maximum context length from 1 million tokens to 1.5 million tokens.

If accurate, the increase would allow the model to process larger codebases, longer conversations, extensive research documents, and more complex workflows within a single session. Developers could benefit from improved memory and continuity when working on large projects.

Stronger Coding and Agent Capabilities

Coding performance appears to be one of the primary areas of improvement. Early testers claim GPT-5.6 delivers better code understanding, stronger planning abilities, and more reliable execution across multi-step development tasks.

Some reports suggest the model performs particularly well in agent-based coding workflows and software development benchmarks. Testers also highlighted improvements in SVG generation, 3D design creation, and robotic simulation tasks.

According to several users, GPT-5.6 Pro may outperform some competing models in coding-related benchmarks, although results vary depending on the specific test.

Higher Quality Comes With Slower Responses

Not all feedback has focused on speed. Some testers reported that GPT-5.6 takes considerably longer to complete demanding tasks.

In one example, developer Conor Dart used the model to create a browser-based 3D game with physics simulation and camera controls. The generation process reportedly took more than an hour but produced significantly better results than previous versions.

Benchmarker Chris compared GPT-5.6 Pro against GPT-5.5 Extra High and found that while GPT-5.6 generated higher-quality outputs, it required substantially more time to finish the task. This has led many testers to conclude that OpenAI may be prioritizing reasoning quality over raw speed.

Competition in the AI Race Continues

The rumored release comes at a time of intense competition in the AI industry. Chinese AI models such as GLM-5.2 and Anthropic’s latest systems are pushing the boundaries of coding and agent performance.

Reports indicate OpenAI may respond not only with stronger models but also with more aggressive pricing. Some sources claim OpenAI’s current token costs are already significantly lower than Anthropic’s, and further reductions could increase pressure across the AI market.

Industry observers believe a combination of lower costs and improved performance could help OpenAI strengthen its position among developers and enterprise customers.

Possible Launch Date Emerges

Several rumors point to June 25 as a possible launch date. Some testers believe OpenAI is currently conducting stealth A/B testing through GPT-5.5 Pro, allowing the company to gather real-world feedback before a wider rollout.

Prediction market traders have also shown growing confidence in a release during the week of June 22 to June 28, with some forecasts assigning high probabilities to a launch within that period.

OpenAI Has Not Confirmed the Reports

Despite the growing excitement, OpenAI has not officially confirmed the existence of GPT-5.6, its specifications, pricing, or launch schedule.

Most of the current information comes from social media reports, alleged leaks, early tester feedback, and industry speculation. Until OpenAI makes an official announcement, all details regarding GPT-5.6 should be treated as unconfirmed.

If the rumors prove accurate, GPT-5.6 could represent a significant step forward for OpenAI’s AI platform. Early reports point to stronger reasoning, better coding performance, larger context windows, and improved agent capabilities. While the model may take longer to generate responses, testers suggest the trade-off could be worth it for users seeking higher-quality results. For now, the AI community is waiting to see whether OpenAI officially unveils GPT-5.6 in the coming days.

Read More:

• Lenovo Aurora GH15 Headphone: RGB Lighting, Detachable Mic, and 1000mAh Battery
• ASUS Just Launched a Mouse Pad Built by Pro Esports Players
• ASUS Launches ROG Harpe II Extreme Edition 20 Gaming Mouse With 24K Gold-Plated Design

(via)

The post GPT-5.6 Rumors Intensify as OpenAI Reportedly Prepares Major AI Upgrade appeared first on Gizmochina.

HONOR has offered a glimpse into the future of smartphone filmmaking by unveiling the first professional cinematic video captured using its upcoming HONOR Robot Phone. The showcase took place at the 28th Shanghai International Film Festival (SIFF), where HONOR is serving as the official mobile photography and videography partner. The project highlights the company’s ambition to bring professional filmmaking tools and workflows to smartphone users.

First Cinematic Video Debuts at SIFF

As part of a special collaboration with ELLEMEN, HONOR used the Robot Phone to create cinematic video portraits of SIFF jury members. The footage demonstrates how the device can deliver a film-like visual experience with rich tones, smooth color transitions, and a strong sense of atmosphere. The project serves as a showcase of how smartphone cameras are evolving beyond traditional mobile photography into powerful storytelling tools.

Built-In Gimbal Delivers Cinema-Like Camera Movement

One of the Robot Phone’s standout features is its integrated physical gimbal stabilization system. Unlike conventional smartphones that often rely on electronic stabilization and image cropping, the Robot Phone uses mechanical stabilization to produce smooth handheld tracking shots while preserving full image quality. This allows users to achieve professional camera movements without sacrificing detail or field of view.

Industry’s Smallest Titanium Alloy Gimbal

HONOR says the Robot Phone is equipped with the industry’s smallest titanium alloy gimbal. Designed for precision and stability, the system is powered by high-performance motors that allow dynamic movement beyond the limits of traditional smartphone camera modules. The result is smoother footage, greater flexibility, and improved stability during shooting.

AI Helps Simplify Professional Video Creation

The Robot Phone combines its hardware innovations with advanced AI-powered imaging features. Intelligent subject tracking keeps moving subjects in focus while maintaining stable footage. HONOR says these AI capabilities are designed to simplify the filmmaking process, making it easier for creators to produce professional-looking videos without complex equipment setups.

Bringing ARRI Technology to Smartphones

The Robot Phone is also the first HONOR device to benefit from the company’s strategic partnership with ARRI, a leading name in professional cinema cameras. The smartphone incorporates ARRI’s LogC expertise at the RAW level, giving creators greater flexibility during editing and color grading. It also supports professional workflows through compatibility with ARRI’s LUT ecosystem in DaVinci Resolve, helping users achieve authentic cinematic color styles.

HONOR’s Vision for Mobile Filmmaking

With the Robot Phone, HONOR aims to blur the line between smartphones and professional filmmaking equipment. By combining AI-powered imaging, advanced stabilization technology, and ARRI-inspired cinema tools, the company hopes to make high-quality visual storytelling more accessible. The SIFF debut provides an early look at how HONOR plans to bring professional filmmaking standards to the next generation of content creators and mobile videographers.

The post HONOR Robot Phone Unveils First Cinematic Video at Shanghai International Film Festival appeared first on Gizmochina.

Enterprise AI agents are getting really good at handling complex, multi-step work on their own, things like sorting through emails, prepping for meetings, or pulling together reports across Outlook, Teams, and Excel. But all that smart chaining of tasks comes with a catch: it burns through a ton of computing power, and the bills can add up fast. Microsoft is now stepping in to make this more practical for everyday businesses.

According to reports, Microsoft has made Copilot Cowork generally available to Microsoft 365 Copilot customers and shifted it to usage-based pricing. Instead of a straight flat fee that didn’t always match heavy usage, companies will now pay based on actual “Copilot Credits”, basically metering the real computational load for each task.

On top of the pricing shift, Microsoft is looking at ways to bring costs down even further. They’re exploring a fine-tuned, self-hosted version of DeepSeek V4 (or another open-source model) as a more affordable option alongside the current Anthropic and OpenAI models. DeepSeek V4 has been delivering strong performance at much lower prices, sometimes a fraction of what frontier models cost per token.

Importantly, any DeepSeek integration would be optional and run entirely on Azure. That means customer data stays inside Microsoft’s secure cloud environment with all the usual enterprise-grade protections, compliance, and data residency controls. It’s a smart way to address potential concerns about using a Chinese-developed model in sensitive business settings.

This whole move shows how the AI industry is maturing. Agentic AI is incredibly useful because it can keep working through long chains of tasks and maintain context, but that same power makes it expensive to run at scale. By offering model choices and metered billing, Microsoft is trying to strike a better balance so more businesses can actually adopt these tools without sticker shock.

It’ll be interesting to see in the coming weeks which lower-cost model they land on and how customers respond. Overall, this feels like a practical step toward making advanced AI agents part of normal work life instead of just a high-end experiment. Cost efficiency and flexibility could be the real keys to widespread adoption.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Via | Image)

The post Microsoft exploring use of DeepSeek V4 as Anthropic and Claude models are too expensive appeared first on Gizmochina.

In a recent essay posted on X, Microsoft’s CEO Satya Nadella raised some serious concerns about the long-term impact of advanced AI on businesses. His main worry? Powerful AI models are getting so good at absorbing specialized corporate knowledge that the very companies creating this expertise could end up at a big disadvantage.

Nadella cautions that without careful steps, a small handful of dominant AI companies could end up capturing most of the economic value. He puts it bluntly: no one wants “a world where every company across every sector is ceding value to a few models that eat everything they see.” He also stresses there’s “no societal permission for an AI future that hollows out entire industries.”

He draws a parallel to the early days of globalization, when outsourcing boosted overall GDP numbers but also eroded industrial bases, jobs, and local expertise in many places, effects we’re still dealing with today. Nadella doesn’t want AI to repeat that same mistake, where value gets pulled away from traditional companies and funneled toward just a few tech giants.

Instead, he’s pushing for a more open and decentralized AI world. Companies need to keep control of their own “learning systems” and build internal AI capabilities. He talks about combining “human capital” (judgment, relationships, and creativity) with “token capital” (proprietary AI trained on their own data). That way, businesses can protect their competitive edge and keep driving real innovation.

This isn’t just Nadella’s view. Other leaders like Snowflake CEO Sridhar Ramaswamy and Box CEO Aaron Levy have made similar points, warning that big companies could become little more than data suppliers for powerful external AI models, losing their ability to stand out.

Nadella’s message is clear: we need to develop it in a way that spreads value more fairly instead of concentrating it in just a few hands. That balance will be key to making sure AI strengthens companies and society rather than weakening the expertise that makes them successful in the first place.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source | Image)

The post Microsoft CEO warns advanced AI may concentrate wealth and expertise in a handful of companies appeared first on Gizmochina.

A pretty exciting breakthrough has surfaced for Apple Silicon. A developer has figured out how to bypass restrictions on the M4’s Neural Engine (ANE) to enable on-device AI model training, something Apple has kept locked down so far.

Normally, the Neural Engine in M4 chips is mainly used for running (inference) pre-trained models efficiently. But security researcher and developer @0x0SojalSec announced on X that they successfully reverse-engineered the limitations. They’ve managed to unlock up to 15.8 TFLOPS of compute power for full training workloads, including backpropagation on transformer models.

Instead of relying on Apple’s usual Core ML or Metal frameworks, the team built a custom Model Intermediate Language (MIL) to talk directly to the Neural Engine. They keep everything in RAM to avoid slow storage writes, which makes the process surprisingly fast and stable. They even found a workaround using the exec() command to restart the process when training gets stuck, allowing it to checkpoint and keep going.

The developer has shared the project on GitHub, showing that it’s possible to train neural networks directly on the ANE hardware with minimal extra dependencies. Early tests show impressive results, like completing transformer training steps in just milliseconds on M4 chips.

Apple hasn’t officially enabled or supported training on the Neural Engine (they market it as delivering up to 38 TOPS for inference only). This reverse-engineering effort shows there’s a lot more potential hidden in these chips than most people realized.

This is big news for the AI community because it could turn regular Macs and iPads into much more capable local training machines for smaller models, reducing the need to rely on cloud services. It also sparks interesting discussions about Apple’s tightly controlled hardware-software approach.

It’ll be fascinating to see what else the community discovers as more people experiment with this. For now, it’s a promising step toward more powerful on-device AI experimentation in the Apple ecosystem.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: @0x0SojalSec on X)

The post Developer finds workaround to train AI locally on MacBooks, bypassing Apple’s software restrictions appeared first on Gizmochina.

Asus is pushing the boundaries with the new ExpertCenter Pro ET900N G3, a regular-looking tower PC that packs Nvidia’s powerful GB300 Grace Blackwell Ultra desktop superchip inside. It’s designed for local AI development, model training, and inference, giving you data-center-level performance without needing an actual server room.

At first glance, it looks like a standard desktop tower (about 23.2 x 58.4 x 56.5 cm), something you could actually fit on a typical desk. But inside, it’s a beast. The system combines a Grace CPU with 72 Arm cores and a Blackwell Ultra GPU, delivering up to 20 petaFLOPS of AI performance.

One of the most impressive parts is the memory: it has a massive 748 GB of unified coherent memory (496 GB LPDDR5X for the CPU + 252 GB HBM3e for the GPU). This setup lets you work with really large AI models, even those approaching trillion-parameter scale, right on your desk.

It also comes with a hefty 1,600W power supply and strong cooling, so it can run at full power for long periods without slowing down. For expansion, you get three PCIe 5.0 slots, support for up to four M.2 2280 NVMe SSDs, and high-speed NVLink-C2C and ConnectX-8 networking. It ships with Ubuntu and Nvidia’s AI tools pre-loaded, with Windows support coming later. On the outside, the ports are practical, with eight USB 3.2 ports, one USB 2.0 port, and dual 10 Gbps Ethernet.

The ExpertCenter Pro ET900N G3 is now available in the US, starting at a hefty $99,999. It’s clearly aimed at researchers, developers, and companies working on generative AI, large language models, simulations, and autonomous systems. It’s basically bringing the spirit of Nvidia’s DGX Station to a more compact form.

This is a clear sign of the trend toward making high-end AI hardware more accessible at the desktop level. Yes, the price is steep, but for teams that need serious local compute power without the hassle of full server setups, this could be a game-changer.

What do you think? Crazy impressive or just too expensive for most people?

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(ASUS)

The post ASUS unveils desktop PC with 748 GB RAM, NVIDIA’s GB300 Grace Blackwell Ultra superchip appeared first on Gizmochina.

TECNO has officially introduced the new POVA 8 5G, bringing a strong focus on battery life, design, gaming performance, and AI-powered tools. The smartphone made its global debut on June 11, 2026, with India becoming the first market to receive the device. Availability in other regions will be announced in the coming months.

Interstellar-Inspired Design With Alive Matrix Display

The TECNO POVA 8 5G continues the brand’s futuristic design language with an Interstellar Spaceship-inspired look. The phone features a geometric rear camera module and a semi-transparent back panel that gives it a modern appearance.

One of the biggest highlights is the new Alive Matrix Display on the rear panel. This customizable LED lighting system can react to incoming calls, notifications, music playback, gaming events, charging status, and other alerts. TECNO says users can choose from 49 preset lighting effects or create their own custom animations.

Massive 8,000mAh Battery Promises Two-Day Usage

Battery life is a major focus of the POVA 8 5G. The device packs a large 8,000mAh battery that is TÜV SÜD certified for up to two days of regular use. TECNO claims the phone can deliver more than 60 hours of calling, 85 hours of music playback, 29 hours of YouTube streaming, and over 14 hours of gaming on a single charge.

The company also says the battery is designed for long-term durability, retaining over 80% of its health after 2,000 charge cycles and offering up to six years of reliable use.

50MP Sony Camera System

For photography, the TECNO POVA 8 5G features a 50MP Sony LYTIA 600 main camera that is co-engineered with Sony for improved light capture and image quality. The camera also supports 2x lossless zoom, allowing users to get closer to subjects without a noticeable drop in detail. On the front, the device includes a 13MP selfie camera for video calls and social media content. TECNO further enhances the camera experience with AI LightMaster 2.0, which can reduce lens flares, reflections, and unwanted shadows in photos.

Dimensity 7100 Power and AI Features

Powering the smartphone is the MediaTek Dimensity 7100 5G chipset. TECNO has paired it with its G1 Signal Enhancement Chip and SE1 Wi-Fi Enhancement Chip to improve connectivity in weak-signal environments. The company claims the setup can deliver stronger cellular and Wi-Fi reception in places such as elevators, basements, and crowded areas.

The phone also includes a 144Hz eye-care display for smoother scrolling and gaming.

Software Support and Extra Benefits

TECNO has equipped the POVA 8 5G with several AI tools, including AI YouTube Summary, AI Writing, AI Theme Generator, AI LightMaster 2.0, and All-Scenario Noise Cancellation. The company is also promising two major Android OS upgrades and three years of security updates.

Additionally, buyers will receive 256GB of free cloud storage for three years, while selected markets will also get a three-month Google AI Plus trial with 2TB cloud storage. With its large battery, unique rear lighting system, and practical AI features, the POVA 8 5G aims to stand out in the competitive mid-range smartphone segment.

The post TECNO POVA 8 5G Launches With Massive 8,000mAh Battery and Alive Matrix Display appeared first on Gizmochina.

Samsung is launching a company-wide AI initiative aimed at integrating artificial intelligence into daily operations across its businesses.

Announced on June 9, 2026, the program, called “AI Transformation” (AX), will introduce new AI tools, employee training programs, and dedicated AI teams throughout the Samsung Group. The effort is designed to increase productivity and encourage wider use of AI across the company’s operations.

As part of the rollout, Samsung plans to give employees access to external generative AI services, including Google Gemini, ChatGPT, and Claude. The tools will be available across subsidiaries and are expected to support tasks ranging from software development and marketing to manufacturing and business operations.

Training is another major part of the initiative. Samsung aims to provide AI education to its entire workforce of roughly 280,000 employees by the end of 2026.

The company is also starting from the top. Around 50 group presidents will attend a two-day AX Bootcamp at Samsung’s Human Resources Development Institute this month. Senior executives, numbering about 2,300, will take part in separate three-day training sessions, with the first round scheduled to finish by mid-August.

Samsung is expected to unveil a group-wide AX Vision during the leadership training program, outlining how AI will be applied across its businesses. The company also plans to establish dedicated AI departments within each subsidiary to oversee adoption and develop internal expertise.

The move reflects the growing importance of AI across the technology industry, where companies are increasingly looking to automate workflows and improve decision-making. This is one of Samsung’s largest efforts yet to expand AI adoption beyond products and into the way the company itself operates.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Chosunbiz)

The post Samsung launches company-wide “AI Transformation” initiative to increase productivity appeared first on Gizmochina.

NVIDIA’s new RTX Spark platform was one of the biggest highlights at Computex Taipei 2026, with major PC brands including ASUS, Dell, Lenovo, and MSI unveiling compact desktop systems powered by the new AI-focused architecture. The launch marks NVIDIA’s first major step into the PC processor market and could reshape the future of AI PCs, creator workstations, and compact desktops.

NVIDIA RTX Spark Platform Takes Center Stage

The RTX Spark platform is an Arm-based system-on-chip (SoC) built around NVIDIA’s Grace and Blackwell technologies. It combines a 20-core CPU with a Blackwell GPU featuring 48 SM units and 6,144 CUDA cores. The platform also supports up to 128GB of LPDDR5X unified memory.

NVIDIA says the chip is designed for local AI development, content creation, gaming, and AI inference workloads. The company also claims graphics performance comparable to an RTX 5070-class GPU in certain scenarios.

ASUS, Dell, Lenovo, and MSI Showcase New Mini PCs

Several manufacturers introduced RTX Spark-powered mini PCs during the event. While each system features a unique chassis design, they share the same core platform and connectivity options.

ASUS ProArt GA10

The ASUS ProArt GA10 features a compact 150 × 150 × 51 mm design and supports up to 140W of cooling capacity. It includes four USB-C ports, HDMI output, and 10 Gigabit Ethernet connectivity.

Dell XPS RTX Spark Desktop

Dell’s new compact desktop also adopts the RTX Spark platform. The system includes four USB-C ports, HDMI output, and 10GbE networking while maintaining a small footprint.

Lenovo RTX Spark Mini PC

Lenovo introduced a newly designed mini PC chassis built around RTX Spark. It offers the same set of connectivity options, including four USB-C ports, HDMI, and 10GbE Ethernet.

MSI EdgeMesa N AI+

MSI’s EdgeMesa N AI+ stands out with its white chassis design. The system targets content creators and AI developers while offering the same RTX Spark hardware foundation.

Shared Features Across All Models

Every RTX Spark mini PC showcased at Computex includes 10GbE networking, 20Gbps USB-C connectivity, PCIe Gen 5 M.2 SSD support, and up to 128GB unified memory.

Why RTX Spark Could Be a Game-Changer for PC Buyers

For consumers, RTX Spark could bring desktop-class AI performance, powerful graphics, and massive memory capacities into much smaller and more efficient PCs. Because the CPU, GPU, and up to 128GB of unified memory work together on a single platform, users may be able to run advanced AI models, edit high-resolution videos, and handle demanding creative workloads locally without relying heavily on cloud services.

If NVIDIA’s performance claims hold up, RTX Spark systems could also offer a compelling alternative to traditional Intel and AMD-powered PCs, increasing competition and giving buyers more choice. 

The trade-off is that some older Windows applications and specialized software may face compatibility or optimization challenges, although Windows on Arm support has improved significantly in recent years.

Read More:

• US- China Tech War: China Just Beat NVIDIA in the Next Big AI Race
• Why Nvidia’s H2+ Humanoid Robot Has the Tech World Talking
• US Expands AI Chip Restrictions Worldwide: What It Means for China

(via)

The post NVIDIA RTX Spark Mini PCs Arrive at Computex 2026 — Could They Replace Traditional Desktops? appeared first on Gizmochina.

The battle for AI leadership is rapidly moving from chatbots to robots, and China has just scored a major victory. Chinese startup Spirit AI announced that its Spirit v1.6 model ranked No.1 on the RoboArena benchmark, outperforming NVIDIA and other leading competitors. 

Spirit v1.6 achieved a score of 1,924, outperforming NVIDIA’s Cosmos3-Nano-Policy, which scored 1,881, and DreamZero, another NVIDIA-backed model, which scored 1,763. The achievement marks the first time a Chinese embodied AI model has reached the No.1 spot on the benchmark.

Why RoboArena Matters

RoboArena is designed to measure how effectively AI systems can translate decisions into real-world robotic actions. Unlike traditional AI benchmarks that focus on language or reasoning, RoboArena evaluates physical task execution, including object manipulation, navigation, tool usage, perception, planning, and adaptability in unfamiliar environments.

The benchmark is widely respected because it uses randomized environments, adversarial testing scenarios, and strict anti-overfitting measures, making it difficult for companies to optimize specifically for the test.

The Growing Physical AI Race

The result highlights how competition in artificial intelligence is increasingly shifting beyond chatbots and large language models toward physical AI. These systems serve as the “brains” behind humanoid robots, robotic arms, autonomous vehicles, and industrial automation platforms.

Spirit AI’s success is being viewed as evidence that efficient model design and high-quality training data can be just as important as massive computing power. The company’s model reportedly demonstrated strong performance in object handling, autonomous navigation, complex tool use, and adapting to new situations with minimal fine-tuning.

China’s Expanding Advantage

China has also been gaining ground across several other embodied AI benchmarks. Companies including Manifold AI, AgiBot, and DexForce currently lead categories related to world modeling, robotic perception, and robotics data generation.

At the same time, investment in the sector continues to surge. Spirit AI recently raised 1.5 billion yuan (about $222 million), while other Chinese robotics startups have secured large funding rounds over the past year.  China’s growing strength in robotics data, funding, and benchmark performance suggests the global race is becoming more competitive than ever.

Read More:

• ASUS Launches ROG Harpe II Extreme Edition 20 Gaming Mouse With 24K Gold-Plated Design
• Why Nvidia’s H2+ Humanoid Robot Has the Tech World Talking
• US Expands AI Chip Restrictions Worldwide: What It Means for China

(via)

The post US- China Tech War: China Just Beat NVIDIA in the Next Big AI Race appeared first on Gizmochina.

Nvidia has announced a major step forward in humanoid robotics with the launch of the H2+ reference robot platform, developed in partnership with China’s Unitree Robotics and Singapore-based Sharpa. The announcement was made by Nvidia CEO Jensen Huang during his Computex 2026 keynote in Taipei.

The project is attracting industry attention because it combines advanced AI, robotics hardware, and software into a unified platform designed to speed up the development of real-world humanoid robots. 

Rather than being a finished commercial robot, H2+ is a reference design that researchers, universities, and robotics companies can use as a blueprint to build and customize their own humanoid robots. Nvidia says the platform is designed to simplify every stage of robot development, from data collection and AI training to real-world deployment.

Combining Hardware and AI

The H2+ platform brings together several technologies into a single package. It uses Unitree’s H2 Plus humanoid robot body, Sharpa’s Wave robotic hands, and Nvidia’s Isaac GR00T AI models. Powering the system is Nvidia’s new Jetson AGX Thor T5000 computer, which is based on the Blackwell GPU architecture.

The robot stands nearly six feet tall and offers a total of 75 degrees of freedom, including highly dexterous robotic hands with 22 active joints. Nvidia says the Jetson AGX Thor T5000 delivers up to 2,070 FP4 TOPS of AI performance and includes 128GB of memory, allowing the robot to process complex tasks in real time.

Nvidia Introduces Cosmos 3

Alongside H2+, Nvidia also introduced Cosmos 3, a new physical AI world model designed to help robots better understand and learn from their environments.

According to Huang, one of the biggest challenges in robotics is collecting enough useful training data. Cosmos 3 can understand both third-person and first-person perspectives, helping generate robot-centric training data that can improve learning and decision-making.

Expansion Beyond China

Nvidia also revealed plans to expand its humanoid robot partnerships beyond China. The company is looking to work with robot makers in the United States, Europe, and South Korea, although partner names have not yet been announced.

The goal is to create standardized robot platforms with built-in security features such as Secure Boot and Confidential Computing, helping ensure that only verified software can run on the machines.

Growing Humanoid Robot Market

The announcement comes as interest in humanoid robots continues to grow. Analysts at Morgan Stanley expect China’s humanoid robot sales to exceed 28,000 units in 2026, potentially making it the world’s largest market for the technology.

With H2+, Cosmos 3, and new international partnerships, Nvidia is positioning itself at the center of the rapidly expanding humanoid robotics industry, supplying the AI hardware, software, and development tools that could power the next generation of intelligent machines.

Read More:

• US Expands AI Chip Restrictions Worldwide: What It Means for China
• Xiaomi is repairing its rainstorm-affected phones, laptops, and accessories for free in China
• China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift

(via)

The post Why Nvidia’s H2+ Humanoid Robot Has the Tech World Talking appeared first on Gizmochina.

The US government has given a new clarification that export restrictions on advanced AI chips apply not only to companies operating in China but also to Chinese-owned firms and subsidiaries located overseas. The move closes a potential loophole that may have allowed Chinese companies to access restricted AI hardware through operations in countries outside mainland China.

Why the US Made This Move

The clarification follows concerns in Washington that Chinese firms could obtain advanced AI processors through overseas subsidiaries despite existing export controls. Some policymakers argued that uncertainty created after the cancellation of the Biden-era AI Diffusion Framework may have opened a pathway for such purchases. The new guidance makes it clear that Chinese-headquartered companies remain subject to licensing requirements regardless of where they operate.

Pressure Grows on China’s AI Industry

The tighter enforcement is expected to make access to advanced US AI hardware more difficult for Chinese companies. NVIDIA’s most powerful Blackwell AI chips remain restricted, while purchases of other advanced processors continue to face regulatory scrutiny. As a result, Chinese technology firms are being pushed to rely more heavily on domestic alternatives.

China Accelerates Its Chip Strategy

In response to years of US restrictions, China has been rapidly expanding its domestic AI semiconductor industry. Companies including Huawei, Moore Threads, Biren Technology, Cambricon, and Alibaba have increased investments in AI processors designed for both training and deploying large AI models. Rather than focusing solely on replacing Nvidia products, many firms are now building complete AI hardware and software ecosystems.

Domestic Alternatives Gain Importance

Industry analysts expect Chinese AI companies to increasingly adopt locally developed accelerators as access to foreign chips becomes more limited. Huawei’s Ascend series and Alibaba’s custom AI processors are among the products attracting attention as China works to reduce dependence on US technology.

What to Expect Next

The latest US action is unlikely to slow China’s AI ambitions. Instead, it may further accelerate investment in domestic chips, software platforms, and AI infrastructure. The competition between the US and China is increasingly shifting from individual products to entire technology ecosystems. In the coming years, the key question may no longer be whether China can replace Nvidia, but how quickly it can build a self-sufficient AI industry capable of competing on a global scale.

Read More:

• China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift
• Huawei aims to launch lithography equivalent to TSMC’s 1.4nm by 2031 in a major defiance of U.S. sanctions
• OnePlus Ace 6 Ultra vs OnePlus Ace 6: Are the Upgrades Really Worth the Extra Money?

The post US Expands AI Chip Restrictions Worldwide: What It Means for China appeared first on Gizmochina.

CHUWI has officially introduced the new CoreBook Air, a next-generation Copilot+ PC designed to deliver powerful AI capabilities, long battery life, and premium portability. According to CHUWI CEO Jesea Yang, the CoreBook Air combines Lunar Lake performance and Copilot+ AI capabilities in an ultra-light 1.2kg design, helping bring advanced AI PCs to a wider audience at a more accessible price point. The company aims to make advanced AI-powered computing more accessible to students, professionals, creators, and remote workers.

Premium Design Built for Portability

The CHUWI CoreBook Air features a slim and lightweight design that weighs just 1.2kg, making it easy to carry throughout the day. The laptop uses a premium CNC-crafted aluminum chassis that provides a durable and professional appearance while maintaining portability. Its compact form factor makes it suitable for students attending classes, content creators working on the move, business professionals, and remote workers who need a dependable travel companion.

Sharp 2.8K Display with Smooth Visuals

The laptop comes equipped with a 14-inch IPS display featuring a 2.8K resolution of 2880 × 1800 pixels. It also supports a 120Hz refresh rate and covers 100% of the sRGB color gamut. This combination delivers sharper text, detailed visuals, smoother scrolling, and an improved video viewing experience. The high-resolution panel is also designed to provide comfortable viewing during long work sessions.

Intel Lunar Lake Performance with Advanced AI

At the heart of the CoreBook Air is the Intel Core Ultra 5 226V processor based on Intel’s Lunar Lake (Series 2) architecture. The chip features 8 CPU cores and 8 threads, with boost frequencies reaching up to 4.5GHz. Users can expect strong computing performance, improved power efficiency, broad software compatibility, enhanced security, and advanced AI processing capabilities.

Graphics duties are handled by the integrated Intel Arc 130V GPU, which CHUWI claims can deliver up to twice the graphics performance of older Intel Iris Xe solutions. The GPU supports hardware-accelerated ray tracing, casual gaming, content creation tasks, multimedia streaming, and GPU-accelerated applications.

Copilot+ Features and AI Acceleration

One of the biggest highlights of the CoreBook Air is its AI performance. The system delivers up to 97 TOPS of total AI computing power and includes an Intel AI Boost NPU capable of up to 40 NPU TOPS. This dedicated AI hardware enables faster on-device AI processing, reduced dependence on cloud services, improved privacy, and quicker AI responses.

Running Windows 11 Pro out of the box, the laptop supports Microsoft Copilot+ PC experiences such as Click to Do, Cocreator, and Live Captions. A dedicated AI key also provides instant access to Microsoft Copilot.

Memory, Storage, Battery, and Connectivity

The CoreBook Air includes 16GB LPDDR5X memory running at 8533 MT/s using a memory-on-package design for faster multitasking, lower latency, and better efficiency. Storage is handled by a 512GB PCIe SSD, while users can upgrade storage later through PCIe 4.0 x4 SSD support.

A 55Wh battery powers the laptop and is rated for up to 15 hours of local video playback. Charging is handled by a 65W USB-C Power Delivery adapter, capable of charging the battery to around 63% in approximately one hour.

For connectivity, the laptop supports Wi-Fi 6E for faster wireless speeds, lower latency, and improved stability. Port selection includes two Thunderbolt 4 ports with up to 40Gbps bandwidth, Power Delivery support, external display connectivity, and eGPU compatibility. Users also get three USB-A ports, an HDMI 2.0 output, and a 3.5mm audio jack.

Pricing and Availability

The CHUWI CoreBook Air is available for pre-order now. The laptop carries an official retail price of $799, but early buyers can take advantage of a limited-time Super Early-Bird discount of $50, reducing the launch price to $749. The promotional offer is available on a first-come, first-served basis while launch stock lasts.

CHUWI CoreBook Air – Specs Sheet

Display

Performance

Memory & Storage

Battery & Charging

AI Features

Ports & Connectivity

Wireless Connectivity

Camera & Multimedia

The post CHUWI CoreBook Air Launches as a Copilot+ PC with Intel Lunar Lake, 2.8K Display and 15-Hour Battery appeared first on Gizmochina.

Xiaomi has introduced its latest open-ear audio wearable, the Xiaomi Clip Earbuds, combining lightweight comfort, powerful sound, AI real-time translation, and long battery life in a stylish clip-on design. The new earbuds are now available on Giztop for a starting price of $159.99 with global shipping support. 

Lightweight and Comfortable Design

The Xiaomi Clip Earbuds are designed for users who want comfort during long listening sessions. The earbuds feature a lightweight clip-on design, with each earbud weighing around 5.5 grams. The open-ear style allows users to stay aware of their surroundings while listening to music or taking calls. This design makes them suitable for workouts, office use, travel, and daily outdoor activities.

The earbuds also come with an IP57 dust and water resistance rating, offering protection against sweat and light rain. Xiaomi has included touch controls on the earbuds for easy music playback, call handling, and other functions. MIUI users also get a pop-up window for quick connection and battery status viewing.

Powerful Audio and Clear Calling Experience

Xiaomi Clip Earbuds are equipped with 11mm dynamic drivers paired with microcrystalline metal-coated diaphragms. The earbuds deliver balanced sound with clear vocals and deep bass. They support a wide 20Hz to 48kHz frequency response for detailed audio performance.

For wireless audio, the earbuds support AAC, SBC, and LHDC 5.0 codecs. They also carry Hi-Res Audio Wireless certification, helping deliver high-quality sound during music playback.

Call quality is another major highlight. Xiaomi uses a three-microphone setup along with a VPU bone-conduction microphone to reduce surrounding noise and isolate the user’s voice. This helps provide clearer conversations even in crowded or noisy environments.

Long Battery Life and Smart Features

The Xiaomi Clip Earbuds offer up to 9 hours of music playback on a single charge. With the charging case, total battery life can reach up to 38 hours. The USB-C charging case also supports fast charging, and a quick 5-minute charge can provide around 2 hours of playback.

Another interesting feature is AI real-time translation support, making the earbuds more useful for travel and communication.

With premium audio features, clear call quality, smart AI tools, and a comfortable open-ear design, the Xiaomi Clip Earbuds offer a modern listening experience for everyday users. 

Read More:

• Xiaomi 17 Max Now Available on Giztop With 8000mAh Battery and 200MP Leica Camera
• Lenovo Legion Y900 13-Inch 2026 Now Available at Giztop Starting at $499
• Lenovo’s New Legion Y700 2026 Gaming Tablet Gets Massive $200 Discount at Giztop

The post Xiaomi Clip Earbuds Now Available on Giztop for $159.99 – Open-Ear Design Meets AI Translation appeared first on Gizmochina.

China has officially launched its first general-purpose household robot, called “Shiguang S1,” in Wuhan’s Optics Valley. The robot was introduced on May 20 by Chinese robotics company GigaBrain and is designed specifically for daily home use. Unlike traditional robots built for factories, the Shiguang S1 is made to assist families with regular household tasks and interact naturally with people inside the home.

Designed to Handle Daily Household Work

The Shiguang S1 can perform several common household activities. According to the company, the robot is capable of folding clothes, cooking meals, and clearing the dining table. It can also chat with elderly people and provide assistance at home. One of its biggest features is its ability to continuously learn and adapt to new tasks, making it more flexible than older task-specific robots.

Powered by Advanced AI Technology

GigaBrain says the robot uses its self-developed embodied intelligent model, which allows it to understand tasks and react to different home environments. Ye Yun, Vice President of R&D at GigaBrain, explained that traditional factory robots usually follow fixed instructions, while the Shiguang S1 can plan actions on its own and respond intelligently to changing situations.

The company also revealed plans to officially launch its next-generation “GigaBrain 1” model in the third quarter of 2026.

Safety and Early User Testing

To improve safety, the robot includes a compliant control mechanism. If it accidentally touches a person or pet while moving, it immediately stops to reduce the risk of injury.

Families in Wuhan can now apply to become early “seed users” through the “Shiguang Vision” WeChat account and the company’s official application page. The launch highlights China’s growing progress in AI-powered robotics, elderly care technology, and smart home automation.

Read More:

• Motorola Edge 70 Pro vs Redmi Note 15 Pro+: Which Mid-Range Phone Is Better
• OPPO Find X9 Ultra vs Honor Magic8 Pro Comparison: Camera, Battery & Performance
• Motorola Razr Ultra 2026 vs Honor Magic V5: One Clearly Offers More Value

(via)

The post China Launches First Universal Household Robot “Shiguang S1” in Wuhan appeared first on Gizmochina.

Kynooe has officially launched its new AI-powered robotic arm on Kickstarter, introducing a fully modular robotic arm designed for everyday users rather than just engineers or robotics experts.

Built for home assistance, smart automation, content creation, and creative workflows, Kynooe combines modular hardware, AI-powered interaction, and app-based controls into a more consumer-friendly robotics platform. The project focuses on making robotics more accessible through a no-code experience and a flexible modular ecosystem.

Beginner-Friendly Setup Without Coding

One of the key highlights of Kynooe is its beginner-focused setup process. The system uses a plug-and-play modular design and can be controlled through a companion app with an interface designed to feel as intuitive as using a smartphone.

Instead of relying on traditional robotics programming or complex engineering setups, users can control and customize the robot through visual interface tools, AI-assisted interactions, and simplified workflows. The no-code approach is designed to lower the barrier to entry for users who are new to robotics.

Fully Modular Design With Expandable Accessories

Unlike traditional robotic arms built for fixed functions, Kynooe features a fully modular architecture that supports interchangeable tools, accessories, and attachments. The platform is compatible with precision instruments, custom modules, and even 3D-printed extensions for specialized use cases.

The company sums up the concept with the tagline “One arm. Infinite applications,” emphasizing the system’s adaptability across diverse workflows and environments. Depending on the configuration, the robot can shift seamlessly between creative assistance, intelligent filming, automation tasks, and productivity-focused applications.

AI Features and Natural Interaction

Kynooe also integrates AI-powered interaction features aimed at making the experience feel more conversational and intuitive. Users can interact with the robot through natural language commands instead of relying entirely on manual controls or traditional programming methods.

One of the major built-in features is on-device AI face tracking. The feature is aimed at content creators and smart filming setups. The robotic arm can automatically track faces and adjust camera movement for smoother recording and hands-free filming support. The company also plans to introduce additional on-device AI features in the future, aiming to improve user experience while maintaining user privacy.

Built for Everyday Scenarios

Rather than focusing solely on industrial or engineering applications, Kynooe is positioned as a robotics platform for real-world, everyday use cases. The system is designed for household assistance, entertainment, smart filming, creative projects, and interactive experiences with family and friends.

The robotic arm adapts across different scenarios depending on the selected modules and accessories, allowing users to customize the system around their own workflows and creative needs.

Open Ecosystem and Kynooe Robot Hub

Beyond hardware, Kynooe is also building an ecosystem focused on embodied AI and community-driven robotics development. The company plans to release an open-source SDK on GitHub that will provide developers and advanced users with access to everything from AI integrations to low-level robotic joint controls.

The ecosystem also includes the Kynooe Robot Hub, a platform where users can train, customize, publish, and share robotic behaviors and workflows with the broader community. The company envisions a future where robotic skills can be shared, downloaded, and deployed as easily as mobile apps.

Development Progress and Manufacturing Timeline

According to the company, Kynooe has already completed structural design work, hardware validation, and small-batch trial production. The project is currently moving toward certification, production validation testing, and mass production preparation ahead of planned shipments later in 2026.

The current roadmap includes ongoing tooling optimization, manufacturing preparation, and compliance testing as the company scales toward full production.

Shipping Plans and Kickstarter Offers

Kynooe is scheduled to officially launch on Kickstarter on May 20. The campaign will feature a range of launch-day discounts, limited-edition rewards, and multiple pledge options designed for different types of users, including creator-focused bundles and modular accessory packages.

According to the company, the first batch of shipments is currently expected to begin around September 2026, followed by ongoing fulfillment and second-batch deliveries planned for October 2026.

More information about the project is available on the official Kickstarter campaign page.

Verdict

Rather than building yet another industrial robotic arm, Kynooe aims to reimagine robotics as a flexible everyday companion for creators, smart homes, and personal automation. Its combination of modular hardware, AI-powered interaction, and beginner-friendly controls makes it one of the most ambitious consumer robotics projects currently entering the crowdfunding space.

The post Kynooe Launches World’s First Fully Modular AI-Powered Robotic Arm on Kickstarter appeared first on Gizmochina.