studioglobal

Trending AI Posts & Hot Research Catchup - Page 327

TrendingDiscover
  • Illustration of a programmable silicon photonic chip carrying entangled qubits as light travels through integrated waveguides

    Guizhen Chip’s 16-Qubit Photonic Chip: What the USTC Preprint Actually Demonstrated

    last month17 sources
    0

    The August 4, 2026 arXiv preprint reports a four photon, 16 qubit GHZ graph state, with genuine entanglement verified across 10 qubits, plus a four qubit cluster state demonstration.

    Open original page
  • A consumer wearing a powered hip-and-thigh exoskeleton while hiking on a rocky outdoor trail

    Hypershell X Ultra S Review: What It’s Really Like to Wear a $2,000 AI Exoskeleton

    last month23 sources
    0

    Hypershell’s X Ultra S costs ¥13,999, weighs 1.8 kg and claims up to 30 km of range in Eco 30.

    Open original page
  • Abstract editorial illustration of Chinese and U.S. AI models competing across a global token-usage network

    Chinese AI Models Led OpenRouter Usage for 15 Straight Weeks

    last month23 sources
    0

    Chinese developed models led OpenRouter usage for the 15th consecutive week, recording 34.25 trillion tokens versus 9.17 trillion for U.S.

    Open original page
  • Abstract editorial illustration of MiniMax H3 unifying text, image, video, and audio inputs for AI video generation

    MiniMax H3 Explained: Open-Weight Video Generation With Native Audio

    last month28 sources
    0

    Released with open weights on August 3, 2026, MiniMax H3 combines text, images, video, and audio in one generation workflow and can produce clips of up to about 15 seconds with native stereo audio.

    Open original page
  • Illustration of China’s national AI compute fabric linking a 100,000-card supercluster with optical networks and regional data centers

    China’s 100,000-Card AI Cluster Signals a Shift From Chips to a National Compute Fabric

    last month27 sources
    0

    China’s reported 100,000 card Zhengzhou cluster marks a shift from building isolated AI data centers to coordinating domestic compute nationally.

    Open original page
  • Illustration of an AI agent phone coordinating apps, tools and user approvals under China’s L3 terminal intelligence framework

    China’s L3 AI Phone Standard Explained: What the First 11 Devices Actually Prove

    last month20 sources
    0

    China’s first July 17, 2026 AI terminal test results named 11 mobile devices—nine smartphones and two tablets—as L3.

    Open original page
  • What is Alibaba’s open source Wan Animate 2, released by the Tongyi Wanxiang team in August 2026, and how does its end to end diffusion Tran

    What is Alibaba’s open-source Wan-Animate-2, released by the Tongyi Wanxiang team in August 2026, and how does its end-to-end diffusion-Tran

    last month14 sources
    0

    Wan-Animate-2 is Alibaba Tongyi Lab’s open character-animation system: it animates a reference character from a driving video, while retaining the character’s identity. Its key advance is that motion is lear...

    Open original page
  • AI-generated illustration of a compact humanoid robot demonstrating mobility and customer service at an automotive exhibition

    Tiangong Omni Debuts as Tiangong 3.0 Takes Aim at Automotive Retail

    last month21 sources
    0

    At WRC 2026, Tiangong Omni debuted as a 1.35 meter, 39 kilogram humanoid built for flexibility in tight environments, while Tiangong 3.0 autonomously greeted visitors, explained vehicles and handled consulta...

    Open original page
  • Εννοιολογική απεικόνιση FOPLP με chip, στρώσεις RDL και διαδρομές απαγωγής θερμότητας

    Why Huatian Technology Is Putting Thermal Design at the Heart of FOPLP

    last month17 sources
    0

    Huatian Technology says its FOPLP approach integrates thermal design with material selection, wiring and package structure instead of treating heat as a post assembly problem.

    Open original page
  • Abstract editorial illustration representing Xiaohongshu’s dots3-note preview multimodal AI model and its long-context MoE architecture

    Why Xiaohongshu Open-Sourced a 280B AI Model—and What It Signals About Its Platform Strategy

    last month21 sources
    0

    Released in August 2026, dots3 note preview is a 280B parameter MoE model with 16B active parameters, a 512K context window and text, image, video and audio input.

    Open original page
  • Editorial illustration representing Qwen3.8-27B running as a local multimodal AI model on consumer hardware

    Qwen3.8-27B Explained: Why the 17GB Local AI Model Is Being Called a “Model Kill Line”

    last month19 sources
    0

    Released on August 14, 2026, Qwen3.8 27B is a 27B Apache 2.0 multimodal model that passed one million downloads in two days and can fit a 17GB quantized build on suitable consumer hardware.

    Open original page
  • Illustration of an AI GUI agent operating phone and computer screens with clicks, typing, and swipes

    Qwen-UI-Agent Explained: Alibaba’s Real-Device AI That Operates Phones and Computers

    last month18 sources
    0

    Qwen UI Agent is Alibaba’s real world GUI agent foundation model, released on August 20, 2026.

    Open original page
Load more