NAll news

All news

Thursday124 items

  1. The ElecProduct release
    AI score60

    Nvidia unveils OpenShell reference architecture at GTC 2026 Berlin for monitoring AI agents

    At GTC 2026 Berlin, Nvidia unveiled the OpenShell reference architecture to support the secure deployment and operation of AI agents. The architecture consists of Vera Rubin NVL72 server racks and BlueField-4 DPUs, with Rubin GPUs handling model inference, Vera CPUs handling agent control and tool execution, and BlueField-4 enforcing security policies at the hardware layer. The architecture also runs Nvidia Sentry security software, which can isolate and block operations within milliseconds when an agent crosses boundaries; the architecture is available now, and Dell, Cisco, HPE, Lenovo and Supermicro are building infrastructure based on it. Nvidia also announced the DSX AI Factory platform at the same event, claiming it can save 54 million liters of water per day in a 2GW facility compared with traditional cooling methods.

    Same story · 9 reports

    #Product updates#Agents#English text is being prepared.

  3. YicaiProduct release
    AI score52

    Zijing Zhikang Releases AI-Native Diagnostic and Treatment Operating System, with 54 AI Doctors Teaming Up to Collaborate

    On October 21, Zijing Zhikang, incubated by the Institute for AI Industry Research at Tsinghua University, released an AI-native diagnostic and treatment operating system that enables AI doctors to collaborate as a team like experts, with human doctors making the final decisions, improving efficiency by 30%. The system has built 54 AI doctors and millions of hyper-realistic AI patients, covering more than 20 clinical departments and tens of thousands of diseases. The AI doctors achieved a test accuracy of over 96% on the MedQA dataset, compared with an average of 87% for human experts. Its dual-layer evolution architecture allows AI doctors to first receive high-throughput training in a virtual world and then be deployed in physical hospitals to work alongside and collaborate with real doctors. The technology originates from the evolvable agent technology of Professor Liu Yang's team at Tsinghua University.

    Same story · 2 reports

    #Product updates#Agents#English text is being prepared.

  4. LeiphoneProduct release
    AI score59

    Qwen Releases AI Glasses N1 Pro and AI Clip-on Earbuds

    Qwen released the AI glasses N1 Pro and AI clip-on earbuds, with the glasses priced from 2,218 yuan and the earbuds from 1,099 yuan, both going on sale the same day. The N1 Pro weighs 44g and is equipped with three chips—Snapdragon AR1, Bestechnic, and Himax—relying on a dedicated low-power chip for continuous environmental perception, and features an eye-tracking module with center-region accuracy controlled within 1 degree. Song Gang, vice president of the Qwen business unit, said AI wearables are an extension of Qwen into the physical world and part of the Personal Agent, with the hardware used to capture physical-world context and then connect to the service capabilities of the Qwen App.

    #Product updates#Agents#English text is being prepared.

  5. 21ic Electronic NetworkProduct release
    AI score58

    Zoom Launches Unified Context Layer and Suite of AI Products

    At its annual Zoomtopia conference, Zoom announced a unified context layer that links conversation and document data across Zoom and third-party applications, making it available for both people and AI agents to access. Accompanying features include the now-available ZoomMate agent team, MCP servers covering the product portfolio, and a search Q&A API, the latter supporting queries to this context layer from tools such as Claude, GPT, Codex, Copilot, and Gemini. Features such as follow-up meetings, enhanced notes, the My Notes workflow, voice translation, and deepfake risk detection are expected to roll out gradually between October 2026 and the first quarter of 2027.

    #Product updates#Agents#English text is being prepared.

  6. WallstreetcnProduct release
    AI score77

    DeepSeek open-sources production-grade inference framework DeepInfer

    On October 20, DeepSeek announced the open-sourcing of its internal production-grade inference framework DeepInfer, which integrates technologies such as distributed inference and speculative decoding and is already compatible with NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs. Bloomberg noted that inference technology is usually regarded by AI companies as a core competitive advantage and is rarely made public; this open-sourcing gives outsiders a look at how DeepSeek keeps operating costs down. The documentation also states that building a complete service still requires a KV cache system, Prefill/Decode disaggregation, and load balancing across data-parallel replicas, which are not fully included in this open-source project.

    Same story · 29 reports

    #English text is being prepared.#Reasoning#DeepSeek

  7. heise onlineProduct release
    AI score52

    Broadcom releases VMware HyperArmor, providing proactive protection for the ESX hypervisor

    At VMware Explore on Tour in Frankfurt, Broadcom released VMware HyperArmor, which provides host-level proactive protection for ESX 8 and later versions. By allowing only known binaries and scripts to run and blocking unauthorized execution and writes, it enables virtual patching without reboots, and it adds TOTP two-factor authentication for SSH logins. The product will be offered as a paid add-on, launching in the first quarter of the fiscal year beginning in November; pricing has not yet been announced. Broadcom also mentioned that it has used Anthropic's Claude Mythos Preview model to review source code for several months, and the related findings have been incorporated into VCF 9.1.1.

    #Product updates#Safety / alignment#Agents

  8. NumeramaProduct release
    AI score68

    Anthropic Launches Claude Duet, a Proactive Personal Agent App

    On October 21, Anthropic released Claude Duet, a standalone app in which Claude runs as a proactive agent that can monitor connected apps, send a summary every morning, and execute tasks in the background within a cloud virtual machine. The product is currently a research preview, available only to Claude Pro and Max subscribers in the United States via a waitlist; it is not charged separately but consumes subscription usage, and it checks the account only once every few hours. The article compares it with Meta's Muse, launched on September 8, and OpenAI's Dots, launched on September 29, and mentions privacy experts' concerns about agents accessing personal data.

    Same story · 44 reports

    #Product updates#Agents#Anthropic

  9. t3nProduct release
    AI score76

    Anthropic releases proactive agent Claude Duet

    Anthropic releases the proactive agent Claude Duet, aimed at professionals and small business managers, delivering a daily morning briefing of the day's key points. It can connect to apps such as Gmail, Slack, Stripe, Canva, LinkedIn, and Shopify to summarize unread messages, check unpaid invoices, draft marketing campaigns, or build online stores, and can use more than 6,000 connectors in the Claude Marketplace. The product is initially available as standalone Android and iOS apps and a web research preview; Pro and Max subscribers can join the waitlist, with usage limits the same as other Claude products.

    Same story · 44 reports

    #Product updates#Agents#English text is being prepared.

  10. The DecoderProduct release
    AI score79

    Anthropic Launches Persistent Agent App Claude Duet Research Preview

    Anthropic has released a research preview of Claude Duet, a persistent agent app, taking aim at Grok Bot, Meta's Muse, and OpenAI's Dots. Duet continuously handles tasks on a cloud virtual machine, connecting to services such as Gmail, Slack, and Stripe through more than 6,000 connectors in the Claude Marketplace; it requires user approval before sending messages, making payments, signing, or deleting, and securities trading and transfers between accounts are not available even with approval. Anthropic acknowledges that early versions may misread message threads, misjudge urgency, and check connected accounts only every few hours.

    Same story · 44 reports

    #Product updates#Agents#Anthropic

  11. Shanghai ObserverProduct release
    AI score58

    AgiBot Open-Sources AGIBOT WORLD 2026 Phase 4 Embodied Reasoning Dataset

    On October 22, AgiBot officially open-sourced the AGIBOT WORLD 2026 Phase 4 Physical Reasoning dataset, with task design shifting its focus from manipulation difficulty to decision-making difficulty. The dataset constructs different state scenarios of the same task around factors such as object count, occlusion relationships, position occupancy, and task progress, covering nine core reasoning capabilities including attribute selection, conditional judgment, manipulation feasibility assessment, target search, long-horizon planning, and process memory, and sets three difficulty levels: Easy, Medium, and Hard. The data describes goals through task instructions, characterizes task structure through task stages, subtask intervals, and action steps, and provides action descriptions along with spatial annotations such as boxes, points, and lines. It will be used to train AgiBot's upcoming Agent framework and new-generation foundation model.

    #English text is being prepared.#Embodied AI#English text is being prepared.

  12. MiniMax BlogFirst-partyProduct release
    AI score61

    MiniMax Code Desktop Launches Mini App

    MiniMax has launched Mini App in local sessions of the MiniMax Code desktop app (macOS / Windows). It is a persistent application with a frontend and a Node.js backend, installed as a plugin, that can call and be called by the Agent, keeping work on a continuously alive page rather than scattered across conversations. This release includes the Mini App runtime (persistent data directory, hosted MCP, Agent API, Host Connectors, and in-page tool approval), /miniapp-creator for generating a page with a single sentence, and the official Mini Apps in the plugin marketplace and the open-source community repository MiniMax-Code-MiniApps. According to the official announcement, account credentials are kept by the host and cannot be accessed by plugins or pages, and content passed from a page to the Agent first enters the input box for review.

    Cover: a hand places a card into a cluster of Mini Apps, including a stock dashboard, a 3D model, a map, and a chat, connected by a line to the Agent

    #Product updates#Agents#English text is being prepared.

  13. WallstreetcnProduct release
    AI score77

    Anthropic Launches Claude Duet, a Proactive Agent Powered by Opus 5.5

    Anthropic announced the launch of Claude Duet, a proactive agent powered by Opus 5.5. The first batch is open as a waitlist to Claude Pro and Max subscribers in the United States, with support for iOS, Android, and the web. Duet comes with a cloud virtual machine and browser, can call on more than 6,000 connectors and plugins in the Claude app marketplace to advance tasks across systems, and inherits the memory and skills of the existing account. Actions such as sending messages, making payments, and signing documents require user approval; operations such as buying and selling securities and transferring funds between accounts are not permitted even with approval; background checks are usually performed every few hours, Slack functionality is limited, Microsoft 365 is not yet supported, and there is no additional charge during the research preview period.

    Same story · 44 reports

    #Product updates#Agents#Anthropic

  14. LeiphoneProduct release
    AI score56

    Volcano Engine Launches First Batch of Enterprise Trials for Seedance 2.5 4K Model, Supplying 20 Studios and Directors Worldwide by Invitation

    Volcano Engine has launched the first batch of enterprise user trials for the Seedance 2.5 4K model, tied to specific works and supplied by invitation to 20 studios and directors worldwide, targeting high-end film and television needs. This version emphasizes more delicate and natural character performances and finer consistency in reference generation and editing, and is the most powerful model in the Seedance 2.5 generation. A film studio that has already conducted trials said the price was not higher than expected; users can first generate sample footage at 480P and, once satisfied, regenerate it in 4K. Among the 20 works currently in use, the earliest, an S-tier IP for a satellite TV channel, will air as soon as next month.

    Same story · 2 reports

    #Product updates#Video#English text is being prepared.

  15. Kling AIFirst-partyProduct release
    AI score62

    Kling teases Kling 4 model

    The official Kling AI account posted a teaser with the caption "Hold on to your hat" and the hashtags #KlingAI #Kling4, without disclosing specific features or a release date. The cited OpenArt post says Kling 4.0 is already live on its platform, supporting video guidance with start, middle, and end frames, up to 10 reference images, and generating and editing videos with text or frames.

    Quoted postOpenArt · @openart_ai

    Kling 4.0 is now live on OpenArt. • Guide your video with start, middle, and end frames • Use up to 10 reference images • Generate with text or frames, and edit your videos • All-in-One Reference: take control further with one model for generating and editing videos Plus, get 50% off Kling 4.0 on Pro and Wonder plans until November 20.

    #English text is being prepared.#Product updates#Video

  16. Rohan PaulProduct release
    AI score76

    Anthropic launches Claude Duet, a proactive agent app

    Anthropic has launched Claude Duet, a proactive agent app, opening it to US Pro and Max subscribers via a waitlist. Duet checks connected accounts every few hours, runs on a cloud virtual machine with a separate browser, uses Claude Opus 5.5, and users currently cannot watch the browser operations in progress. Based on Opus 5.5's API pricing, a check that reads 40K tokens and writes 1.5K tokens costs about $0.19; six checks a day comes to about $34 a month, higher than Pro's $20 monthly price. Before sending messages, making payments, signing documents, or deleting content, Duet shows the specific action and waits for approval; securities trading and transfers between accounts remain prohibited even with approval.

    Same story · 44 reports

    #Product updates#Agents#Anthropic

  17. TMTPost · AGIProduct release
    AI score78

    Anthropic launches proactive Agent product Claude Duet

    Anthropic launches the standalone product Claude Duet, positioned as a proactive Agent. It is currently open as a research preview and requires an application submitted on Claude. The first batch is aimed at Pro and Max users in the United States, offering three entry points: iOS, Android, and claude.ai. Duet is designed based on Claude Opus 5.5. It proactively organizes important matters from connected apps and pushes briefings, and can also take on tasks such as scheduling meetings, building data dashboards, and making forecasts, and can use more than 6,000 connectors and plugins in the Claude Marketplace. Before sending messages, making payments, signing documents, or deleting content, it first displays the pending action and waits for approval. Credit card numbers and ID document information are directly denied access, while health and personal financial information is released only with explicit authorization.

    Same story · 44 reports

    #Product updates#Agents#Anthropic

  18. MacRumors Apple IntelligenceProduct release
    AI score57

    Apple says simplicity and Siri AI are HomeView's core selling points

    In an interview with The Verge, Apple's product marketing executive said that Siri AI on HomeView focuses on completing household commands with natural voice and minimizing follow-up questions, and that simplicity is a key feature of this home hub. HomeView can recognize up to 6 people and display corresponding information; Siri requests involving personal context require a device that supports Apple Intelligence. HomeView was announced on October 13, and both the desktop and wall-mounted versions are priced at $299 in the US, going on sale on November 13.

    #Industry news#Speech#English text is being prepared.

  19. NVIDIA：NewsroomFirst-partyProduct release
    AI score69

    NVIDIA Cosmos Expands New Tools for Training, Evaluating, and Deploying World Models

    NVIDIA expands the Cosmos platform with new training tools, simulation checkpoints, and Cosmos Data Factory blueprints to help developers build world models. Cosmos Data Factory offers five cloud-native blueprints, now available through GitHub, Microsoft Azure, Nebius, and CoreWeave; Porsche is using it to build a shared autonomous driving data flywheel, expected to reduce repetitive real-vehicle testing by up to 90% and shorten the development cycle by up to 85%. Cosmos Framework's training, optimization, and deployment tools are extended as agent skills to open-source models such as GR00T, Alpamayo, QWEN, and Pi0.5, and the Cosmos 3 NIM microservice delivers up to 2x faster inference under validated configurations.

    #Product updates#English text is being prepared.#English text is being prepared.

  20. Kling AIFirst-partyProduct release
    AI score63

    Kling 4.0 launches on OnSolo with support for generating 30-second native video in a single pass

    Kling 4.0 is now live on OnSolo and can generate 30 seconds of native video in one pass, maintaining motion stability through fast action and large-scale camera movements. This version supports stereo sound and more precise lip sync, a 21:9 widescreen aspect ratio, and up to 15 references and 10 keyframes stacked in a single job to maintain consistency of characters, shots, and story.

    Quoted postOnSoloAI · @OnSoloAI

    Kling 4.0 is now live on OnSolo. 30 seconds of native video in one take. Steady motion through fast action and big camera moves. Stereo sound with sharper lip sync. 21:9 widescreen. Stack up to 15 references and 10 keyframes in one job, so your characters, shots and story hold together. Make your next short drama, film or FMV with it today.

    #Product updates#Video#English text is being prepared.