studioglobal

Trending AI Posts & Hot Research Catchup - Page 295

TrendingDiscover
  • How does Moonshot AI’s 47 page Kimi K3 technical report argue that AI progress depends on scaling both pre deployment model size and post de

    How does Moonshot AI’s 47 page Kimi K3 technical report argue that AI progress depends on scaling both pre deployment model size and post de

    last month13 sources
    0

    K3’s core argument is that frontier progress requires scaling two complementary resources: model capacity before deployment, and the compute spent after deployment on long reasoning, tool use, and agentic ro...

    Open original page
  • How did Chinese startup Mineng Technology’s tens of millions yuan Series A financing support development of its self designed spiking neural

    How did Chinese startup Mineng Technology’s tens of millions yuan Series A financing support development of its self designed spiking neural

    last month19 sources
    0

    The reported financing appears to fund commercialization rather than a conventional GPU like chip program: Mineng/MinoTec says the proceeds will support mass production and iteration of medical grade modules...

    Open original page
  • Editorial illustration representing evidence-driven AI coding workflow evaluation

    Better Harness: Qoder’s Evidence-Driven Framework for Improving AI Coding Workflows

    last month15 sources
    0

    Better Harness is Qoder’s open source workflow reviewer for coding agents: it evaluates the engineering environment around an agent, turns supported gaps into scoped fixes, and can be rerun to verify progress.

    Open original page
  • Editorial illustration of China’s AI office-agent competition among Tencent, Alibaba, and 360

    China’s AI Office-Agent Race: Tencent’s Lead, Alibaba’s Reset, and 360’s SMB Bet

    last month17 sources
    0

    Tencent’s WorkBuddy led China’s desktop AI office agent market with 20.97 million monthly PC visits in June 2026, while Alibaba consolidated QoderWork, Wukong, and MuleRun into Qianwen Office and 360 used a...

    Open original page
  • Editorial illustration of an AI agent collaborating in a Git-based software development workflow

    CodeBuddy NPC and AI Native Git: How Tencent Cloud Turns Git Workflows Into AI Delivery Loops

    last month22 sources
    0

    CodeBuddy NPC is designed to let developers @mention an AI role in an Issue or pull request, then have it work asynchronously through planning, coding, PR submission, testing and CI driven repair.

    Open original page
  • Εικονογράφηση ενός συστήματος τεχνητής νοημοσύνης για αυτόνομη οδήγηση με οπτική ανάλυση δρόμου

    How BYD’s HyWorldVLA Blends Pixel and Latent World Models for Autonomous Driving

    last month15 sources
    0

    BYD’s HyWorldVLA reported a 90.59 PDMS score on NAVSIM v1 by combining pixel level supervision with prediction in a compressed latent space.

    Open original page
  • Editorial illustration of Tencent WorkBuddy’s rapid growth in China’s AI office-agent market

    How Tencent WorkBuddy Reached 20.97M Visits and Reshaped China’s AI Office Race

    last month23 sources
    0

    Tencent WorkBuddy grew from 8.85 million PC side visits at its March 2026 launch to 20.97 million in June, surpassing ByteDance Trae and Alibaba QoderWork combined.

    Open original page
  • How does Tencent Cloud’s CodeBuddy NPC, launched on July 29, 2026, implement an AI Native Git paradigm in which developers @mention on deman

    How does Tencent Cloud’s CodeBuddy NPC, launched on July 29, 2026, implement an AI Native Git paradigm in which developers @mention on deman

    last month20 sources
    0

    CodeBuddy NPC’s core idea is to make AI an authenticated, event driven participant in the existing Git development system—not a chat window beside it.

    Open original page
  • Editorial illustration of AI coding costs, comparing metered token billing with subscription allowances

    Why Cheap Chinese LLM APIs Can Cost More for Heavy AI Coding

    last month14 sources
    0

    GLM 5.2’s $1.40 per million input tokens and $4.40 per million output tokens are inexpensive on a unit basis, but a metered API has no spending ceiling.

    Open original page
  • Editorial illustration of a sparse mixture-of-experts AI model and connected agent workflow

    Ling-3.0-flash Shows Why Sparse MoE Models Can Beat Bigger AI Systems on Cost

    last month14 sources
    0

    Ling 3.0 flash does not prove that bigger models no longer matter, but it shows how a 124B parameter MoE can target strong agent performance while activating only 5.1B parameters per token.

    Open original page
  • Editorial illustration of robotic orthopedic surgery and a knee implant, representing Tinavi’s proposed MicroPort Orthopedics acquisition

    Tinavi’s MicroPort Deal: Why a Surgical-Robot Maker Wants the Implant Business

    last month22 sources
    0

    Tinavi’s proposed 62% purchase of Shanghai MicroPort Orthopedics is designed to combine a surgical robot platform with repeat use hip and knee implants and overseas commercialization capabilities.

    Open original page
  • Editorial illustration of an embodied AI robot processing audio, video, speech, and motion signals

    Why AgiBot’s WITA-Omni Scored 85.21 on Daily-Omni

    last month25 sources
    0

    AgiBot reports that WITA Omni Preview led Daily Omni with 85.21% average accuracy and first or joint first results in six of eight measures; its apparent advantage is a focus on jointly reasoning over audio,...

    Open original page
Load more