source&pool
A daily wire of long-form journalism, video, and discourse — filed, tagged, and laid out flat.
VOL. I·NO. 01
THURSDAY, SEPTEMBER 24, 2026
Hacker News3806X 主题热门3702CNBC65YahooFinance63aihot619to5Mac49MacRumors47IGN44Verge42Kotaku38TechCrunch27AndroidAuthority26Engadget23NintendoLife22Gematsu209to5Google17USAToday16Wccftech16ArsTechnica15BusinessInsider15Eurogamer15Guardian15WarhammerCommunity15FoxBusiness14Polygon14Fortune13Investor'sBusinessDaily13PushSquare13TechPowerUp12NPR11SeekingAlpha11CBS10Gizmodo10GSMArena10NBC10ABC9CNN9Notebookcheck9VideoGamesChronicle9AppleInsider8BleepingComputer8CNET8MotleyFool8Mashable8NintendoEverything8VideoCardz8Yahoo8bgr7CoinDesk7Fox7PureXbox7NewYorkPost6Conversation6Aftermath5AlJazeera5Deadline5HollywoodReporter5PetaPixel5PlayStationLifeStyle5Pokemon5SouthChinaMorningPost5TechSpot5Tom'sGuide5WIRED5AndroidPolice4GamesIndustry.biz4InsiderGaming4Motor14Nature4XBOXWire4SamMobile4SlashGear4Hacker4UploadVR4Variety4WSB-TV4AndroidCentral3BellofLostSouls3DroidLife3EventHubs3GearPatrol3Hackaday3HuffPost3Lifehacker3PCMag3PokeBeach3RockPaperShotgun3RPGSite3SeattleTimes3SFGATE3WhatHi-Fi?3WindowsCentral3404Media26abcPhiladelphia280Level2ABC7LosAngeles2AZFamily2Benzinga2Currently2DaringFireball2DCRainmaker2DigitalFoundry2Futurism2GameInformer2GAMINGbible2HouseDigest2Jalopnik2LosAngelesTimes2MyNintendo2politico.eu2qz2Road&Track2SimsCommunity2Slate2TimeExtension2TODAY2TweakTown2YahooFinanceUK224/7WallSt.1BusinessInsiderAfrica1ageofempires1Alternet1AndroidHeadlines1ArizonaSports1BikeRadar1BloodyDisgusting1BostonGlobe1BusinessTimes1Yahoo!FinanceCanada1CalMatters1CarBuzz1cbn1ChromeUnboxed1Chron1ClaimDepot1ColoradoSun1Skin.ClubCommunity1ChristianScienceMonitor1CyberSecurityNews1Cyclingnews1DailyDownforce1DailyKos1DarkHorizons1Decrypt1Defense1denver71Designboom1DigitalCameraWorld1DirtonDirt1DSOGaming1Electrek1empireonline1GameGPU1erictopol.substack1Fangoria1ForexFactory1FOX191franchisetimes1DetroitFreePress1GameRant1GameWorldObserver1GamingOnLinux1AAAGasPrices1GeekWire1GeekyGadgets1Global1GosuGamers1Gothamist1Hackster.io1Hodinkee1HoustonChronicle1Independent1InsideEVs1InterestingEngineering1Invezz1iPhoneinCanada1MacObserver1Magic:Gathering1MakeUseOf1Mashed1Mercury1MLive1MorningBrew1Motorsport1MP1st1BloombergLaw1SemiAnalysis1Newsshooter1Newsweek1NintendoWire1nrn1NYT1CrudeOilPricesToday1OneMileataTime1OregonPublicBroadcasting1OregonLive1PageSix1PersonaCentral1Phoronix1QuantaMagazine1Realtor1RoadtoVR1RockstarINTEL1Salon1SeattleRed1Semafor1SanFranciscoChronicle1SimpleFlying1GhostHowls1SlippedDisc1SoraNews241SpaceNews1statnews1YahooTech1the5krunner1DailyBeast1DailyMeal1Drive1Hindu1Intercept1Register1Times1TimesofIndia1TimesUnion1TMZ1TopGear1PCMagUK1VisualCapitalist1Vulture1WCVB1WFMZ1WHYY1WKYT1YGOrganization1YourTango1
  1. 001Hacker NewsSEP · 23English

    GPT-6 Astra is the best vision model

    OpenAI's GPT-6 Astra vision model achieves state-of-the-art object detection performance, scoring 82.1% mAP@50 on Roboflow Vision Evals and outperforming competitors like Qwen3.8 Max and GPT-5.6 Sol. The model excels at computer vision tasks including object detection, visual reasoning, box prompting, and segmentation, combining fine-grained detail detection with semantic understanding for effective auto-annotation and classification across diverse visual scenarios.

    By Piotr Skalski
  2. 002Hacker NewsSEP · 20English

    A design brief as one image: can a 27B vision model build the animation?

    A 27B vision model was tasked with building canvas animations from design briefs provided as single PNG images. Six quantized versions of Qwen3.8-27B and two Bonsai variants were evaluated across three design cards using 17 objective checks per page, with results showing Opti performing competitively with Q4_K_M despite 28% smaller file size, while Bonsai struggled with JavaScript errors and token budget exhaustion.

    By airylizard
  3. 003Hacker NewsSEP · 17English

    Show HN: Knowhere-Rebuilds the document structure that chunking throws away

    Knowhere is an open-source document parsing and retrieval system that reconstructs document hierarchy and structure for AI agents, supporting both text and vision-based parsing tracks that converge into a unified memory schema for RAG and agentic workflows.

    By Ontos-AI