Article title: Morning Brief: Google Launches Gemini 4 Neon Model, NetEase UU Remote to Launch Paid Membership in December, and More
Main text:
Google Launches Gemini 4 Neon Model
On November 18, Google published the Gemini 4 Neon AI model on its official blog, aimed mainly at complex tasks such as software development, enterprise data analysis, and scientific research. The new model supports a 1-million-token context window, can comprehensively process data in multiple formats including spreadsheets, PDFs, charts, audio, and video, and generates structured results.
On the programming side, Gemini 4 Neon has been optimized for long-running agent tasks and can perform cross-file debugging, code refactoring, and complex project development. In the Google Antigravity development tool, Neon can also serve as a task coordinator, responsible for planning tasks and dispatching multiple Gemini 4 Hydrogen sub-agents to execute in parallel, improving the efficiency of handling complex development work.
Google also introduced Gemini 4 Neon for cyber, specifically aimed at cybersecurity defense, which can be used for code vulnerability discovery, risk verification, and patch remediation. This version is planned to be gradually opened to eligible cybersecurity defenders through the Fairwind Program within days of release.
Gemini 4 Neon began gradual rollout on the day of release. Developers can use it through the paid tier of the Gemini API in Google AI Studio, and enterprise users can access it through Gemini for business. Google AI Pro and Ultra subscribers can also select this model in Google Antigravity and the Gemini app.
In terms of API pricing, Gemini 4 Neon charges $3 per million input tokens and $15 per million output tokens, with cached input priced at $0.15 per million tokens. Source

NetEase UU Remote to Launch Paid Membership in December
On November 19, NetEase UU Remote released an announcement announcing that it will officially launch two paid membership tiers, VIP and SVIP, on December 1, 2026, offering advanced features such as higher image quality, frame rates, and multi-screen control. The company also promised that the basic remote control service will remain free, with no limit on the number of connections or usage duration, no queuing, and no ads.
According to the published membership benefits, the free version supports remote images up to 1080p at 30 frames and continues to provide basic features such as file transfer, remote wake-up, direct microphone connection, privacy screensaver, port mapping, and terminal tools. VIP supports images up to 4K at 144 frames, a 500Mbps bitrate, HDR, multi-screen control, wireless secondary screen, and high-speed relay, priced at 30 yuan per month, 24 yuan per month with continuous monthly billing, and 238 yuan for an annual card. SVIP adds global remote control nodes on top of this, priced at 98 yuan per month, 88 yuan per month with continuous monthly billing, and 758 yuan for an annual card.
To give back to existing users, NetEase UU Remote also announced that users who successfully registered before December 1, 2026 can receive a 50% discount on the VIP annual card, at a discounted price of 119 yuan, equivalent to about 9.9 yuan per month. Eligible users can find the "Old Friend 50% Discount Benefit" entry in the client's "Message Center," and after claiming it, view the usage instructions through "My Coupons."
In addition, NetEase UU Remote revealed that HarmonyOS and Linux clients will also be launched later, but no specific time has been announced. Source


Google Play to Mandate Foldable Screen Compatibility for Apps
Google announced on the official Android Developers blog on November 18 that it will implement new Google Play adaptive app quality requirements starting in April 2027, mandating that ordinary apps meet at least the Tier 3 basic standard in the Android foldable screen compatibility guidelines. Apps that do not meet the requirements will not be able to be published or updated on Google Play and may face reduced store visibility, but existing apps that have not been updated can still be downloaded.
Under the Tier 3 standard, apps must fully utilize the available display area and must not rely on black bars or compatibility mode to run. When the device rotates, folds, unfolds, or the window size changes, apps must also preserve scroll position, entered text, and media playback progress, and ensure normal use in multi-window mode.

At the same time, starting in April 2027, in addition to the existing 2 phone screenshots, developers must also submit 3 foldable screen screenshots to Google Play, covering the unfolded inner screen in 1:1 and 4:3 ratios, as well as the folded outer screen in a 10:16 ratio. The relevant screenshots can be obtained through physical devices or the Android emulator, and Google Play Console has currently opened uploads.

Google also plans to require newly published and updated apps to target Android 17 (API 37) starting in August 2027, further eliminating developers' exemptions from restrictions on screen orientation and window resizing for large-screen devices. At the same time, games will adopt a separate large-screen compatibility standard, and its Tier 3 minimum requirements will also officially take effect in August 2027. Source
AMD Launches Instinct MI450X AI Accelerator
On November 12, AMD announced the launch of the Instinct MI450X GPU accelerator, further expanding the Instinct MI400 series product line. The new product is aimed at the AMD Helios rack-scale AI computing platform and, while maintaining the same computing performance and memory bandwidth as the MI455X, reduces HBM4 memory capacity from 432GB to 288GB, a reduction of about one third.

According to the specifications published by AMD, the MI450X uses the fifth-generation CDNA architecture, is equipped with 256 workgroup processors (WGP) and 8 HBM4 memory stacks, and has a peak memory bandwidth of 23.3TB/s. Its FP4 peak computing performance reaches 40.3 PFLOPS, consistent with the MI455X. The interconnect capabilities of the two products are also the same, both supporting up to 3.6TB/s scale-up bandwidth and 600GB/s scale-out bandwidth.
AMD said the MI450X is mainly aimed at AI workloads requiring no more than 288GB of memory, designed to help customers optimize operating costs while meeting throughput and response time requirements. For customers who need to run larger models, longer contexts, or higher-concurrency tasks, the MI455X provides more memory space. Both accelerators support the Helios rack-scale architecture and the ROCm software platform, allowing customers to choose different memory configurations according to actual deployment needs. Source
Nginx Reportedly Exposed to 5 Remote Code Execution Vulnerabilities
On November 18, the security research team Xint Security Research publicly disclosed on its official website a vulnerability attack chain named NGINX2Root, involving 5 Nginx remote code execution (RCE) vulnerabilities and 1 Linux kernel SCTP subsystem privilege escalation vulnerability. Under specific configuration conditions, attackers can obtain code execution permission in the Nginx worker process without authentication; further combined with the kernel vulnerability, they may also break through Docker container isolation and obtain root privileges on the host machine. Xint has successfully demonstrated this attack chain in a test environment.
According to Xint's technical report, the 5 Nginx vulnerabilities involve functions such as HTTP/2, HTTP/3, reverse proxy, and mail proxy, and are independent attack paths under different configurations. Among them, the more broadly impactful CVE-2026-123007 originates from an integer overflow issue in HTTP/2 flow control, which can lead to out-of-bounds read and write of heap memory. This vulnerability has existed since Nginx 1.9.5, released in 2015.
In response to these vulnerabilities, Nginx maintainer F5 has released security updates. Both the Nginx Open Source 1.30.7 stable version and the 1.31.9 mainline version include fixes, and Nginx Plus has also released corresponding fixed versions. For administrators who are temporarily unable to upgrade, disabling HTTP/2 can mitigate the risk of the two vulnerabilities CVE-2026-123007 and CVE-2026-123633, but cannot protect against the remaining three Nginx vulnerabilities. Source
Microsoft Announces Microsoft To Do Will Be Fully Discontinued in June Next Year
Microsoft announced on its official blog on November 18 that it will officially discontinue its to-do app Microsoft To Do on June 30, 2027, covering Windows, macOS, iOS, Android, and the web. Microsoft plans to integrate personal task management features into Microsoft Planner, but personal Microsoft accounts cannot use this alternative and will need to choose another to-do app.
According to Microsoft's published schedule, Planner for work and school accounts will begin rolling out a redesigned "My tasks" interface in December 2026, supporting centralized management of personal tasks and assigned work tasks. However, existing Microsoft To Do tasks and lists will not automatically migrate to new personal plans in Planner, and existing tasks can only be viewed through Planner until June 30, 2027. Microsoft said it will publish data export guidance later, including supported data types, specific requirements, and limitations.

The history of Microsoft To Do can be traced back to Wunderlist, launched by German developer 6Wunderkinder. Microsoft acquired the company in June 2015, with media reports at the time estimating the deal at $100 million to $200 million. Subsequently, Microsoft To Do, developed with the participation of the original Wunderlist team, launched in preview in April 2017 and gradually replaced Wunderlist, which ceased service in 2020. Source

MSI EdgeMesa N AI+ mini PC officially launches
On November 18, MSI launched the EdgeMesa N AI+ mini AI PC in the Chinese market, equipped with the NVIDIA RTX Spark N1X superchip and offered in two configurations: 64GB unified memory with a 1TB SSD, and 128GB unified memory with a 4TB SSD, priced at 33,999 yuan and 56,999 yuan respectively.
The EdgeMesa N AI+ integrates a 20-core NVIDIA Grace CPU and a Blackwell RTX GPU with 6,144 CUDA cores, delivering up to 1 PFLOPS of FP4 AI compute, targeting applications such as local AI model inference, agent development, and content creation. Its chassis measures 152 × 147 × 53mm and uses an aluminum alloy shell with an anti-fingerprint coating. Inside, it features an AI Pulse cooling system consisting of a vapor chamber and dual turbo fans, supporting sustained operation at 140W TDP, with officially rated full-load noise of 32dB.

For expansion, the EdgeMesa N AI+ provides one PCIe 5.0 ×4 M.2 SSD slot and is equipped with an AzureWave AW-EM637 wireless card supporting Wi-Fi 7 and Bluetooth 5.4. The chassis has four USB-C ports at 20Gbps, three of which support DisplayPort video output, plus one HDMI 2.1 FRL port and one 10GbE RJ-45 port, supporting up to four displays simultaneously. The device runs Windows 11 and comes with a 240W external power adapter. Source
Other news worth a glance
- Huawei announced via its official Weibo that it will launch the Mate 90 Air phone and WATCH Air smartwatch at the HUAWEI Mate X8 and all-scenario new product launch event at 14:30 on November 24. Both new products use slim and light design as their main selling point; the Mate 90 Air is positioned as Huawei's thinnest Mate bar phone, while the WATCH Air sets a new slimness record for Huawei smartwatches with a 6.9mm body thickness. According to Huawei's published teaser information, the WATCH Air features a high-gloss titanium alloy unibody, sapphire glass crystal, and 1.4mm four-sided equal-width screen bezels, offered in three colorways. According to leaks from digital bloggers, the watch will also feature a 1.92-inch AMOLED square screen using LTPO 2.0 technology, with a screen-to-body ratio of 86%, peak brightness of 3,500 nits, and a watch body weight of about 27 grams. It is expected to house a high-silicon stacked special-shaped battery with up to one week of battery life, while health and sports features remain consistent with the WATCH 6 and GT 7 Pro. As for the Mate 90 Air, Huawei's online store has announced three colorways—Rose Gold, Moon Silver, and Obsidian Black—and two storage configurations: 12GB+256GB and 12GB+512GB. According to leaks, the new phone is expected to use a 7-inch quad-curved screen with equal depth, a centered circular camera module on the back with a triple-camera full-focal-length Red Maple imaging system, a midframe without injection-molded break points, a thickness in the 6mm range, and a 66W charger and side-mounted fingerprint recognition. Source
- According to Huawei's published teaser information, the WATCH Air features a high-gloss titanium alloy unibody, sapphire glass crystal, and 1.4mm four-sided equal-width screen bezels, offered in three colorways. According to leaks from digital bloggers, the watch will also feature a 1.92-inch AMOLED square screen using LTPO 2.0 technology, with a screen-to-body ratio of 86%, peak brightness of 3,500 nits, and a watch body weight of about 27 grams. It is expected to house a high-silicon stacked special-shaped battery with up to one week of battery life, while health and sports features remain consistent with the WATCH 6 and GT 7 Pro.
- As for the Mate 90 Air, Huawei's online store has announced three colorways—Rose Gold, Moon Silver, and Obsidian Black—and two storage configurations: 12GB+256GB and 12GB+512GB. According to leaks, the new phone is expected to use a 7-inch quad-curved screen with equal depth, a centered circular camera module on the back with a triple-camera full-focal-length Red Maple imaging system, a midframe without injection-molded break points, a thickness in the 6mm range, and a 66W charger and side-mounted fingerprint recognition. Source
Recent updates from SSPAI
- We have opened a dedicated page summarizing the awards and results of the "Outstanding Author Selection." The "Outstanding Author Selection" is a mechanism the editorial team has implemented since 2023, selecting existing authors with outstanding contributions during each quarter as well as emerging promising new authors after the quarter ends. Winners receive an exclusive badge and prizes. Go to view
- The 4.7-inch e-ink device Xiaozhi Pico, balancing portable reading and independent development, is now in stock on SSPAI. Go to purchase
- The SSPAI iOS app now supports iPhone Duo. Download and take a look
- Limited-time offers on SSPAI membership, genuine software, and paid columns. Don't miss out
Good articles you may have missed
- Hiking Travel Guide | Yulong Tiger Leaping Gorge and Paradise Yubeng
- Necessary but Unbalanced: A Brief Review of macOS 27.3's New Permission Control Mechanism
- New Gadgets 256 | What have SSPAI's editors bought recently?
- Grade Apple's 2026, and join in
Download the SSPAI 2.0 app and follow the SSPAI official account to unlock a new reading experience 📰
Practical, useful genuine software, presented to you by SSPAI 🚀