New-generation Steam Controller announced by Valve
China Unicom launches “Unicom Cube” customizable service
Google Translate introduces “Pronunciation Practice” feature
News Worth a Quick Look
Ledo L80 unveiled
On April 28, Ledo Auto launched the Ledo L80 in Hangzhou, positioning it as a “smart dual-cabin, large five-seat flagship SUV.” The vehicle measures 5145 mm in length and offers two intelligent driving solutions: a pure vision setup powered by Orin-X, or a combination of the “Shenji chip” with LiDAR. It comes in six exterior colors and four interior themes. The model emphasizes cargo space and scenario-based expansion, supporting features such as carrying three bicycles, a customized camping kit, and a three-section convertible bed, all offered as official accessories.
The Ledo L80 starts at a pre-sale price of 245,800 RMB, with the BaaS (battery-as-a-service) version starting at 159,800 RMB. Test drives will begin on May 1, with the official launch set for May 15. Source
OnePlus Ace 6 Supreme Edition released
On April 28, OnePlus released the OnePlus Ace 6 Supreme Edition, powered by the Dimensity 9500 platform and featuring a self-developed “Fengchi gaming engine” along with an “Ice Glacier cooling system.” The display is a BOE co-developed 165 Hz high-refresh-rate panel with 1800 nits global brightness, 0.1 ms grayscale response time, 96% first-frame brightness ratio, and 3840 Hz PWM dimming. The device also includes the “Lingxi touch chip,” supporting a 4000 Hz instantaneous touch sampling rate and 480 Hz multi-touch sampling, along with a gyroscope adjustable up to 500 Hz. The phone is rated IP66, IP68, IP69, and IP69K for dust and water resistance. It is equipped with an 8600 mAh “Glacier battery” supporting 120 W fast charging. In terms of imaging, it features a 50 MP main camera with a 1/1.55-inch sensor. The OnePlus Ace 6 Supreme Edition is now available across all channels, starting at 3,099 RMB (12GB + 256GB).
OnePlus also introduced the “OnePlus Gunshen Gaming Controller,” featuring a hybrid touch-and-press design with esports-grade micro-switches, priced at 449 RMB, with a 50 RMB discount when purchased together with the phone. Source
New-generation Steam Controller announced by Valve
On April 28, Valve announced that its new-generation Steam Controller will go on sale at 1:00 AM Beijing time on May 5. The controller can be used independently without a Steam Machine and is compatible with all devices running Steam or the Steam Link app, including PC, Mac, mobile devices, and Steam Deck. In terms of configuration, the controller includes standard control buttons, a Steam button, a quick access button, four assignable grip buttons, two full-size TMR magnetic joysticks with capacitive touch, and a 6-axis IMU gyroscope. It also features four haptic motors, including two LRA motors inside the touchpads and two high-output LRA motors in the grips. Each side is equipped with a 34.5 mm square touchpad with haptic feedback and pressure sensitivity, with configurable click strength. The back of the controller also includes two capacitive sensing areas.
The controller supports 2.4GHz wireless connectivity (4 ms polling rate), comes with an 8.39 Wh lithium battery, offers over 35 hours of battery life, weighs approximately 292 g, and is priced at $99. Source
China Unicom launches “Unicom Cube” customizable service
On April 28, China Unicom launched the “Unicom Cube” customizable communication service, centered around a drag-and-drop configuration experience. The service allows users to freely combine service components with high flexibility, supporting online self-adjustment and dynamic changes. In terms of pricing, the service adopts tiered data pricing: the first 20 GB per month is charged at 1.5 RMB/GB, usage between 20 GB and 50 GB is priced at 1.2 RMB/GB, and usage above 50 GB is charged at 1 RMB/GB. Voice calls are billed on a pay-as-you-go basis at 0.1 RMB per minute, while SMS and MMS are charged at 0.1 RMB per message.
The service has a base monthly fee of 39 RMB, with China Unicom promising a true pay-for-what-you-use model, ensuring users will not encounter overage charges common in traditional plans. Source
Google Translate introduces “Pronunciation Practice” feature
On April 28, marking the 20th anniversary of Google Translate, Google announced a new pronunciation practice feature. After obtaining a translation result, users can tap the practice button to read aloud, with AI analyzing speech and providing real-time feedback. The feature is based on the Gemini Understand and Ask capability introduced earlier this year and is currently available to Android users in the United States and India, supporting English, Spanish, and Hindi. Source
News Worth a Quick Look
According to Engadget citing The Information, Google has signed a confidential agreement with the U.S. Department of Defense allowing its AI models to be used for “any lawful government purpose,” including handling sensitive tasks in classified environments. The agreement restricts the use of AI for domestic mass surveillance, but Google cannot veto lawful government actions and must assist in adjusting AI safety settings and filters when requested. Prior to the agreement, more than 600 Google employees had written to CEO Sundar Pichai urging the company not to involve AI in classified military projects. Google responded that API access remains governed by standard terms and that the cooperation supports national security in a responsible manner. Source
According to 9to5Google citing leaker Lanzuk on Naver, Samsung is considering a new rear camera module design for the Galaxy S27 series, replacing the design used for the past three years. Compatibility with Qi2 magnetic accessories is cited as a key reason. Camera specifications for the next generation may also be revised, though the change remains uncertain, with reports suggesting “recent cost issues” could delay the full redesign. Source
Google’s Nearby Share (AirDrop-like feature) has now expanded support to the international versions of OPPO Find X9 Ultra and vivo X300 Ultra. Source
Alibaba launches HappyHorse 1.0 video generation model
WeChat warns of “755” spoofed scam calls
JBL releases two new audio products
CHERRY unveils MX 8.3 Pro mechanical keyboard
Microsoft internally launches the Windows K2 project to optimize the Windows 11 experience
LincPlus introduces new NAS product LincStation E1
Apple launches a 12-month installment annual subscription plan
GitHub Copilot will fully shift to a usage-based pricing model
News Worth a Quick Look
Alibaba launches HappyHorse 1.0 video generation model
On April 27, Alibaba announced that its video generation model HappyHorse 1.0 has entered limited rollout testing. According to Alibaba, HappyHorse 1.0 is open for testing to professional creators, enterprise clients, and general users. Professional creators and enterprise users can register via the official HappyHorse website and the Alibaba Cloud Bailian platform, while general users can access it through the “HappyHorse” entry at the bottom of the homepage in the latest version of the Qwen App.
In terms of features, HappyHorse 1.0 supports text-to-video, image-to-video, multi-shot storytelling, multi-aspect-ratio adaptation, and video super-resolution output. Alibaba states that the model can generate videos up to 15 seconds in length and supports multilingual content creation, including Cantonese, English, French, and Korean. In terms of pricing, the official HappyHorse website lists 720P video generation at 0.9 RMB per second, while the same capability in the Qwen App is priced as low as 0.44 RMB per second. Source
WeChat warns of “755” spoofed scam calls
On April 27, WeChat issued an official notice warning that scammers have recently used number spoofing tools to impersonate “Tencent Customer Service” or “WeChat Security Center” when calling overseas users.
According to the notice, scammers may accurately state users’ names and overseas phone numbers during calls to gain trust, then threaten users by claiming that their WeChat accounts will be suspended, inducing them to report to the police. The scammers may then offer to “assist” by transferring the call to a fake “police hotline,” where they further request bank account details and verification codes to complete the fraud. In other cases, scammers may use claims such as “linked违规 public accounts” or “suspected illegal fundraising” to deceive users.
WeChat emphasized that such spoofed calls usually begin with “755,” which differs from the official customer service number “0755.” The company clarified that its hotline has never proactively made international outbound calls and will not notify users of account violations via phone. All such information is communicated only through in-app system notifications or penalty alerts. WeChat also stated that it will never request sensitive personal information or assist in transferring calls to law enforcement agencies.
WeChat advises users to hang up immediately upon receiving such calls and avoid providing any information to prevent falling victim to fraud. Source
JBL releases two new audio products
On April 27, JBL announced two new audio products: the LIVE 780NC over-ear headphones and the BandBox Solo speaker.
The LIVE 780NC headphones come in five color options and feature an ergonomic silicone headband and metal hinge structure. They are equipped with 40mm dynamic drivers, support spatial audio and Adaptive Noise Cancelling 2.0, include a six-microphone array, and support Bluetooth 6.0. In terms of battery life, they can last up to 80 hours with noise cancellation turned off. The headphones will go on sale on April 28 at a price of 1,399 RMB.
The BandBox Solo speaker is designed for practice and creative scenarios. It features JBL sound tuning, a built-in tuner and metronome, and integrated effects, allowing one-touch sound adjustments without the need for pedals. It also supports wired headphone connections for practice and USB-C direct recording, along with AI-powered track separation and app-based smart controls. The battery life is approximately 6 hours, and the product is priced at 2,199 RMB. Source
CHERRY unveils MX 8.3 Pro mechanical keyboard
On April 27, CHERRY launched the MX 8.3 Pro mechanical keyboard, positioned as a high-end product focusing on build quality and performance, priced at 2,299 RMB.
The keyboard adopts an orange-and-black color scheme and continues with an 80% TKL layout. It is equipped with the new-generation MX Lumina Brown switches and supports tri-mode connectivity, including Bluetooth, wired, and 2.4GHz. In USB-C wired mode, it supports up to an 8kHz polling rate, while the 2.4GHz mode supports up to 4kHz polling.
In terms of battery life, the MX 8.3 Pro features a built-in 5000mAh battery, with an official claim of up to 1400 hours of usage, and supports advanced SOCD key functionality.
Regarding structure and materials, the MX 8.3 Pro uses a CNC-machined aluminum chassis. The front features a full-color LCD control screen and an X-shaped multifunction knob. Internally, it adopts a gasket structure, consisting of an FR4 gold-plated positioning plate, Poron foam, IXPE switch pads, PET acoustic pads, a hot-swappable PCB, switch socket foam, and bottom sound-dampening foam. It also comes with double-shot PBT keycaps with a fine matte finish. Source
Microsoft internally launches the Windows K2 project to optimize the Windows 11 experience
Microsoft Windows chief Pavan Davuluri stated at an internal briefing in March this year that Microsoft will optimize the system based on user feedback, aiming to make Windows 11 “the most stable operating system in the world.”
According to reports, Microsoft has internally launched an initiative called the “Windows K2” plan. The project was initiated in the second half of 2025 and focuses on addressing user concerns about bloated AI features and declining system reliability. The K2 plan revolves around three core pillars: Performance, Craft (code quality), and Reliability, continuously improving system performance through Windows Insider feedback, telemetry data, and user research.
In addition, the plan introduces “Community” as a fourth pillar. Microsoft stated it will rebuild the Windows enthusiast ecosystem, including bringing back offline Windows Insider events and encouraging development teams to respond more actively to user feedback on social media.
In terms of performance optimization, Microsoft is benchmarking against SteamOS for gaming performance1, aiming to adjust the system’s underlying architecture over the next one to two years to achieve similar results. Meanwhile, internal teams will gradually migrate to the WinUI 3 framework to improve native UI performance, and rebuild components such as the Start menu, Run dialog, and Control Panel, adding more customization options like UI scaling and menu hiding. Source
LincPlus introduces new NAS product LincStation E1
On April 27, LincPlus, a sub-brand of Techvision, announced the launch of its new ARM-based NAS product, LincStation E1.
The device adopts a 2+2 hybrid storage design, offering two 3.5”/2.5” SATA drive bays and two M.2 2280 NVMe SSD slots. In terms of hardware, the LincStation E1 is powered by a processor featuring 4× Cortex-A55 cores (2.0GHz) and a Mali-G52 GPU, supporting 4K video decoding, and comes with a memory configuration of 4GB DDR4 + 64GB eMMC.
On the software side, the device runs the self-developed LincOS operating system and supports interoperability with Android, iOS, and Windows devices.
For external I/O, the LincStation E1 includes one USB-A 5Gbps port, two USB-A 480Mbps ports, one 1GbE RJ45 port, and one HDMI 2.1 TMDS port, along with a wireless card supporting both 2.4GHz and 5GHz bands. Source
Apple launches a 12-month installment annual subscription plan
Today, Apple announced a new subscription model for the App Store, allowing developers to offer monthly payments with a 12-month commitment.
According to Apple, this subscription format allows users to pay monthly while committing to a full 12-month period, enabling developers to break down annual subscriptions into lower-barrier monthly payments. The feature is currently available for configuration in App Store Connect and can be tested via Xcode, but has not yet been officially rolled out to users in the App Store.
Regarding payment mechanics, Apple stated that users can cancel subscriptions at any time, but cancellation only prevents auto-renewal after the current cycle; payments within the committed period must still be completed. Users can view completed and remaining payments in their Apple Account, and the system will send reminders via email and push notifications for upcoming charges.
Apple noted that this feature is expected to launch alongside iOS 26.5 and corresponding system updates next month, and will be available for devices running iOS 26.4 and above. However, the subscription model will not initially be available in the United States and Singapore, with reasons and future rollout timelines yet to be announced. Source
GitHub Copilot will fully shift to a usage-based pricing model
GitHub announced that all Copilot plans will no longer use the “Premium Request” (PRU) counting model, and will instead allocate a certain number of GitHub AI Credits per subscription each month, with paid users able to purchase additional credits. Usage will be calculated based on token consumption, including input, output, and cached tokens, with specific rates aligned to the public API pricing of each model.
GitHub stated that Copilot has evolved from an in-editor assistant into an agentic platform capable of executing long-running, multi-step autonomous coding sessions and traversing entire repositories using the latest models. The existing PRU-based pricing is no longer sustainable, and shifting to usage-based billing will help maintain long-term service reliability while reducing restrictions for heavy users.
Following this change, base subscription prices for GitHub Copilot will remain unchanged, and Copilot Pro plans will include equivalent monthly AI Credits. Code completion and Next Edit Suggestions will continue to be included in all plans without consuming AI Credits. Copilot code review features, however, will consume both AI Credits and GitHub Actions runtime.
For existing annual subscribers, GitHub will provide transition arrangements. Users with annual Pro or Pro+ plans will continue under the original request-based pricing terms until their subscription expires. However, starting June 1, model multipliers will increase significantly. After expiration, users will be automatically moved to Copilot Free, with the option to upgrade to a monthly paid plan or convert the remaining value of their annual plan into credits on a prorated basis. Source
News Worth a Quick Look
Recently, data miner Billbil-kun revealed that Microsoft plans to introduce a new subscription tier called “Starter Edition” for Xbox Game Pass. This plan will offer 10 hours of Xbox cloud gaming per month, along with in-game perks and access to a library of over 50 games, playable across consoles, PC, and other devices. However, the Starter Edition will not support online multiplayer, making it the only tier in the current Xbox Game Pass lineup without multiplayer functionality. The report suggests it may launch in mid-May this year, though pricing and full benefits have yet to be officially announced. Source
On April 27, Sony announced price increases for the PlayStation 5 series and related accessories across Southeast Asian markets including Singapore, Malaysia, Thailand, Indonesia, the Philippines, and Vietnam, effective May 1. Indonesia will see the largest increase, with the PS Portal priced at 5,199,000 IDR, up 44.5% from 3,599,000 IDR. The PS5 and PS5 Digital Edition will be priced at 11,399,000 IDR and 9,999,000 IDR respectively. In Singapore, the PS5 will be priced at 849 SGD, with the Digital Edition and PS5 Pro at 764 SGD and 1,167 SGD respectively, and the PS Portal at 347 SGD. Sony stated that the price adjustments are due to global economic pressures. Source
According to a report by 9to5Google on April 26, citing informed sources, the upcoming update will cover all Google Workspace apps. The new icon design will align with Google’s gradient “G” logo and products such as Gemini, Google Photos, and Google Maps, reinforcing AI-related visual identity. The updated icons will no longer be restricted to Google’s four standard colors, instead adopting more distinctive color schemes and shapes to address user concerns about the lack of recognizability in previous designs. Source
Editor’s note: Compared to Windows 11, which runs games natively, games on SteamOS generally rely on the Proton compatibility layer. After optimization, performance loss in complex scenarios is typically around 10% to 20% at most. ↩︎
@Vanilla: To be honest, when I first opened Maestri’s official website, I thought it was just another app trying to ride the AI hype. However, after actually downloading it and using it to complete a small project, I found that Maestri performs quite well in certain scenarios, helping us better leverage AI Agents to get work done. Next, I’ll use a recent investment research project of mine as an example to see what Maestri can really do.
When you open Maestri, you’ll notice that its main interface is a whiteboard canvas. You can add elements such as terminals, tags, attachments, file trees, portals, formats, and drawings. Among them, the file tree can be thought of as folders, portals as web pages, formats as text boxes, and the rest are fairly self-explanatory.
Looking at the toolbar at the top of the canvas, in addition to the elements mentioned earlier, there is also a selection tool that allows you to move or resize elements. All buttons on the toolbar have corresponding keyboard shortcuts, usually a single letter key, making them very quick to access.
Then, at the bottom right corner of the canvas, there are three buttons: floors, minimap, and zoom.
The floor feature is something Maestri built using macOS system capabilities. It allows you to create multiple floors, essentially independent workspaces. When creating a floor, you can either start fresh or directly duplicate the base layer. In the end, you can merge them or designate one floor as the final output layer—it has a bit of a Git-like flavor.
The minimap is a thumbnail view of the entire canvas, helping you quickly locate different areas. The zoom function is even more straightforward, letting you control the zoom level using plus and minus buttons.
Currently, Maestri’s terminal supports Claude Code, Codex, Gemini CLI, OpenCode, and Shell, and you can choose among them when inserting a terminal element.
When you insert multiple terminals on the canvas, you can connect them with lines, allowing them to communicate with each other. In addition to connecting terminals, you can also link tags and portals to terminal elements, enabling AI Agents to read and edit the content within those elements.
Maestri actually relies on Maestri Skills to enable communication between AI Agents and to operate on connected tags and portals. As a result, its capabilities are limited to what Maestri Skills support—for example, AI Agents cannot create new notes and write content into them, nor can they establish links with elements like file trees or text boxes.
Overall, Maestri addresses the issue of frequent context switching and lack of data interoperability when working with multiple Agents. Through its “canvas” approach, it allows multimedia files within a project to be displayed simultaneously. Especially on large-screen devices, it can significantly improve reading and workflow efficiency.
Maestri is available for free download on its official website. A $18 purchase unlocks unlimited workspaces, authorization for two computers, keyboard navigation, and more.
Hour by Hour: A Scheduling App with Natural Language Input
Platform: iOS
Keywords: Scheduling, Natural Language Input
@ElijahLee: Hour by Hour is a scheduling app developed by an indie developer. Its standout feature is the ability to input schedules using natural language, with the app automatically recognizing event names and time arrangements, making it easy to plan out an entire day.
The inspiration for the app came from the developer’s experience creating travel itineraries. For example, if you need to catch a flight at 11:30, what time should you arrive at the airport, when should you leave home, and when should you set an alarm? To perfectly organize such a schedule—handling time calculations, convenient input, and event associations within a single app—the developer created Hour by Hour.
Adding events via natural language input is one of the app’s key features. You can type “Swim at 3pm,” and the app will automatically recognize a swimming session starting at 3 PM. Then, if you input “Leave 45 mins before,” the app will calculate that you should leave at 14:15, and the two events will automatically be linked. By enabling notifications from the top-right corner of the schedule planning page, the app will send reminders when events begin, helping you remember when it’s time to head out.
When creating a schedule, you can customize the schedule name, choose an icon, and select dates in the future. The app offers a wide range of customization options. For names, you can add both a title and a subtitle; icon choices span categories such as general, weather, transportation, food and drink, social activities, work, and exercise; date selection includes today, any future date, daily, weekdays/weekends, and recurring cycles.
For frequently repeated trips or business travels, you can quickly duplicate an entire itinerary. Within a single itinerary, you can naturally set up multi-day schedules. You can also assign different colors to distinguish between itineraries. The app also integrates with Apple Calendar to display events. Overall, the feature set is very comprehensive.
You can download Hour by Hour for free from the App Store. A one-time purchase of 68 RMB unlocks premium features, including unlimited schedules (the free version is limited to three), AI image generation (currently in testing), and more. Natural language input currently supports English only.
Voicebox: Clone Your Voice
Platform: macOS / Windows
Keywords: AI, Voice Cloning
@化学心情下2: Have you ever had this experience: wanting to record your own podcast, but stumbling over your words in front of the microphone, or making mistakes even when reading from a prepared script? This kind of nervousness when facing an audience has troubled content creators like me as well. But in the age of AI, cloning your voice and turning your written scripts into podcast content is no longer difficult. For example, a tool I recently discovered called Voicebox makes it easy for you to “appear” with your voice.
Simply put, Voicebox is a tool that can clone your voice and generate audio with the same tone. After opening the app, you first need to download a local AI model for speech generation and transcription. For instance, I directly chose the Qwen TTS 1.7B parameter model, though you can select one based on your computer’s hardware and the language you need. Once the download and model loading are complete, you can start voice sampling.
Before cloning your voice, you need to create or import a voice profile, which will be used to generate audio with the same tone. After clicking “Create Voice,” you’ll enter the recording interface. Paste the reference script into the “Reference Text” field, choose the target language and local model, and then start recording.
Once the voice profile is created, you can let the local speech model clone the “voiceprint” from your recording. In the “Voice Profiles” section, select the profile you just recorded, then paste the text you want to convert into speech into the input box below. Choose the language, local model, and voice effects, then click the generate button on the right side of the text box. After a short wait, you’ll hear audio generated based on your voice profile.
The generated audio can also be further processed. For example, you can apply effects—Voicebox offers presets such as “Robot,” “Radio,” “Echo Chamber,” and “Deep Voice.” The “Audio Channels” feature lets you manage output devices and channels. In “Settings” under “Generation,” you can also configure options like automatic chunking limits, fade-in and fade-out between segments, and audio normalization.
Beyond generating individual audio clips, Voicebox can also combine multiple audio segments into a story to create a complete audio project. For example, you can import multiple voice profiles with different styles and turn an eBook into an audiobook.
You can download Voicebox for free from its official website, with support for both macOS and Windows platforms.
SwiftMTP: A Better Way to Connect macOS and Android via Cable
Platform: macOS
Keywords: MTP, Android, macOS, File Management
@克莱德: With the combined push of national subsidies, local AI models, and tools like OpenClaw, many Android users—myself included—have gradually moved from the familiar Windows ecosystem to macOS. If you haven’t fully committed to Apple’s ecosystem, or simply prefer Android over iOS, transferring files between Android and macOS is something you’ll inevitably have to deal with.
For small daily files, we do have wireless solutions like LocalSend or Blip. But for syncing media libraries or handling large file imports and exports—the “heavy lifting”—a wired connection is still the way to go. And wired options are actually quite limited. If you don’t want to spend big money on simple tasks, nor bother with outdated tools like HandShaker (which hasn’t been updated in years) or the rather plain-looking OpenMTP, you might want to try SwiftMTP, brought to us by SSPAI contributor @Neighbor_Z.
From my less-than-a-year experience using a Mac, anything with “Swift” in its name tends to feel very “native,” and SwiftMTP is no exception. Compared to OpenMTP, which I had been using before, SwiftMTP adopts a UI design more in line with the visual style of macOS 26 Tahoe. It also embodies the lightweight and responsive nature of native Swift apps (with an installed size of around 15MB, as checked via mole). On top of existing open-source MTP backend solutions, it fixes several key performance and functionality issues. For example, in similar tools, transferring files or folders sometimes led to encoding problems when filenames contained emoji or extended CJK characters, resulting in garbled text or even missing files. SwiftMTP improves the backend encoding logic, making it more reliable—especially for users working with Chinese-language filenames.
In terms of user experience, SwiftMTP offers favorites and device management in the sidebar. The favorites section comes preloaded with commonly used directories such as Photos, Bluetooth, and Downloads, and you can also add frequently used folders from the right-hand pane. If you manage multiple Android devices, you can connect and control several devices simultaneously via the bottom section.
On the performance side, SwiftMTP improves the file traversal logic, significantly reducing the time previously required to prepare for file transfers. Transfer speeds are noticeably better than similar tools, and you can cancel transfers at any time. Also worth mentioning is the device information panel on the right—it lets you clearly see the connection protocol and transfer speed between your device and Mac. In scenarios involving long-duration, high-speed file transfers, having this kind of visibility is reassuring.
SwiftMTP feels like an enhanced and polished version of existing macOS + Android MTP solutions. With its native design and more stable, feature-rich transfer experience, it quickly became my go-to replacement for OpenMTP—a must-have app for macOS setups. The only drawback I’ve noticed so far is that the preview view in the right-hand browsing pane is still somewhat limited at this stage.
@大大大K: If you ask what’s trending lately, it’s definitely “raising shrimp” and “raising horses.” I set up an Ubuntu Server virtual machine on my NAS and deployed OpenClaw and Hermes Agent just to experiment. But they’re not always well-behaved—sometimes they inexplicably break themselves. Whenever that happens, I have to check the gateway status or manually fix things. Most of the time, though, it only takes a few commands. Booting up a computer just to SSH in feels like overkill, so I found Haven, a tool that lets me SSH remotely right from my phone.
Although it’s an SSH client, Haven doesn’t compromise on aesthetics. It adopts the Material You design and supports dynamic color theming. Functionally, it’s more than just an SSH client—it also integrates a local terminal, supports remote desktop connections via VNC and RDP, and includes file management features. This all-in-one approach really suits my needs, since these are all essential tools for remote operations.
Using a terminal or SSH on a phone isn’t exactly convenient, and Haven makes several optimizations to address this. For example, terminals often require key combinations, so Haven pins commonly used keys like Tab, Esc, Ctrl, and Alt above the keyboard. It also provides directional controls including Home and End, along with copy-paste shortcuts. Even better, Haven can natively call virtual terminal tools to maintain persistent sessions. If you’ve installed tools like tmux or screen on your remote system, you should definitely enable this feature in settings—it prevents processes from being terminated when the connection drops.
Beyond basic SSH connections, Haven includes a built-in file browser and SSH key storage, allowing you to interact with the remote file system through a graphical interface. Even more impressively, it supports running a Linux VM directly on your Android device (provided you have a local terminal tool installed), greatly expanding its versatility.
On the security side, Haven supports fingerprint unlock, screenshot blocking, tunneling, and other features. If you’re connecting to production systems, having these extra layers of protection certainly provides peace of mind. Overall, Haven is an excellent SSH client: open-source, free, and powerful enough for advanced users, while still being easy to use for beginners like me. If you need to manage SSH from your phone, you can download Haven for free from GitHub or F-Droid.
Gradient Weather: A Beautiful Yet Practical Weather App
Platform: Android
Keywords: Weather
@Peggy_: Generally speaking, a weather app is something every mobile operating system comes with by default. Even so, a wide variety of weather apps continue to be developed, each offering its own differentiation—for example, CARROT Weather, known for its sharp wit, or HeWeather, which is trusted by many for its data accuracy. Of course, there is also a category of apps that focus on design as their main selling point, and the Gradient Weather we’re introducing today stands out with its visual appeal.
When you first open the app, you can choose whether to grant Gradient Weather access to your location. If you prefer not to, you can still add cities by searching for them manually. After completing this step, you can customize the app’s appearance on the onboarding screen—from theme colors to font selection, offering a fair degree of personalization.
On the main interface, in addition to the most important current temperature, Gradient Weather places emphasis on the hourly forecast for the next 24 hours. You can tap on any specific hour to view additional details such as wind speed, UV index, and humidity. Following that is the more familiar 7-day weather forecast, helping you understand upcoming weather trends.
Scrolling further down reveals more detailed information for the current day. Gradient Weather adopts a card-based design, combining large text with colorful icons to effectively present different types of weather data. Each category uses a distinct chart style, reducing visual clutter. Of course, if you’re not satisfied with the default layout, you can customize it to prioritize the information you care about most.
In the settings interface, Gradient Weather provides even more customization options centered around information display. From the size of weather elements on the home screen, to the layout of future forecasts, to the number of days shown—you can adjust everything. The app also lists its data sources in the settings, making it easy to assess their reliability. A particularly clever touch is the built-in collection of weather-themed app icons: you can either choose your favorite manually or let the icon change automatically based on current weather conditions.
In addition, Gradient Weather offers different notification types for various weather conditions, which you can later adjust based on what matters most to you.
@Snow: Although China’s perfume market has grown rapidly in recent years, compared to some Western countries, perfume is still a relatively niche consumer product domestically. Enthusiasts often share scent preferences in WeChat groups or on Xiaohongshu, and record their testing notes in local journals or apps like Fragrantica. But when it comes to deciding which fragrance to “wear” before heading out, you probably still have to go through a bit of an internal struggle. Earlier this month, an app called Kaori launched on the App Store—it may help you manage complex scent memories while also easing the dilemma of choosing what to wear.
Open Kaori and tap the “+” button in the bottom right corner to “Add Perfume,” and you can start registering your collection. Kaori supports customizing details such as perfume name, brand, volume, scent profile, concentration, purchase location, purchase date, usage preferences, related memories, and more. It isn’t limited by third-party databases, so whether it’s a niche fragrance or a custom blend, it can have a place in your personal archive.
When entering a perfume name or adding an image, the app will automatically call a large model for matching. If a match is found, Kaori will provide suggestions, letting AI fill in part of the data for you. The AI can also analyze uploaded perfume images and isolate the main subject—so long as the shooting angle isn’t too unusual, Kaori can maintain a clean and consistent visual presentation.
One thing to note is that the “scent profile” field in Kaori does not refer to listing top, middle, and base notes. Instead, it uses the “fragrance wheel” model for classification. For those new to perfumes, if the AI fails to match automatically, you may need to familiarize yourself with how the fragrance wheel works.
The user preferences set when registering perfumes include three dimensions: season, time of day, and personal liking. These serve as the foundation for Kaori’s intelligent fragrance recommendations. The app combines real-time weather, temperature, and seasonal data from your location to suggest three perfumes best suited to the current moment. You can tap an item to quickly log it, or tap the “+” button in the bottom right to manually select a fragrance based on scent categories.
In the “Memories” section, you can review all your recent wearing records and understand your latest scent preferences. Within the perfume archive, you can also track usage frequency for each fragrance and gain insight into the contexts in which you tend to favor them—this may also help inform your future purchasing decisions.
In addition, Kaori offers beautifully designed sharing templates. You can place your daily outfit photos, travel experiences, or casual perfume snapshots into these templates, add your olfactory notes and text, and share them with fellow enthusiasts.
However, as a newly released app, Kaori still has some limitations. For example, its AI model currently cannot match Chinese names— even for well-known fragrances like Chanel Bleu, you’ll need to input “Bleu” to get a match. Additionally, Kaori does not distinguish between “tested” and “owned” fragrances like Fragrantica does. If, like me, you’re very cautious about purchasing full bottles but have a large collection of samples or decants, Kaori may feel somewhat limited for managing them.
Kaori includes an in-app subscription: 8 RMB per month or 68 RMB per year. The paid version unlocks unlimited perfume entries, a full calendar feature, watermark-free sharing, and iCloud backup.
OpenAI has released the GPT-5.5 model, focusing on stronger capabilities in programming, computer operation, and deep research.
DeepSeek has unveiled a preview of its next-generation V4 model and open-sourced it simultaneously, emphasizing enhanced reasoning, programming, and Agent capabilities. It supports up to 1 million tokens of context and aims to directly compete with flagship models from OpenAI and Google at a lower cost.
The standalone chat app XChat from X is now officially available, allowing login via X accounts. The app is not yet available in mainland China.
The read-later tool Matter introduced API and CLI support this month.
U.S. asks diplomats to warn other countries about China’s AI distillation practices
Domestic AI firms told not to accept unapproved U.S. investment
Hong Kong steps up e-cigarette ban promotion ahead of Golden Week
FILCO keyboard parent company shuts down
Authorities respond to claims that Doubao accessed 2026 Shandong civil service exam results early
News Worth a Quick Look
X launches messaging app XChat
On April 24, X officially launched its standalone messaging app XChat to the public, currently available only on iOS. XChat uses X accounts for login and supports text messaging, file sharing, voice and video calls, as well as group chats. Meanwhile, due to low usage and widespread spam on X’s existing Communities feature, it will be shut down, with XChat serving as the new hub for migrating community members.
In terms of privacy, XChat claims to be free of ads and tracking mechanisms, and offers features such as message editing and recall, disappearing messages, and screenshot protection. X also states that all messages are protected by a PIN code and use end-to-end encryption, though the industry remains skeptical. During early testing, experts warned that XChat’s security might not match that of mature encrypted apps like Signal, and the security of the official release still requires further evaluation.
X’s Head of Design, Benji Taylor, revealed that the current version is only the beginning of X’s push into instant messaging. The company is also internally testing a standalone payments app. Previously, Elon Musk had envisioned integrating messaging, payments, content creation, shopping, and AI into a single “super app” similar to WeChat. However, the launch of XChat suggests a shift toward a more modular product ecosystem.
U.S. asks diplomats to warn other countries about China’s AI distillation practices
According to Reuters, the U.S. State Department sent a diplomatic cable last Friday to embassies worldwide, instructing American diplomats to warn foreign governments that Chinese AI companies such as DeepSeek may be “extensively stealing” intellectual property from U.S. AI firms through model distillation. Distillation refers to using outputs from large models to train smaller, lower-cost models. The cable also indicates that separate diplomatic démarches will be made to the Chinese government on this issue.
The document names companies including DeepSeek, Moonshot AI, and MiniMax, claiming that unauthorized distillation allows foreign firms to release comparable products at significantly lower costs, potentially undermining safety mechanisms and ideological neutrality in original models. Previously, OpenAI had raised similar concerns to U.S. lawmakers.
In response, the Chinese Embassy in the U.S. dismissed the claims as baseless and described them as a deliberate attack on China’s AI industry. DeepSeek has previously stated that its V3 model was primarily trained on naturally collected web data, not synthetic data generated by OpenAI models.
Domestic AI firms told not to accept unapproved U.S. investment
According to Bloomberg, regulators are planning to restrict domestic tech companies—especially well-known AI startups—from accepting U.S. capital without prior government approval.
In recent weeks, agencies including the National Development and Reform Commission have issued guidance to several private tech firms, requiring them to reject unapproved U.S. funding. Companies such as Moonshot AI, which is seeking to expand financing, and StepFun, which is preparing for a Hong Kong IPO, have reportedly received such instructions. StepFun is also restructuring its overseas corporate setup in line with regulatory requirements. Additionally, authorities have imposed similar restrictions on ByteDance, prohibiting the transfer of existing shares to U.S. investors without approval.
These measures stem from concerns over the outflow of domestic technology. In December 2025, Meta announced the acquisition of AI agent startup Manus. Although registered in Singapore, Manus was founded by Chinese entrepreneurs. Prior to the acquisition, the company restructured and relocated its domestic staff to Singapore to complete the deal without regulatory review from Beijing. Authorities are now conducting a joint investigation into the transaction and its implications, and one of its co-founders has reportedly been restricted from leaving the country.
Previously, the U.S. had already implemented rules in 2025 banning its capital from investing in Chinese semiconductor, quantum computing, and AI companies.
Hong Kong steps up e-cigarette ban promotion ahead of Golden Week
According to the South China Morning Post, as the Golden Week holiday approaches, Hong Kong has intensified promotion of its upcoming e-cigarette ban at airports, land border crossings, and mainland subway stations, while also collaborating with mainland media to raise awareness.
Under the revised Smoking (Public Health) Ordinance, starting April 30, 2026, possession of specified alternative smoking products—including e-cigarettes, heated tobacco products, and herbal cigarettes—will be prohibited in public places. Anyone found using or carrying activated alternative smoking products will be presumed to possess such items. Violations constitute a criminal offense; for small quantities intended for personal use, offenders will face a fixed penalty of HKD 3,000. Tourists will receive electronic penalty notices and can pay fines via Alipay or WeChat.
Authorities have also promoted earlier measures, including expanding statutory no-smoking areas to within three meters of entrances to designated premises such as care homes, schools, and hospitals. Smoking is also prohibited while queuing, including when waiting in lines of two or more people for public transport, within designated boarding areas (such as bus shelters), and when entering venues like cinemas, museums, and stadiums.
FILCO keyboard parent company shuts down
According to GIGAZINE, Japan’s Diatec, best known for its mechanical keyboard brand FILCO, announced on its official website that it has ceased operations as of April 22, 2026.
In the announcement, Diatec apologized to consumers for the “sudden” news and thanked users for their long-term support. The company emphasized that all personal user data collected through e-commerce sales and after-sales support had been destroyed on April 22.
FILCO is a landmark brand in the history of mechanical keyboards. Its classic Majestouch series is known for durability and excellent typing feel, long regarded as a benchmark among high-end mass-produced mechanical keyboards, with strong reputation and recognition among enthusiasts.
In recent years, Diatec had continued to iterate on its products. In 2022, it launched the Majestouch Convertible 3, supporting both wired and wireless connectivity, offering multiple switch options including red, brown, blue, and silent red, along with various layouts. In 2023, it also introduced the Majestouch Xacro M10SP, a split mechanical keyboard with 10 macro keys.
Authorities respond to claims that Doubao accessed 2026 Shandong civil service exam results early
According to Yangtze Evening News, on the evening of April 23, a user claimed online that they had accessed the 2026 Shandong public institution exam results using Doubao, even though the results were officially scheduled for release the following day. Screenshots of the results were shared before the post was later deleted.
In response, staff responsible for application policy consultation stated that since results were to be released on April 24, the query portal had been tested the night before. “Some users accidentally accessed it,” they said, adding that the test portal was promptly shut down after discovery and caused no adverse impact. Analysts believe the incident may have occurred because Doubao inferred the result URL based on patterns from previous years, while the official system began testing without adequate safeguards, leading to the so-called “early access.”
Such incidents caused by predictable URLs are not uncommon. In January 2011, Microsoft’s financial results were discovered hours early by a data company, forcing an early release. In May 2024, the U.S. Bureau of Labor Statistics uploaded CPI data 30 minutes ahead of schedule to a publicly guessable URL, though it went unnoticed. In November 2025, the UK’s Office for Budget Responsibility leaked its spring budget report a week early in a similar way, with at least seven IP addresses accessing it, leading to the chairman’s resignation.
News Worth a Quick Look
Analyst Dan Nystedt claims that the standard version of iPhone 18 (expected to be delayed to 2027) will feature 12GB of RAM, a 50% increase over the iPhone 17’s 8GB.
OpenClaw 2026.4.24, released on April 26, sets DeepSeek V4 Flash as the default model.
On April 24, Meta issued an internal memo announcing plans to lay off about 10% of its workforce (around 8,000 employees) on May 20 and cancel hiring for 6,000 open positions to offset heavy AI investment costs. As of early this year, Meta employed nearly 79,000 people globally. Affected U.S. employees will receive 16 weeks of pay, plus two additional weeks for each year of service, along with 18 months of healthcare coverage for themselves and their families. Around the same time, Microsoft announced its first voluntary buyout (early retirement) program for U.S. employees, open to senior directors and below whose combined age and tenure equal at least 70.
According to TechCrunch, surging demand for running local AI models has led to a complete sellout of the M4 Mac mini through official channels, driving up prices in the second-hand market. The base model, originally priced at $599 (16GB RAM, 256GB storage), is currently unavailable for delivery or in-store pickup on Apple’s website, while higher-capacity versions are backordered until June. Multiple outlets note this is the first time the base model has completely sold out. On platforms like eBay, lightly used units are now listed above $700, with near-new units priced between $715 and $795.
Recent recommendations: [TV] For All Mankind Season 5, [K-Drama] Yumi’s Cells Season 3, [K-Drama] Bloodhounds Season 2, [UK] The Young Offenders Season 5, [CN]Eight Thousand Miles of Clouds and Moon, [JP]Mysterious Footage Sent to TV Stations, [Film] 5 Centimeters per Second, [Film]The Sato Couple Story, [Anime]Awashima Hyakkei, [Anime]Ina Botan, Drunk Like a Lily, [Anime]Kirio Fan Club
A few notable trailers: Mud Man first trailer, The Death of Robin Hood new trailer, Masters of the Universe: He-Man Rising new trailer, Spider-Man Noir new trailer, Cold War 1994 final trailer
Some industry news: Minions & the Big Monster confirmed for release, The Sheep Detective Squad set for mainland China release on May 16, Star Wars: The Mandalorian & Grogu set for May 22 release in mainland China, Elden Ring live-action film scheduled for March 2028, Ultraman Origins release poster unveiled
[TV] For All Mankind Season 5
Keywords: Drama / Sci-Fi
Also known as: For All Mankind Season 5
Runtime: ~60 minutes per episode × 10 episodes; Douban link
For the stars of all mankind.
@潘誉晗：Season five sets its timeline in the 2010s. More than a decade has passed, and the Mars base has become highly developed and well-established. “Happy Valley” has evolved into an extraterrestrial colony with thousands of residents. Meanwhile, authorities on Earth aim to impose stricter regulations and laws to exert tighter control over the red planet, naturally triggering dissatisfaction and protests among Martian residents. When a body is discovered in the Martian wilderness, a new crisis emerges for this growing colony.
By season five, the story expands in scale, continuing its signature measured pacing as it gently unfolds a distant space narrative. The cross-generational arc, especially Ed’s storyline, is deeply moving. Still committed to guarding this land of belief, Ed feels the generational divide with younger people. On a political level, the tension between Earth and the Martian colony becomes especially compelling. Mars was once humanity’s dream destination, but as it gains the ability to stand independently, its relationship with its home planet grows increasingly complex and sensitive.
[K-Drama] Yumi’s Cells Season 3
Keywords: Drama / Comedy / Romance
Also known as: 유미의세포들 시즌 3 / Yumi’s Cells 3
Runtime: ~60 minutes per episode × (ongoing); Douban link
Cells left unused for too long will fall asleep~
@潘誉晗：In season three, Yumi has become a popular novelist. Her love stories have gained countless fans, but her success also brings a crisis to her cells—her life has become too smooth and uneventful, causing her “cells” to fall into dormancy. That changes with the arrival of this season’s male lead, Shin Sun-rok. As Yumi’s editor, he is unusually calm and restrained for his age, with strict work principles that often clash with Yumi’s personality. Under this stimulation, the long-dormant cells begin to awaken one by one.
As Korea’s first live-action and animation hybrid series, Yumi’s Cells continues its dual narrative style—Yumi’s real-life story alongside the perspective of her brain cells. The cell world remains the show’s biggest highlight. For audiences used to viewing emotions through a live-action lens, this innovative storytelling offers a direct way to step into the “cell village,” reflect through the cells’ actions and dialogue, and better understand their own inner needs.
Every punch is thrown to protect family and friends.
@潘誉晗：The story continues from the previous season. After dismantling a loan shark syndicate, Kim Geon-woo and Hong Woo-jin pursue their boxing dreams. But life is far from peaceful. Geon-woo’s skills catch the attention of Baek Jeong, who runs an underground fighting league called “IKFC” with millions of online members. He profits by streaming matches irregularly. Geon-woo rejects the lucrative offer, but that refusal brings even greater danger. Baek Jeong, having already announced Geon-woo’s participation, resorts to illegal tactics—kidnapping his mother and harming those around him—to drag him into the abyss.
Three years after its debut, the hit series returns with its second season. While the story still has some flaws, its action remains compelling enough to overshadow them. The underground fighting setting expands the scope of crime globally, raising the stakes and making the action sequences even more intense. The cage fights blend combat techniques seamlessly, delivering adrenaline-pumping scenes.
[UK] The Young Offenders Season 5
Keywords: Comedy
Also known as: The Young Offenders Season 5
Runtime: ~30 minutes per episode × 6 episodes; Douban link
Two idiots’ ridiculous teenage years.
@潘誉晗：Jock escapes from prison and sneaks back into Ireland, calling Conor to pick him up. After waiting by the coast all day, Conor finally sees Jock appear—floating with a swim ring. The two lovable “young offenders” reunite once again.
Based on the Irish comedy of the same name, The Young Offenders focuses on Conor and Jock, who have never been considered model youths. From high school to adulthood, they remain troublemakers—dirty uniforms, stolen bikes, ruined graduation ceremonies were just the beginning. Even after serving time in prison, they remain unreliable. They try new jobs and relationships, but still make a mess of life. But who says you have to become someone extraordinary to live a meaningful life?
Conor and Jock are like two naive fools—arguing over trivial matters yet fiercely loyal to each other. They stumble through chaos and mischief, growing along the way, while audiences find laughter and comfort in their absurd adventures.
@潘誉晗：In 1937, the Battle of Shanghai breaks out. Nationalist officer Zhang Yunkui leads his troops in a heroic resistance, but his entire brigade is ultimately wiped out. In his final moments, Zhang ensures that a conscripted cook, Wan Fu, escapes and asks him to deliver news of his death to Nanjing. The message devastates Zhang’s family, especially his pregnant wife Yu Jiao. But just as despair sets in, they discover a newspaper photo showing Zhang alive in a hospital. Wan Fu, along with Yu Jiao and Zhang’s father, sets out for Shanghai to find him. Meanwhile, Zhang, who miraculously survives, is falsely accused of desertion by Nationalist authorities. After failing to clear his name, he joins the Communist Party and continues fighting for his country.
Produced by the same team behind The Age of Awakening, the series premiered earlier this month on CCTV-8. While it also focuses on wartime stories, its greatest strength lies in its portrayal of a rich ensemble cast. It highlights ordinary people struggling amid war—from those driven by money, to those fighting for survival, and ultimately to those standing firm with dignity. Even the smallest figures shine with profound humanity.
[JP] Mysterious Footage Sent to a TV Station
Keywords: Drama / Thriller / Horror
Also known as: 放送局に届いたある映像
Runtime: ~30 minutes per episode × 3 episodes; Douban link
Do you want to become a complete person?
@潘誉晗：A radio station employee, Ito, receives two letters, each containing a videotape along with a note that reads “Please play.” After watching them, he finds the content nearly identical—both tapes show human-shaped figures, one male and one female, with different body parts marked by red dotted lines. The unsettling background music carries a vaguely religious tone: “Let us strive to become complete humans.” Curious, Ito asks someone to analyze the audio. The producer tells him the music contains a special frequency—963 Hz, also known as the “God frequency.”
Renowned Japanese filmmaker Iwasawa Hiroki delivers another standout work in the form of a mockumentary. Old videotapes mailed only on Sundays, a mysterious drifter with seemingly supernatural abilities, an orphanage hiding secrets, and a red-clad woman lurking in the footage—these elements may seem familiar, even cliché. Yet when combined with the director’s themes of “fragmentation” and “completeness,” they amplify the horror to an unsettling degree. The series is short and sharp, but leaves a lingering impact.
[Film] 5 Centimeters per Second
Keywords: Anime adaptation / Drama / Romance
Also known as: 秒速5センチメートル / 5 Centimeters per Second
The speed at which cherry blossoms fall is 5 centimeters per second.
@SHY：Takaki Toono and Akari Shinohara were close childhood friends who were separated after moving away, promising to reunite under the cherry blossom tree. Years later, Takaki is buried in work and his personal life is in disarray, while Akari has moved on with new encounters. As the day of their promise approaches, will their diverging paths ever cross again?
Adapted from Makoto Shinkai’s early classic, director Okuyama Yoshiyuki—originally a photographer—brings a music video-like aesthetic to the feature film, creating breathtaking visual beauty. Hokuto Matsumura, who voiced Souta in Suzume, leads the cast, with careful casting and set design recreating iconic scenes from the animation, retelling this tender coming-of-age love story in a new way. When the remixed theme song One more time, One more chance plays, it is sure to stir familiar emotions among audiences.
As the original animated version runs only 63 minutes, the live-action film expands the narrative. Screenwriter Suzuki Fumiko preserves the original three-part structure while adding well-integrated new elements. Compared to the animation’s subtle emotional tone, the live-action version presents feelings more directly. While it sacrifices some of the original’s intentional ambiguity, it establishes its own identity. As time buries past promises, we ultimately learn to treasure memories and move forward.
@利兹与青鸟：At 22, Sato Sachi is lively and cheerful; at 23, Sato Tamotsu is serious and gentle. The two meet and fall in love at a university coffee club. What seems like a harmonious relationship takes an unexpected turn—Sachi, who studied law to support Tamotsu, passes the bar exam first and becomes a lawyer. Meanwhile, Tamotsu continues juggling part-time jobs and exam preparation, also taking on the role of a homemaker. This imbalance becomes a source of tension, leading to frequent arguments over trivial matters. Despite this, they happily marry after Sachi becomes pregnant, firmly believing that “marriage won’t change their relationship.” But will that belief hold true?
The film builds its narrative from everyday details, piecing together a non-traditional marital story. The couple may quarrel over mundane issues or find joy in small moments, yet both are weighed down by life’s hardships and societal expectations. These grounded, relatable moments add warmth to the story. By juxtaposing Sachi’s divorce cases with the couple’s own relationship, the film explores conflicts that often have no clear resolution, leaving a lingering sense of melancholy. The pairing of Yukino Kishii and Hio Miyazawa is visually charming, and their performances bring additional depth to this familiar theme.
[Anime] A Hundred Scenes of Awajima
Keywords: Manga adaptation / School / Youth
Also known as: 淡島百景 / A Hundred Scenes of Awajima
Runtime: 24 minutes per episode × 12 episodes, updated weekly on Thursdays
All the stars that shine brightest burn out the fastest.
@SHY：Awajima Opera School, known for producing countless stars, is a dream destination for many young girls. Preparatory student Tabata Wakana, dorm leader Takehara Kinue burdened by a friend’s expectations, elegant scholarship student Okabe Emi, and attention-seeking Ibuki Keiko—these vivid personalities are about to write their youth on this stage.
Although set in an opera school, creator Takako Shimura focuses not on the glamorous spotlight, but on the undercurrents beneath it. There is no single protagonist—everyone gets their moment, portraying the bittersweet experiences both on and off stage. Interwoven perspectives piece together independent yet connected stories, forming a complex web of relationships that span decades, reflecting on the ripple effects of life choices and recalling an irreplaceable period of youth.
Directed by Morio Asaka, known for works like Cardcaptor Sakura, NANA, and Chihayafuru, the series excels at capturing delicate emotions. His strong direction compensates for production limitations, rearranging the source material into a more cohesive narrative flow while highlighting contrasts between light and shadow. This gentle watercolor-like tragicomedy unfolds like a slow, contemplative prose collection—without grand spectacle, yet capable of leaving lasting ripples in the heart.
[Anime] Botan Kamiina Fully Blossoms When Drunk
Keywords: Manga adaptation / Slice of life / Food
Also known as: 上伊那ぼたん、酔へる姿は百合の花 / Botan Kamiina Fully Blossoms When Drunk
Runtime: 24 minutes per episode × 12 episodes, updated weekly on Fridays
@SHY：First-year college student Botan Kamiina moves into the dormitory and, during a spring outing with her roommates, accidentally comes across dorm leader Ibuki Tonami drinking alone. Drawn to her demeanor, Botan—out of curiosity—tries alcohol for the first time. From then on, she begins exploring different drinks, and as the two become drinking companions, their relationship gradually evolves.
Finally leaving high school behind, it’s only natural to try something a little more “adult.” Nicknamed “Liqun & Tsingtao,” the series unfolds in a tipsy, free-flowing atmosphere. The original creator, a programmer-turned-manga artist known as Hei, is a devoted fan of Encouragement of Climb and even contributed as an animator; this influence is evident in the character design and overall tone. Usually well-behaved, Botan reveals her true nature after drinking—her intentions aren’t entirely about the alcohol, leaving others flushed and flustered. Beneath the ambiguous atmosphere, the use of negative space creates a sense of airiness, capturing the subtle charm of everyday life.
Founded last year with mono Girls, the new animation studio Soigne continues its strong momentum, delivering quality that feels celebration-worthy. The handheld DV-style opening captures fleeting moments with shaky camerawork, while the ending—featuring rotating vocal performances by the characters—is fully handled by Yoshinari Go. Episode 2, storyboarded by Yusuke Yamamoto with key animation solely by Yusuke Matsuo, is especially mesmerizing. This well-balanced brew feels like a strong drink—intense at first sip, yet leaving a lingering sweetness, and before you realize it, you’ve already had too much.
[Anime] Kirio Fanclub
Keywords: Manga adaptation / Slice of life / Comedy
Also known as: 霧尾ファンクラブ / Kirio Fanclub
Runtime: 24 minutes per episode × 12 episodes, updated weekly on Thursdays
@SHY：Best friends Aimi Miyoshi and Nami Someya both fall for their classmate Kirio and form a rather peculiar alliance. Despite having no real interaction with him, the two come up with increasingly absurd schemes to catch his attention—yet achieve little success. But is this grand, dramatic crush really as straightforward as it seems?
Don’t be misled by the title—Kirio, the nominal center of attention, is mostly just a background figure, rarely even showing his face. The real focus lies in the comedic double act between the two female leads. Though they constantly talk about Kirio, their true intentions may be something else entirely. Their confessions, often omitting the object, might not even be directed at him.
The awkwardness extends beyond the main duo—every character here is eccentric. One-sided affections connect into an N-pointed star, with true feelings remaining elusive, giving the story a surprising hint of mystery. Mixing elements of BG, BL, and GL into a chaotic blend, it creates a tangled web of relationships that feels impossible to untangle—making you want to dedicate a song of unrequited love to everyone involved.
The anime adaptation preserves the original’s narrative tricks, delivering quirky humor that feels delightfully offbeat. Even with dialogue-heavy scenes, it never becomes dull, and each episode ends with a punchy twist. It may feel confusing at first, but once the groundwork is laid, the fun only grows. Whether these relationships are rivalry or romance is left for you to decide.
More
[Film] Case 137 @利兹与青鸟：A 20-year-old man is critically injured after being struck in the head by a riot-control projectile during a protest on the streets of Paris. Stéphanie takes on the case, only to face distrust and hostility from both the public and fellow police officers. The film meticulously portrays the complexity of collaborative investigations, professional interrogations, and evidence verification, blending chaotic real-world footage with carefully crafted cinematic shots to balance realism and artistry. While reflecting on French social issues and the struggles of marginalized groups, it also introduces moments of warmth through a stray kitten rescued by the protagonist.
[Film] Love Like Mine @利兹与青鸟：Recently awarded Best Actress at the 44th Hong Kong Film Awards, the film tells the little-known story of Amei, a woman with disabilities, and Akin, a sex worker. It focuses on Amei’s psychological growth and emotional journey—as she struggles against her mother’s overprotection and her own physical limitations, she seeks autonomy over her body and courageously pursues love. The characters are richly and delicately portrayed, though the pacing and editing feel somewhat conventional. The subject matter carries strong social relevance, continually questioning how society and able-bodied individuals should treat people with disabilities.
[JP] Rosanjin’s Kitchen @利兹与青鸟：Kitaoji Rosanjin, renowned in Japan for his culinary mastery, lived in seclusion in his later years while maintaining his reputation. Arrogant and notoriously difficult to interview, he initially intimidates a cautious editor, yet is drawn to the straightforward personality of reporter Yoneko. He invites her to cook for the Prime Minister and later guides her into the mountains to search for premium ingredients. Told across four seasonal episodes, the series reveals a different side of Rosanjin—his artistry in cuisine and his aesthetic philosophy—captured through understated yet classically styled cinematography that highlights both exquisite food and the beauty of nature.
[K-Drama] Undercover Miss Hong @潘誉晗：To investigate the financial flows of a well-known securities firm, 35-year-old financial regulator Hong Jinbao disguises herself as a 20-year-old recruit and infiltrates the company. Complications arise when she discovers her ex-boyfriend is now her new boss. As she carries out her mission, she must also keep her true identity hidden from him. Set in 1997, the story is steeped in retro charm, blending light comedy with plenty of humorous moments, while also delivering a touching portrayal of friendship between the female lead and her roommate.
[TV] XO, Kitty Season 3 @潘誉晗：As graduation approaches, the energetic Kitty has already planned her future—focusing on her studies, life, and her unresolved romantic tension with Min-ho. Unlike typical teen dramas, this series follows an American high school girl studying abroad in South Korea while pursuing love. The show devotes significant attention to her cultural exploration and adaptation, adding a layer of social depth beyond its romantic storyline.
[UK] The Unchosen @潘誉晗：In a remote British countryside lies a secluded Christian community where Rosie lives with her husband and daughter, strictly adhering to its doctrines. Her life changes when her daughter nearly drowns, leading her to meet a mysterious man—who turns out to be a wanted criminal. His arrival shatters the community’s fragile order, prompting Rosie, who has been deeply conditioned by religious beliefs, to begin questioning herself and awakening to a new sense of identity. The series explores female self-growth and carries notable real-world significance.
[JP] Rebirth: The Last Hero @Sholmes：Company president Neo resorts to ruthless tactics to force shop owners out in order to acquire their land. While visiting a shrine, he is pushed down the stairs and loses consciousness. Upon waking, he finds himself transported 14 years into the past, now living as the son of the shopping district’s chairman. As he navigates this new identity, he also searches for the truth behind his attempted murder. Beneath its mystery-driven narrative lies a reflection on human nature, contrasting the warmth of ordinary people with the arrogance of those in power.
[Anime] One-Room Manga Café Daily Life @SHY：Meiko Morita, a girl from the countryside, seizes a rare opportunity to transfer to a prestigious girls’ school in Tokyo and moves into a student-run manga café. Mistaken for a manga artist of the same name, she teams up with her eccentric roommates to save the struggling café, leading to a lively and chaotic daily life. A classic Kirara-style four-panel manga adaptation, the series may feel somewhat formulaic in its characters and story, but its high concentration of “moe” charm and Easter eggs referencing other works from the same publisher make up for it—perfect for a light and wholesome viewing experience.
On April 23, DC’s new film Clayface released its first trailer and is set to premiere in North America on October 23. Directed by James Watkins, and starring Tom Rhys Harries, Naomi Ackie, and Max Minghella, the story follows a B-movie actor who injects himself with a substance to maintain his performance, only to have his face and body transform into a walking mass of clay. Source
On April 22, A24 released a new trailer for The Death of Robin Hood, set to hit North American theaters on June 19. Directed by Kurt Sutter, and starring Hugh Jackman, Jodie Comer, and Bill Skarsgård, the film follows a war-weary, solitary Robin Hood who is gravely injured and falls into the hands of a mysterious woman who offers him a chance at redemption. Source
On April 23, Masters of the Universe: He-Man Rising released a new trailer and will premiere in North America on June 5. Directed by Travis Knight and starring Nicholas Galitzine, the story follows a long-forgotten prince who awakens under the call of a mysterious sword and must become the legendary He-Man to save both his former homeland and the present world. Source
On April 23, the series Spider-Man Noir released a new trailer and will premiere on Prime Video on May 27, available in both black-and-white and color versions. Starring Nicolas Cage, the story focuses on an aging, down-on-his-luck private detective in 1930s New York, who is also the city’s only superhero, forced to confront his past. Source
On April 22, the film Cold War 1994 released its final “power struggle” trailer and will premiere on May 1. Directed by Sunny Luk, and starring Daniel Wu, Terrance Lau, Wu Kang-ren, and Tse Kwan-ho, with special appearances by Chow Yun-fat, Aaron Kwok, Tony Leung Ka-fai, and Louis Koo, the story is set on the eve of Hong Kong’s 1997 handover, where every character must face a final choice amid a major power reshuffle. Source
More
The TV anime Dragon Ball Super: Beerus released a “super battle” trailer. This enhanced version of Dragon Ball Super tells the story of the God of Destruction Beerus awakening and descending upon Earth, leading to a universe-shaking battle with Goku. The ending also hints at Frieza’s revival. The animation reconstructs the storyline and significantly redraws the visuals. It is set to air in Fall 2026. Source
The first trailer for Evil Dead 6: Burning has been released. Directed by Sébastien Vaniček (Infested) and co-written with Florent Bernard and Sam Raimi, the film stars Souheila Yacoub, Hunter Doohan, Luciane Buchanan, Tandi Wright, and Viktor Ndukwe. It is scheduled for release in North America on July 10. Source
The Taiwanese series Smart Town released its first trailer. Directed by Hsieh Chun-yi and starring Gigi Leung, Chen Yan-fei, and others, it builds a new narrative world based on multiple short stories by Junji Ito, including Tomie, Souichi’s Convenient Curse, Slug Girl, The Hanging Balloons, and Fragments of Horror. The series will premiere on Netflix in 2026. Source
The first teaser for Silo Season 3 has been released. Starring Rebecca Ferguson and directed again by Michael Dinner, the series is adapted from Hugh Howey’s bestselling sci-fi novels. The story continues from Season 2, where Juliette survives but loses her memory, now facing new threats. It will premiere on Apple TV+ on July 3.
The film Secret Mission has announced its release date. Written and produced by Wang Wei and directed by Chung Siu-hung, it stars Celina Jade and Shawn Yue, with special appearances by Vivian Chow, and also features Jiang Luxia and Qu Jingjing. The film tells the story of a female counter-terrorism special operations unit and is set to release on June 19.
📽 Weekly Film & TV News Roundup
Minions & the Big Monster Confirmed for China Release
On April 23, the animated film Minions & the Big Monster was officially confirmed for release in mainland China, alongside a trailer and poster, with the release date yet to be announced. Directed by Pierre Coffin and written by Brian Lynch, produced by Illumination Entertainment, the story follows the mischievous Minions as they set out to create their dream monster movie—only to accidentally gather a lineup of real monsters, leading to chaotic battles. Source
The Sheep Detective Squad Set for May 16 China Release
On April 24, the film The Sheep Detective Squad released a China mainland trailer and poster, announcing its release on May 16. Directed by Kyle Balda (Minions series), the story features Hugh Jackman as a mysteriously murdered shepherd. A group of fluffy sheep, inspired by his detective stories, begin investigating clues, searching for the real culprit in a comedic mystery adventure. Source
Star Wars: The Mandalorian & Grogu Set for May 22 China Release
On April 20, Star Wars: The Mandalorian & Grogu released its China mainland trailer and poster, confirming a May 22 release date, day-and-date with North America. Directed by Jon Favreau and starring Pedro Pascal, Jeremy Allen White, Sigourney Weaver, and Martin Scorsese, the galaxy’s coolest bounty hunter and cutest Force user set out on a new journey. Source
Elden Ring Live-Action Film Set for March 2028 Release
On April 20, the live-action adaptation of Elden Ring was scheduled for release in North America on March 3, 2028, alongside the announcement of its main cast. Directed by Alex Garland and starring Kit Connor, Ben Whishaw, Cailee Spaeny, Tom Burke, Havana Rose Liu, Sonoya Mizuno, and Jonathan Pryce, the film is co-produced by A24 and Bandai Namco. Source
Ultraman Origins Release Poster Announced
On April 21, the documentary Ultraman Origins, created to commemorate the 60th anniversary of the Ultraman series, announced its theatrical release in Japan on July 3. Planned by Hirokazu Kore-eda, the film features contributions from leading filmmakers and creators including Guillermo del Toro, Hideo Kojima, Hideaki Anno, and Shinji Higuchi, each offering unique perspectives to explore the question: “What exactly is Ultraman?”
OpenAI has announced the GPT-5.5 series, describing it as the most intelligent and intuitive model version to date. Compared to GPT-5.4, the new models deliver stronger performance in coding and debugging, online research, and generating documents and spreadsheets, while also supporting cross-tool collaboration to execute complex tasks. Users can input multi-step, unstructured requests, and the model can automatically handle task planning, tool invocation, result verification, and uncertainty management. In Codex scenarios, GPT-5.5 significantly reduces token usage and introduces more robust safety mechanisms.
GPT-5.5 will be gradually rolled out to ChatGPT Plus, Pro, Business, and Enterprise users, and will be integrated into the Codex platform. The Pro tier will offer more advanced capabilities. Source
Hanvon launches Clear7C color e-ink e-reader
On April 23, Hanvon launched the Clear7C color e-ink e-reader. The device features a 7-inch Kaleido 3 color e-ink display with a resolution of 1680×1264, supporting 150 PPI in color mode and 300 PPI in black-and-white mode. It also includes physical page-turn buttons on one side. The Clear7C weighs approximately 175g and is just 3.9mm thick, with a strong focus on portability and one-handed usability.
In terms of hardware, the Clear7C is powered by an octa-core processor, paired with 4GB of RAM and 64GB of storage, and includes a 2700mAh battery. The company claims up to 120 hours of reading time. It runs on Android 14. Source
On April 23, Onyx Boox introduced the Leaf5+ monochrome e-ink e-reader. The device features a 7-inch Carta 1300 panel with a 300 PPI resolution, supports stylus input, and uses etched tempered glass for the display, along with physical page-turn buttons. The Leaf5+ is about 4mm thick and weighs 192g, and also includes a new light guide plate, with the company claiming up to 90% screen uniformity and a color temperature adjustment range of 3000K–5500K.
For hardware, it is equipped with a Qualcomm octa-core processor, 4GB LPDDR4X RAM, and 64GB of storage, along with a 2000mAh battery. The device is priced at RMB 1,749. Source
iFLYTEK unveils AINOTE2 e-ink device
On April 23, iFLYTEK unveiled the AINOTE2 e-ink device, described as an “ultra-thin e-ink tablet.” It features a 10.65-inch monochrome e-ink display with a resolution of 1920×2560 and a pixel density of 300 PPI, supporting handwriting input via the iFLYTEK I-Pencil electromagnetic stylus. The device uses a magnesium-aluminum alloy body, measuring about 4.2mm thick and weighing 295g.
In terms of hardware, the AINOTE2 comes with 4GB of RAM, 64GB of storage, and a 4000mAh battery. It runs on Android 14 and includes several custom AI features from iFLYTEK, such as content summarization, speech-to-text transcription, and integrated note-taking. The device is priced at RMB 3,999. Source
Panasonic releases LUMIX ZS300 camera
On April 23, Panasonic announced the LUMIX ZS300. The device features a 1-inch CMOS sensor with approximately 20.1 million effective pixels, paired with a Leica-certified 15× optical zoom lens (equivalent to 24–360mm). It supports a minimum focusing distance of 3 cm and includes 5-axis hybrid image stabilization. The camera is available in black and silver, with a built-in flash and 22 creative filters.
In terms of features, the LUMIX ZS300 supports 4K video recording, with a 4K Photo mode capable of 30 fps high-speed burst shooting, as well as FHD 120fps slow-motion recording. For connectivity, it supports Bluetooth and Wi-Fi for pairing with smartphones and transferring images, and also supports charging via a USB Type-C port. The product will officially go on sale at 12:00 on April 27, with a starting price of RMB 4,998. Source
Google introduces two AI chips: TPU 8t and 8i
At the Google Cloud Next 2026 conference, Google unveiled two new generations of custom AI ASIC chips: TPU 8t and TPU 8i, optimized for training and inference scenarios respectively, while also supporting cross-scenario workloads.
The TPU 8t is designed for large-scale, compute-intensive training tasks, focusing on improved compute throughput and vertical scaling bandwidth, delivering higher efficiency in complex model training. The TPU 8i, on the other hand, is optimized for low-latency inference and AI agent workloads, offering higher memory bandwidth and integrating a CAE (on-chip acceleration engine). It is equipped with 384 MB of on-chip SRAM and 288 GB of off-chip HBM, reducing data transfer overhead during execution and lowering latency.
At the cluster level, the TPU 8i Pod scale has increased from 256 chips in the previous generation to 1,152 chips. A single Pod delivers 8.67× higher FP8 compute performance and 5.74× greater HBM memory capacity. Google states that the overall energy efficiency of the new TPU 8t and TPU 8i is approximately double that of the previous-generation TPU Ironwood. Source
On April 23, Tencent Cloud released a security advisory regarding a supply chain poisoning risk affecting the open-source PyPI inference framework Xinference. The advisory states that affected versions may contain malicious code when installed or imported, potentially exfiltrating sensitive data—including cloud platform credentials, API keys, SSH keys, cryptocurrency wallet information, database credentials, and environment variables—to attacker-controlled command-and-control servers.
Tencent Cloud noted that such risks typically originate from tampering within the software supply chain and may pose significant threats to AI inference environments and related infrastructure. Affected users are advised to immediately uninstall compromised versions, scan systems for backdoors and residual files, and rotate critical credentials and keys as soon as possible. A comprehensive security audit is also recommended to assess potential impact and strengthen system defenses. Source
News Worth a Quick Look
X product lead Nikita Bier said that Premium subscribers on iOS can now early-access a new feature: users can pin specific topics on the “Home” feed, and Grok will personalize timeline recommendations based on those topics and past interactions. Powered by Grok’s understanding of platform content and personalization algorithms, the feature aims to deliver a highly customized feed experience, with an Android rollout expected soon. In addition, X announced it will shut down the Communities feature on May 6 due to a continued decline in user activity, and recommends migrating communities to group chats on XChat. Meanwhile, competing platforms like Threads and Mastodon continue to strengthen community-driven feeds. Source
Memory manufacturer Kioxia introduced the EG7 series SSDs on April 23. The lineup uses QLC NAND across all models and does not include a DRAM cache. Performance-wise, EG7 offers up to 7000 MB/s sequential read and 6200 MB/s sequential write speeds, with random performance reaching up to 1,000K IOPS. Capacities include 512GB, 1TB, and 2TB, with form factors in M.2 2230, 2242, and 2280. The series targets the consumer market and has already been supplied to multiple OEMs for use in thin-and-light laptops and desktops. Source
Google confirmed during its Cloud Next 2026 keynote on April 22 that its partnership with Apple stems from a multi-year agreement signed in January 2026. Under the deal, Google will act as the preferred cloud provider, delivering core technical support, with results expected to materialize by the end of 2026. Source
Rambus announced a chipset solution for LPDDR SOCAMM2 modules on April 22, designed to provide control, telemetry, and power management for JEDEC-compliant SOCAMM2 server memory modules. The goal is to meet the high stability and efficiency demands of AI servers. SOCAMM2 introduces a modular, replaceable design to replace traditional soldered LPDDR memory, combining LPDDR’s energy efficiency with improved serviceability and scalability for data centers. Rambus’ LPDDR5X SOCAMM2 chipset supports speeds up to 9.6 Gb/s, emphasizing reliability and efficiency under high bandwidth workloads. Source
According to Mark Gurman, Apple is developing six new product categories, including AI-powered AirPods, smart glasses, wearable clip-on devices, smart displays, desktop robots, and security cameras. Source
Foldable phones haven’t had an easy time over the past two years.
According to IDC’s annual report, China’s foldable smartphone shipments reached about 10.01 million units in 2025, a year-over-year growth of 9.2%, compared to 30.8% in 2024—growth nearly halved in just one year. Meanwhile, foldables still account for only around 2.5% of the global smartphone market. After seven years, the category remains stuck in a niche position.
IDC, Counterpoint, and Omdia all reached a similar conclusion in their year-end reports—the real restart for foldables won’t come until 2026. Their reasoning is also largely the same: new players entering the market, supply chain upgrades, and the adoption of new materials. In other words, the industry has already accepted one thing—the next breakthrough for foldables won’t come from the product itself, but from external forces.
I see it differently.
After years of using foldables on and off—never quite sticking with them—I’ve come to believe the real reason foldables haven’t broken through has nothing to do with specs or supply chains. It’s something more fundamental: every manufacturer is still building foldables with a “phone-first” mindset. The screens fold thinner and unfold larger, but the system is still a phone system, the interaction model is still phone-based, and even many apps and native features feel largely unchanged. The physical form of the screen has evolved, but the thinking behind it hasn’t.
But here’s the problem—once a screen unfolds to seven or eight inches, approaching the size of a small tablet, user expectations are no longer about “a bigger phone.” What we really want is something new: a device that lets us comfortably read long articles, sketch properly, edit videos on the couch, or experience AI in a more fluid, ambient way. These needs used to be fulfilled by tablets—but tablets are too large to carry in a pocket. Phones fit in your pocket but can’t deliver these experiences. Foldables were supposed to address exactly this in-between space.
Most foldables over the past seven years have been trying to solve a problem that may not have been the right one to begin with. They focus on “making phones bigger,” yet rarely question “how to fill this gap.” What the industry truly needs isn’t another foldable with better specs, but the first foldable that breaks away from phone-centric thinking. In many ways, Huawei Pura X Max rethinks this problem from the ground up.
This first-of-its-kind wide-format foldable doesn’t follow the tall, narrow proportions of traditional smartphones, nor does it unfold into an uninspiring square-like screen. Instead, it introduces a new “wide” aspect ratio—one that doesn’t belong to any previous category of device. According to Huawei, the ratio is derived from the common denominator of popular content formats, aligning with the √2:1 ratio of A4 paper. While that may sound mathematical, in practice it’s quite intuitive—4:3 images, 3:2 photos, and 16:9 videos all display in near-native proportions. As a result, many long-standing limitations and assumptions are broken.
For the first time, the evolution of the display returns to the content itself.
Put Agile, Efficient Content Consumption in Your Pocket
For a long time, foldable phone users have shared an unspoken habit: if you want to actually use the phone, you unfold it. The outer screen is often reserved for calls, QR codes, or checking the time—real usage only begins once the inner screen is opened.
But this habit isn’t something users naturally developed—it was conditioned. At its root is an assumption the entire industry has accepted for too long: the value of a foldable lies in having “a bigger screen.” Once that becomes the core selling point, product design inevitably follows the same direction—unfolded is the main stage, the outer screen is just a supporting role. The narrow outer display and increasingly thin bodies, despite claims of offering both “slab and foldable experiences,” are ultimately byproducts of this obsession. It’s not that they were poorly designed—it’s that they were never truly taken seriously in the first place.
More subtly, this design bias feeds back into user psychology. Foldables already differ from traditional phones in system interaction—split screen, multitasking, hover modes, layering—features that only fully shine when the inner display is open. Over time, users develop a quiet anxiety: if I bought a foldable but keep using it closed, am I wasting its potential? As a result, even simple tasks that could easily be handled on the outer screen—reading a slightly longer article or replying to a longer message—end up pushing users to unfold. No one forces you to do it, but you start to feel like you should.
This psychological burden—“it’s a waste if I don’t unfold it”—is more exhausting than any physical limitation. It adds unnecessary friction to the experience. Unfolding feels ceremonious; not unfolding feels wasteful. In a crowded subway, you can’t unfold it with one hand; in a coffee queue, unfolding feels over the top. Most of the time, you just want to get things done while the device is closed—and traditional foldables, both in design and in the mental model they impose, haven’t made that feel natural.
The Huawei Pura X Max takes a different approach: its outer and inner displays share nearly the same wide aspect ratio. It sounds simple, but the experience shift is significant. For the first time, the outer screen is treated as a fully capable primary interface—you no longer have to wait to unfold to “start using” the device, nor do you need to repeatedly decide whether you should unfold. The outer display itself is a wide screen that stands on its own. You unfold only when you want to, based on the situation—not out of habit.
High-Density Information in Fragmented Moments
Commuting, standing in line, waiting for someone—these short fragments of time are when we use our phones most often, and they’re exactly where the outer screen should take over.
Open a content app, and cards can be displayed in multiple columns on the wide outer screen of the Pura X Max, letting you scan several items at once. The essence of feed-based content is to browse more options in less time and quickly decide what to open. With more horizontal space, visual scanning becomes effortless. This level of information density used to belong only to large screens—now it’s available on a device you can hold in one hand.
Thanks to the “wide small screen” design, when folded, the device also has balanced weight distribution and fits naturally in your hands. For fundamental phone tasks—replying to messages, jotting down notes, frequent scrolling through social feeds—the experience is not only smooth but also lighter and more comfortable.
What these scenarios share is that they previously implied you needed to unfold the device for a “better” experience. Now, the outer screen delivers it directly. It’s no longer a transition—it’s enough on its own.
For the Moments That Truly Deserve It
And when a more immersive moment arrives, that’s when you unfold the inner screen.
Watching a movie on the couch over the weekend—this is the classic use case. Most films, short videos, variety shows, and sports broadcasts are shot in 3:2, 16:9, or wider formats, yet on traditional square-ish foldable displays, they leave large black bars. The Pura X Max’s wide inner screen better matches these formats, allowing your brain to naturally enter a “nothing beyond the screen matters” state. For the first time, the core logic of a cinema experience is fully realized on a phone.
Gaming is another scenario worth unfolding for. In titles like auto-battlers, the expanded layout feels much closer to a tablet or PC—wider boards, more spacious UI elements, and better visibility without constant switching. Seeing the full picture at once is the core need of such games—and one of the hardest things to achieve on smaller screens.
Reading a long article or flipping through an e-book—these are also moments worth unfolding for. You’re not opening the device to “justify the product,” but because the content itself deserves more space.
On the Pura X Max, folding and unfolding are no longer opposing modes—they’re simply two levels of content intensity. Quick, fragmented, one-handed tasks are handled perfectly by the outer screen; immersive, longer, more engaging experiences naturally call for the inner display. The device hands this choice back to the context itself.
This is the fundamental difference between a tablet mindset and a phone mindset. Under a phone mindset, the outer screen is subordinate to the inner one, and folding is a compromise. Under a tablet mindset, both screens are just different sizes of the same canvas—primary and secondary views flowing seamlessly between each other. What the Pura X Max achieves isn’t just making foldables “a bit better” for content consumption—it borrows from tablet design thinking to answer a long-unresolved need: combining the expansiveness of a tablet with the portability of a phone.
And Creation, Too
Once the screen gets bigger and the aspect ratio feels right, the first thing that changes isn’t just how you consume content—it’s how you create. In the past landscape of mobile devices, “doing” has always been assumed to belong to tablets and computers. Phones, constrained by smaller, narrower screens, have always been a compromise for any creative task that needs room to breathe—drawing, video editing, layout design. You could doodle a little on a phone, but for anything serious, you’d return to a tablet or a computer; you could cut together a short vertical clip on your phone, but longer horizontal videos meant switching devices.
This division—“consumption on phones, creation on tablets”—is itself a byproduct of phone-first thinking. It assumes the phone is a consumption terminal, while creation belongs elsewhere. But look at it another way: if a foldable phone already unfolds to something close to tablet size, why shouldn’t it take on the responsibilities of a tablet?
Huawei Pura X Max gives a serious answer to that question.
The most direct proof is that drawing apps finally make sense on a phone. Apps that were once exclusive productivity tools on tablets—offering watercolor, oil painting, even brush-like calligraphy strokes—now feel usable. In the past, trying to draw on a six-inch slab phone wasn’t limited by brush realism, but by the lack of space to actually draw.
With the wide display unfolded, the screen is large enough to accommodate a full canvas. Paired with the customized HUAWEI M-Pen 3 mini stylus, a sketchbook finally fits into your pocket. More importantly, the screen’s color accuracy and brightness are good enough to support creative feedback—you see the same soft watercolor tones outdoors as you do with your eyes, without needing to recheck them later on a professional device.
AI-powered creative tools further merge the pipeline between sketching and generation. By embedding AIGC capabilities directly into Huawei Notes, you can sketch on one side of the wide screen and instantly see AI-generated results on the other. A rough doodle and a simple prompt can quickly turn into a usable design draft. What once required multiple devices and apps is now compressed into a single screen.
For video creation, the focus shifts to editing. Vertical vlog editing adopts a side-by-side layout, while horizontal long-form editing uses a top-and-bottom layout—both offering significantly larger preview areas than traditional phones. The bottom toolbar is fully visible without dragging. A light press with the stylus brings up a radial menu, allowing quick cuts, speed adjustments, and subtitle additions in one flow. What used to be a rushed compromise on phones now approaches a desktop-level workflow.
Combining both “viewing” and “doing,” the purpose of the Pura X Max becomes clear—it enables both the consumption and creation strengths of a tablet to fully exist on a device that fits in your pocket. Under phone-first thinking, consumption and creation are separated; under tablet-first thinking, both coexist. The Pura X Max begins as a portable tablet, but its larger ambition is to define something entirely new.
Accompanied: An AI That Doesn’t Need to Be Summoned
The Pura X Max doesn’t just rethink the screen—it also rethinks interaction through its “ambient AI.” The question it answers is this: should interaction on this screen still follow the rules of traditional phones?
In recent years, nearly every manufacturer has talked about “AI phones.” But in practice, most AI still behaves like a summoned tool—you have to speak, tap, or remember how to activate it. AI on phones feels more like a silent utility than a constant companion.
This isn’t entirely the fault of manufacturers. Conversational AI only entered mainstream awareness three or four years ago. By then, the slab phone interaction model had already been established for nearly two decades—home screen, app icons, full-screen apps. It’s a deeply ingrained habit for billions of users, not something easily changed.
So when AI was introduced to phones, options were limited: package it as a standalone app, or as a floating window triggered by voice or a button. Either way, it’s squeezed into an existing framework—not because the screen lacks space, but because AI arrived too late to claim its own place.
Tablets, however, were different from the start. Split-screen multitasking, floating windows, parallel app workflows—these computer-like interaction patterns already exist naturally. When AI arrives on a tablet, it doesn’t need to be forced in; it has space to coexist. Reading an article while AI summarizes on the side, writing while AI refines text in real time—these feel natural.
The Pura X Max brings this tablet-style interaction into a phone form. When a foldable is redesigned with tablet thinking, something like a “persistent side AI panel” doesn’t need to be invented—it simply fits.
This is where Xiaoyi’s ambient AI comes in. It supports three modes: background presence, side-panel persistence, and window expansion. The most intuitive change is the side panel—AI no longer waits to be summoned. It stays alongside your main content, observing what’s on screen and proactively suggesting what you might want to do next.
A few moments that stood out to me.
A colleague posts in a Feishu group: “Meeting with the business team tomorrow at 10:30.” Xiaoyi sees it and adds it to your calendar automatically—no need to open the calendar app. While reading an article, if you encounter an unfamiliar term, simply highlight it and the explanation appears on the side. After finishing the article, a quick glance shows suggested follow-up topics. When opening a map app to go somewhere unfamiliar, you can also ask Xiaoyi on the side where to park, avoiding another search.
None of these ideas are entirely new in isolation. What matters is that they are proactively presented, rather than something you remember to look for later. Once the interaction direction flips, AI shifts from being a “tool” to an “environment”—not something you use, but something you exist within.
This isn’t something manufacturers never considered for slab phones—it’s that there was no room for it. Not physical screen space, but interaction space. The old framework simply doesn’t accommodate always-on, parallel AI without clashing with established habits. The Pura X Max’s tablet-first approach creates that space, allowing AI to settle naturally.
Conclusion
Back in 2010, before the first iPad was released, Steve Jobs asked a question on stage: is there room for a third category between phones and computers? At the time, it sounded rhetorical—but years later, it defined an entire product category. The iPad succeeded not because it was a “bigger iPhone” or a “more portable Mac,” but because it was given its own identity.
More than a decade later, the same question deserves to be asked again—is there a third category between phones and tablets?
That’s the strongest impression I had after using the Pura X Max. Throughout this piece, I’ve used the term “tablet thinking” as a way to explain why it feels different from other foldables. But that only tells part of the story. What the Pura X Max is really doing isn’t just applying tablet thinking to a foldable—it’s finding a place for foldables between phones and tablets.
It borrows from tablets—wide aspect ratios, parallel interaction models, persistent AI—but it is ultimately neither a portable tablet nor a bigger phone. It’s something new: a form made possible only because the screen can fold. It’s small enough to carry, large enough to use seriously, and interactive enough to let AI truly accompany the user. These three things used to belong to separate devices—the Pura X Max brings them together.
Foldables have existed for seven years, and the industry has been measuring them with old metrics—size, thinness, durability. But what determines whether a product endures is not how well it scores on old scales, but whether it makes us realize those scales are no longer sufficient. The Pura X Max breaks that ruler. It’s not just a better foldable—it’s the first to make us reconsider what a foldable should be.
Back to the question at the beginning—can the next breakthrough in foldables come from the product itself? I think the answer is already clear.
The first time I saw this plush bag at a Sanrio store in Beijing Hanguang Department Store, I didn’t feel a strong urge to buy it—I assumed I could easily find it online later. But when I decided I wanted one to celebrate Pompompurin’s 30th anniversary, I realized that most listings on Taobao priced in the hundreds were just AI-generated images, while those with real photos were cheap knockoffs costing only a few dozen yuan. Just as I was about to give up, a friend found this bag on the Misutabai official website in Hong Kong, which supports SF Express shipping to mainland China. There were only two or three left in stock, so I quickly placed an order while still hesitating. As a result, I received the bag on Pompompurin’s birthday.
The bag turned out to be even smaller than I expected—roughly the same size as the Chiikawa mini crossbody bags from Miniso. It can barely fit an iPhone 17 and definitely can’t fit my OPPO Find X9, but it easily holds a Chiikawa phone 🌝.
Compared to Miniso’s version, which uses a fabric strap sewn onto one side, this plush bag features a leather strap with metal lobster clasps on both ends. The zipper is also a metal yellow star, which feels much more refined than Miniso’s acrylic zipper. However, if you want to remove the strap and display it as a plush toy, you’ll need to stuff the bag to maintain a nice shape—unlike Miniso’s version, which holds its form even when empty.
If you want to identify an authentic version, you can look at other details. For example, the “Chiikawa × Sanrio Characters” logo on the back is made of leather, and the product tag is square-shaped, not round like many listings on Taobao.
This series also includes other combinations like Hachi × Cinnamoroll, Chiikawa × Hello Kitty, and Flying Squirrel × Kuromi. If you’re interested, it’s worth looking into. Finally, congratulations to Pompompurin for currently ranking No.1 in mainland China in the 2026 Sanrio Character Popularity Contest!
@路中南: Petlibro Gravity Dump Water Fountain 2S
Reference price: ¥200
I replaced the old Petkit water fountain at home for three main reasons:
There are two cats at home. The younger one loves to “wash its paws” in the fountain, which means the older one ends up drinking dirty water—leading to occasional stomach issues;
The Petkit fountain is too subtle—its indicator light isn’t very noticeable, and when things get busy, I often only realize it’s empty when the cats start searching for water;
Last year, I even cleaned out quite a bit of grime from the pump; and the replacement filters have become too expensive.
This Petlibro fountain works on a completely different principle compared to Petkit’s “filter the same tank of water repeatedly.” Instead, it divides water into portions and replaces “old water” in batches. Each cycle discards the previous water and dispenses fresh water, which is why it’s called a “dump-style” fountain.
That’s essentially its core function. The clean water tank holds about 3L, and since I have a direct drinking water purifier at home, refilling it is convenient. The wastewater tank at the bottom is basically ready-made “used water,” which I can pour straight into the fish tank—so nothing goes to waste, similar to how I reuse water from my robot vacuum. After running this fountain continuously for over three weeks, I’m overall quite satisfied:
It actually has a heating function. In the cold winters of Jiangsu and Zhejiang, my Guangdong-born cats aren’t very tolerant of low temperatures. The old Petkit fountain didn’t have heating, and drinking cold water could sometimes cause stomach discomfort or even vomiting. I haven’t been able to test this feature yet due to the current season, but its 8.5W heating power should be sufficient for a device of this size.
There are multiple ways to trigger water dispensing, and the output volume can be customized. If you like tinkering, you can create a highly customized “water schedule,” though I ultimately chose the simplest “interval dispensing” mode, along with a fallback low-water auto refill. The “motion-triggered dispensing” depends heavily on placement—its sensing range is so wide that sometimes it activates just because a cat or person walks by at a distance, making it overly sensitive.
The companion app looks clean and provides data on water output and pet drinking volume—something the old Petkit fountain didn’t offer.
Surprisingly, it includes a 4000mAh battery, which allows it to keep running for quite a while during power outages, and even makes it portable (USB-C charging).
Overall, getting a pet water fountain like this for around ¥200 is already quite impressive, though it’s not perfect. First, its smart home connectivity isn’t built into the device itself (likely for battery and power considerations). Instead, it comes with a USB-like gateway that needs to be paired separately and plugged into a power source before remote control via the app is possible. Second, the beeping sounds and the noise from the dumping mechanism can be quite noticeable at night. If you’re sensitive to noise while sleeping, you may want to customize the dispensing schedule.
On the very first day it arrived, the younger cat immediately gave it a “test run”—by washing its paws in it.
@Microhoo: DJI Pocket 4
Reference price: ¥2999
Back when the Pocket 3 went viral, I was tempted for a while but didn’t act. The main reason was that I felt its core significance was more about finally “filling the gap” in the all-in-one gimbal camera category after a couple of iterations—it gave media professionals and vlog creators a truly usable tool. The stabilization quality from a three-axis mechanical gimbal is something no smartphone OIS+EIS combination can replicate; the physical advantages of a 1-inch sensor in low light and dynamic range, along with the added “ritual” value compared to a phone, firmly established it as a capable video tool.
But its audience has always been those with clear, specific needs—its strengths require the right scenarios to shine. Without that context, its shortcomings become obvious. Especially at the time of its popularity, smartphone imaging also saw a sudden leap forward, so for me, it wasn’t as convenient as just using a phone. Still, I was very much looking forward to the next generation, because DJI usually understands what users want. If I had to sum up the Pocket 4 in one sentence—it finally addresses most of those “almost there” shortcomings.
Pocket 4 adds two physical buttons for zoom and customizable functions, while the rest of the design and operation remain nearly identical to its predecessor. The real focus is on improvements in image quality and performance. The most noticeable change is in still photography.
To be honest, the photo function on the Pocket 3 felt more like an emergency option. While the large sensor delivered excellent detail, its 9.4MP resolution lagged behind even many mid-range smartphones, making it feel outdated. With Pocket 4 upgraded to 37MP, its overall imaging quality now rivals dedicated cameras. Combined with highly usable built-in filters, beauty effects, and Live Photo output, even users without professional intent can get better photo and video results than most smartphones in many scenarios.
Photo straight out of camera
This generation also introduces a feature I find particularly interesting—an expansion accessory port on the back of the gimbal, supporting both data communication and power delivery. An official external fill light is already available, and when connected through this port, its power, color temperature matching, and brightness adjustments are all controlled directly by the camera, eliminating the need for separate batteries or manual parameter tuning.
It may sound like a small detail, but in complex lighting scenarios where you need to capture usable footage quickly, the reduction in operational burden is very noticeable. More importantly, the design of this interface suggests room for future expansion—if DJI continues building out an accessory ecosystem, the Pocket 4’s use cases could expand even further.
Slow motion is another major upgrade, and arguably the most impressive performance improvement. While the Pocket 3’s 4K/120fps was already top-tier in its class, the Pocket 4 pushes this to 4K/240fps—roughly equivalent to 10× slow motion compared to standard 24fps video—entering territory previously reserved for professional cameras. After shooting a few clips, the ability to stretch fleeting moments into something cinematic feels far beyond simply “slowing things down.”
240fps slow-motion video still frame
Comparing frames from 240fps video to still photos, while there is some drop in dynamic range and resolution, the results remain highly usable—it’s hard to believe such performance comes from a device this compact.
Storage and transfer are often overlooked but critical to workflow efficiency. In an era where storage is precious, the Pocket 4’s 107GB of built-in storage eliminates the need for additional memory cards for most users. Wired USB 3.1 and wireless Wi-Fi 6 also significantly reduce waiting time when transferring footage. The dynamic range has improved from 12 stops to 14 stops, color format upgraded from D-Log M to full D-Log, and audio now supports four-channel spatial recording with independent channel control in post—features that benefit more advanced creators.
Overall, with these combined improvements, I feel the Pocket 4 has evolved from a “specialized tool for a niche audience” into a device that many more people can pick up and use directly. It retains the core advantages of the Pocket 3 as an all-in-one gimbal camera, while no longer requiring users to compromise as much for its shortcomings. This year feels like a turning point for this category, with multiple brands preparing to launch competing products. While it’s hard to say whether the Pocket 4 has an absolute advantage, given DJI’s experience and track record in handheld imaging, it will likely remain the most broadly appealing option.
That said, competition and pressure are real. I see the Pocket 4 as a refinement and reflection after the success of its predecessor, while the briefly teased dual-camera version may be the more compelling answer for professional creators who loved the earlier models. I’m looking forward to it.
@Clyde: Tolek Punk Power “Famicom” Power Strip
Reference price: ¥249
A few years ago, before everyone at the office started hitting middle age, we all somehow ended up buying those tall, boxy power strip organizers from Bull—great for hiding a bunch of plugs on the desk. They did solve the clutter problem, but after sitting there for a while, they felt like a waste of space—and more importantly, they just didn’t look good. So when someone at home needed a power strip, I generously gave mine away.
Then I bought this Tolek punk-style “Famicom” power strip.
Compared to the brick-style modular one we sell, this “Punk Power Famicom” version doesn’t have as many tricks. If you’re nostalgic, you might be drawn in by the “Famicom” design before buying—but after placing the order, it’s best to reset your expectations and just treat it as what it really is: a power strip. The device has three buttons, all centered around the front display, which looks large but actually only occupies about 40% of the width on the left side. The big red button on top turns the display on and off, while the two smaller round buttons control the display zones—left button controls the right side, and right button controls the left side. Yes, it’s counterintuitive. If anyone from Tolek is reading this, feel free to explain the logic.
The display shows some basic runtime information, including a colored bar at the top indicating current power load, and real-time current and voltage readings below. For a power strip, it’s more of a visual gimmick—you’re unlikely to rely on it for diagnosing real issues. The numbers exist more to serve “desk aesthetics” than functionality. The left side of the display is even more so—you can toggle between a smiling face, a crying face, and a looping Tetris animation. None of them are particularly sophisticated or emotionally engaging. Before buying, I even imagined it might support firmware flashing or custom display content—now I think I must’ve been drunk.
So again, keep your expectations grounded: it’s a power strip. That said, the color scheme is classic, and the design has a retro charm. Writing this reminded me of my old Keychron Q1 Pro (SSPAI limited edition) with a broken key. Putting them side by side actually looks pretty fitting:
Finally, one reason I didn’t just ask a colleague for the Pixel 65 modular power strip was that my current desktop power needs are quite high. Even though this model has four rear outlets that support both two- and three-prong plugs, in practice, four still feels limiting. The two USB-C ports and one USB-A port on the front handle high-frequency charging tasks, but that’s about all it can manage. Compared to this, the three-outlet modular version would likely be even less practical.
One last responsible reminder: this unusual design does come with balance and weight distribution issues. If you’re plugging in standard cables, it’s fine—but bulkier power adapters can cause problems. You’ll need to make sure the cables at the back don’t constantly pull it backward, or it might shift around. Honestly, it would’ve been great if they had added suction cups to the bottom.
@waychane: Aqara FP310 Presence Sensor
Reference price: ¥249
Good-looking design and flexible installation—that’s my overall impression of Aqara’s recently released AI multi-dimensional presence sensor FP310. As part of the “infrastructure” of a smart home, I generally believe you can never have too many sensors, so I picked one up and installed it in the laundry room.
In terms of design, the FP310 combines elements of traditional motion sensors with the form factor of the earlier FP2 presence sensor. The sensing unit has a slightly cylindrical shape, with an adjustable-angle base at the bottom, and the overall size is roughly comparable to AirPods Pro.
Since it is battery-powered, the FP310 isn’t restricted by power outlet placement, making installation highly flexible. More notably, it introduces magnetic mounting for the first time—the base contains built-in magnets, allowing it to attach directly to metal surfaces. If you don’t have a suitable magnetic surface, that’s not a problem either: the FP310 comes with a magnetic plate that can be attached to a wall using adhesive, allowing you to mount the sensor onto it.
Included magnetic plate
Functionally, the FP310 uses the same millimeter-wave radar detection as the FP2, making it more precise than traditional motion sensors. It can accurately detect presence even when a person is stationary. In the Aqara app, you can adjust sensitivity based on room size, set the timeout for detecting absence, and configure the radar detection range (0–6 meters). However, it does not support zone-based detection and is limited to single-area monitoring.
In addition, the FP310 integrates temperature, humidity, and ambient light sensors, enabling more advanced automation scenarios—for example, controlling air conditioning based on temperature or turning lights on/off based on brightness. The Aqara app also allows you to adjust the sampling frequency of these sensors or disable them entirely to extend battery life.
In terms of connectivity, the FP310 supports both Zigbee and Thread (Matter), with the latter expanding its compatibility across platforms. Since I only needed integration with Apple Home, I connected it via Zigbee.
As for buying advice, given that the FP310 is battery-powered and lacks zone detection, I wouldn’t recommend installing it in large spaces like living or dining rooms where people stay for extended periods. Instead, it’s better suited for smaller areas such as kitchens, bathrooms, or laundry rooms, where its multi-sensor capabilities can be effectively combined with devices like lights or bathroom heaters for automation.
On April 22, Huawei’s Harmony Intelligent Mobility Alliance held its spring launch event, unveiling several new and updated vehicle models.
Among them, the AITO M6 is positioned as a next-generation smart SUV, with a starting price of RMB 259,800. It comes standard with a 896-line LiDAR system, offers seven exterior color options and multiple interior configurations, and measures 4960×1985×1736 mm with a 2950 mm wheelbase and a drag coefficient of 0.239Cd.
Inside, the vehicle features newly designed ergonomic seats, with zero-gravity mode for the front row and ventilation, heating, and power adjustment for the second row, along with a 17.3-inch entertainment screen. The pure electric version is equipped with a 100 kWh battery pack, offering up to 760 km CLTC range, while the extended-range version delivers up to 1,605 km combined range. The entire lineup includes Huawei ADS 4.1 intelligent driving system, an advanced chassis system, and multiple structural safety designs, along with 35 storage compartments and a 202L front trunk. Source
Product appearance images, from the original source
The Z7 and Z7T were officially launched, with starting prices of RMB 219,800 and RMB 229,800, respectively. The new models offer various color, wheel, and interior combinations, measuring 5036×1976×1465 mm with a 3000 mm wheelbase. They feature a floating display, Harmony ALPS Health Cabin 2.0, and zero-gravity seats for both driver and passenger.
Both models are built on Huawei’s Whale 800V battery platform, offering single- and dual-motor configurations, with a top speed of 242 km/h and a maximum range of 905 km. They come standard with dual-path 896-line LiDAR and a full-dimension collision avoidance 4.0 system, supporting multi-scenario perception and obstacle avoidance. The trunk capacity can expand up to 1694L and is complemented by a front trunk and additional storage solutions. Source
Product appearance images, from the original source
The Zhijie V9 was announced for pre-orders at the spring launch event, with a starting price of RMB 399,800. Customers can place a RMB 5,000 deposit to receive a RMB 10,000 discount on the final payment.
This model is built on Huawei’s Tuling MPV platform, featuring a double-wishbone front suspension and H-arm multi-link rear suspension, with rear-wheel steering standard across all versions and a minimum turning radius of 5.35 meters. Inside, it includes a 12.3-inch instrument panel, dual 17.2-inch central displays, and a 21.4-inch ceiling-mounted screen, along with full-scenario NFC, acoustic glass, and independent audio zones. The second-row zero-gravity seats support multiple modes, and the cabin offers 52 storage spaces. The vehicle also features a constant-temperature smart refrigerator and an onboard oxygen generator with multi-zone adjustment.
In terms of safety, all models come standard with 13 airbags and reinforced structural designs. The first batch of display vehicles has already arrived at dealerships. Source
The flagship luxury SUV AITO M9 has also officially opened for reservations, offering a standard version and an Ultimate extended version, with starting pre-sale prices of RMB 499,800 and RMB 669,800, respectively.
The standard version measures 5285/2026/1845 mm with a 3125 mm wheelbase, while the extended version measures 5402/2026/1845 mm with a 3236 mm wheelbase.
The new model offers 4-, 5-, and 6-seat configurations and supports a 360° cinematic viewing system. It is built on the latest Tuling fully active chassis platform, featuring a Gemini full-redundancy architecture with over 30 redundancy designs. The powertrain uses Huawei’s dual silicon carbide electric drive system, delivering over 900 horsepower, and debuts the Huawei Whale battery platform.
For intelligent driving, it is equipped with a six-LiDAR matrix and ADS 5 system, along with full-color smart light projection, Vehicle-to-Environment communication system 3.0, and integrated “space-ground-network” connectivity. Source
Product appearance images, from Harmony Intelligent Mobility
Framework introduces a range of new products
On April 22, modular computer maker Framework unveiled several new hardware products.
The new Framework Laptop 13 Pro continues the company’s modular design philosophy, offering full Linux support (with a pre-installed Ubuntu option) and positioning itself as a “MacBook Pro for Linux users.” It is powered by third-generation Intel Core Ultra “Panther Lake” processors, with options including Ultra 5 325, Ultra X7 358H, and Ultra X9 388H (AMD Ryzen AI 300 is also available). The latter two Intel configurations support LPDDR5X only, and to maintain memory upgradability, Framework has introduced LPCAMM2 memory modules running at 7467 MT/s. The Laptop 13 Pro also features a PCIe Gen5 M.2 2280 SSD slot and a performance-optimized 13.5-inch 2880 × 1920 LTPS matte touchscreen with a 30–120Hz refresh rate, 100% sRGB coverage, and up to 700 nits brightness. It comes with a full CNC aluminum chassis, a 1.5mm travel keyboard, a haptic touchpad, side-mounted Dolby Atmos speakers, a 74Wh battery rated for 1,000 cycles, and a 100W GaN power adapter. Pricing starts at $1,199 for the base configuration. Source
Product appearance images, from the original source
Another release is the OCuLink 8i development kit for the Framework Laptop 16. This kit converts PCIe Gen4 ×8 DGFF to a standard OCuLink 8i interface, enabling high-speed external PCIe devices (such as external GPUs) without the performance loss typically associated with protocols like Thunderbolt. As a result, the Framework Laptop 16 becomes the first laptop to support OCuLink 8i. The development kit includes an OCuLink adapter board, a graphics module base, and a PCIe OCuLink dock, allowing users to install various expansion cards and reuse existing Framework discrete GPU modules.
In addition, Framework previewed a wireless keyboard with an integrated touchpad, developed in collaboration with its long-term partner Lite-On. This device combines keyboard and cursor input in a single unit. The keyboard features a slim yet durable design, with a “transparent exploration” bottom panel. It offers 19mm key spacing and 1.5mm key travel, along with a 68.8 × 85.6 mm precision touchpad supporting multi-touch gestures. Powered by the Nordic nRF54LM20A controller and running open-source ZMK firmware, it supports USB-C wired connections as well as Bluetooth and 2.4GHz wireless connectivity for up to four devices. Source
Product appearance images, from the original source
Hisense launches the Xiaomo E5S TV series
On April 22, Hisense unveiled the Xiaomo E5S series TVs, including the RGB Mini LED–focused Xiaomo E5S Pro and the upgraded Mini LED model Xiaomo E5S. Both models bring notable improvements in display, local dimming, anti-reflection, audio, and smart interaction capabilities.
Among them, the Xiaomo E5S Pro adopts RGB Mini LED display technology, supporting direct output of RGB primary colors. According to the company, it achieves 100% BT.2020 color gamut, 100% color purity, and 89% color volume, while reducing harmful blue light wavelengths by 42% at the source and lowering overall power consumption by 30%. The model is equipped with up to 4,680 local dimming zones, reaches XDR Pro brightness of 3000 nits, and features the Hi-View AI image processor, supporting 3D light-color control, natural halo optimization, and near-4K reconstruction for low-resolution content.
For the display, the Xiaomo E5S Pro uses the Ink Crystal Screen 2.0, with a reflectivity as low as 2.6%, supporting native 4K at 180Hz and a 330Hz interpolation mode, along with HDMI 2.1 support. The audio system is co-developed with Devialet, featuring a 2.1.2 channel setup with peak power of 165W and a built-in subwoofer. The Xiaomo E5S model, meanwhile, includes 1,560 local dimming zones and up to 4200 nits peak brightness, while also featuring the Ink Crystal Screen 2.0, Hi-View AI processor, Devialet audio, and native 4K 180Hz refresh rate.
Product appearance images, from Hisense
In terms of pricing, the Xiaomo E5S Pro is priced (after subsidies) at RMB 4,499 for 55-inch, RMB 5,999 for 65-inch, RMB 7,799 for 75-inch, RMB 8,999 for 85-inch, and RMB 14,999 for 100-inch models. The Xiaomo E5S is priced (after subsidies) at RMB 3,999 for 55-inch, RMB 4,999 for 65-inch, RMB 6,499 for 75-inch, RMB 7,799 for 85-inch, and RMB 12,999 for 100-inch models. Both series are eligible for trade-in subsidies and launch-period benefits such as installation and warranty services. Source
Razer releases the Atlas Pro glass mousepad
On April 22, Razer announced the Atlas Pro glass gaming mousepad. This product reduces the overall thickness of a glass mousepad to under 2mm for the first time, with the glass layer measuring 1.1mm. The Atlas Pro measures 500mm × 400mm, features a full anti-slip rubber base, and CNC-machined rounded edges. Its glass surface offers low friction, is coated with a dirt-resistant protective layer, and reaches a Mohs hardness of 9H. With 2μm micro-texture etching, it is optimized for precise tracking by optical mouse sensors. The product is priced at RMB 1,199. Source
Product appearance images, from the original source
SpaceX gains priority rights to acquire Cursor
Recently, SpaceX, AI company xAI, and social platform X are preparing for a combined initial public offering (IPO), while also working closely with Cursor to build a world-class AI platform for coding and knowledge work. As part of the collaboration, Cursor will grant SpaceX the right to acquire the company later this year at a valuation of $60 billion, or alternatively receive $10 billion for continued cooperation. Source
Amazon invests an additional $5 billion in Anthropic and secures computing capacity supply
Microsoft lowers pricing for some Xbox Game Pass plans
News Worth a Quick Look
OPPO Find X9 Ultra officially unveiled
On April 21, OPPO unveiled the OPPO Find X9 Ultra, powered by the Snapdragon 8 Gen 5 Elite mobile platform, paired with LPDDR5X memory and UFS 4.1 storage. It features a 6.82-inch QHD+ flat display with a resolution of 3168×1440, supporting up to a 144Hz refresh rate, with 800 nits typical brightness and 1800 nits peak brightness. The camera system includes a 200MP wide-angle lens, 50MP ultra-wide lens, 200MP telephoto lens, and 50MP periscope telephoto lens, along with a 50MP front-facing camera. It is equipped with a 7050mAh battery, supporting 100W wired fast charging and 50W wireless charging. The device comes in three color options—Velvet Canyon, Polar Glacier, and Earth Tundra—with pricing starting at RMB 7,499 (12+256GB). Source
Beats releases Solo 4 JENNIE Special Edition and a 3-meter USB-C cable
On April 21, Beats introduced the Onyx Black JENNIE Special Edition Beats Solo 4. The product features a minimalist monochrome design, includes two detachable black bow accessories, and has a custom music note engraving on one UltraPlush ear cushion, along with a matching carrying case. It will go on sale at 9 AM on April 24, priced at RMB 1,799, available via Apple.com and the Beats official store on JD.com. Source
At the same time, Beats also announced a new addition to its accessories lineup: a 3-meter USB-C to USB-C cable. It features a tangle-free braided design with reinforced internals, supports up to 240W charging, and can handle data transfer, audio output, and CarPlay connectivity. The cable is compatible with USB-C devices across both Apple and Android ecosystems, and comes in four colors: Bolt Black, Surge Gray, Speed Blue, and Rush Red. It is now available in over 50 countries and regions worldwide, with orders open via apple.com.cn at a price of RMB 229. Source
ChatGPT Images 2.0 launched by OpenAI
On April 21, OpenAI released the ChatGPT Images 2.0 model along with the gpt-image-2 API. The model brings significant improvements in instruction following, visual composition, and text rendering. It supports multilingual text generation, including Chinese, Japanese, Korean, Hindi, and Bengali. Supported aspect ratios range from 3:1 to 1:3, with up to 2K resolution available via the API. The model’s knowledge cutoff is December 2025. When “thinking” mode is enabled, it can perform web searches and includes self-verification capabilities, and can generate up to 8 coherent images in a single request.
Yes, these are official sample images.
The model is now available to all ChatGPT and Codex users, while advanced outputs with “thinking” capabilities are limited to ChatGPT Plus, Pro, and Business subscribers. The gpt-image-2 API is also live, with pricing depending on selected image quality and resolution. Source
CATL introduces Shenxing fast-charging and multiple new power batteries
On April 21, CATL announced its third-generation Shenxing supercharging battery, third-generation Qilin battery, and an integrated charging-and-swapping energy solution. The new Shenxing battery supports up to 10C equivalent and peak 15C ultra-fast charging, reaching full charge in just 6 minutes under standard conditions. The third-generation Qilin battery achieves an energy density of 280Wh/kg and supports up to 1,000 km range, featuring thermal-electric separation safety technology. The lineup also includes the Qilin condensed-state battery, second-generation Xiaoyao hybrid battery, and sodium-ion batteries, with the latter scheduled for mass production in Q4 2026. By the end of 2026, CATL plans to build 4,000 integrated charging and swapping stations for passenger vehicles and heavy trucks, enabling flexible charging, battery swapping, and on-demand power distribution. Source
On April 20, GitHub Copilot announced adjustments to its individual subscription plans. Effective immediately, new user sign-ups for Pro, Pro+, and Student plans are suspended. The usage limits for the Pro+ plan have been increased to more than five times that of the Pro plan, with specific limits viewable in real time via VS Code and Copilot CLI. In terms of model access, the Pro plan will no longer support the Opus series models, while the Pro+ plan will retain only Opus 4.7, removing Opus 4.5 and Opus 4.6. Affected users may cancel their subscriptions via the billing settings page before May 20 and request a refund corresponding to the remaining subscription period. Source
Amazon invests an additional $5 billion in Anthropic and secures computing capacity supply
On April 21, Amazon announced an additional $5 billion investment in Anthropic and signed a long-term compute supply agreement. This round brings Amazon’s total direct investment in Anthropic to $13 billion. Under the agreement, Amazon may invest an additional $20 billion if certain business milestones are achieved. Anthropic plans to spend over $100 billion on Amazon Web Services over the next decade to procure custom AI chips, including Graviton, Trainium2, and Trainium4, addressing infrastructure demands driven by the growth of Claude users. The two companies aim to deliver 1 gigawatt of compute capacity by the end of 2026, with a long-term supply cap of 5 gigawatts. Source
Microsoft lowers pricing for some Xbox Game Pass plans
On April 22, Microsoft announced price reductions for certain Xbox Game Pass plans. The monthly fee for Game Pass Ultimate will drop from $29.99 to $22.99, while the PC Game Pass subscription will decrease from $16.49 to $13.99. At the same time, starting in 2026, future installments of the Call of Duty series will no longer be included in Game Pass Ultimate or PC Game Pass at launch. Instead, they will be added during the following holiday season (approximately one year later). Existing Call of Duty titles already in the library will remain unaffected. Source
News Worth a Quick Look
OpenAI has rolled out a Chronicle research preview feature for the macOS version of Codex, available to Pro subscribers. This feature uses background agents to periodically capture screen content and generate a local memory base, enhancing the model’s contextual understanding of the current task environment—such as on-screen error messages, open documents, and past workflows. The stored memory can be reviewed and edited by users. Due to its background operation, enabling the feature may increase rate limit consumption. It is now available to Pro subscribers. Source
Google Labs has open-sourced the DESIGN.md specification from its Stitch platform. This format provides AI agents with a unified, machine-readable design guideline—including brand colors, typography rules, component styles, layout logic, and interaction patterns—allowing AI to generate UI across projects while maintaining consistency with brand design standards. Source
