source&pool
A daily wire of long-form journalism, video, and discourse — filed, tagged, and laid out flat.
VOL. I·NO. 01
WEDNESDAY, SEPTEMBER 23, 2026
Hacker News3671X 主题热门3560CNBC71YahooFinance549to5Mac52aihot50MacRumors50IGN38Kotaku37Verge37TechCrunch27Gematsu22NintendoLife209to5Google19AndroidAuthority19Eurogamer19BusinessInsider18Engadget16WarhammerCommunity15PushSquare14USAToday14Polygon13Guardian13Wccftech13ArsTechnica12Fortune11Investor'sBusinessDaily11NBC11AppleInsider10FoxBusiness10Gizmodo10Notebookcheck10NPR10SeekingAlpha10TechPowerUp10CBS9CoinDesk9ABC8bgr8CNN8MotleyFool8Mashable8NintendoEverything8VideoCardz8GSMArena7WIRED7Yahoo7CNET6Fox6PureXbox6Conversation6AlJazeera5BleepingComputer5NewYorkPost5PetaPixel5SamMobile5TechSpot5VideoGamesChronicle5Deadline4GameInformer4XBOXWire4Pokemon4RockPaperShotgun4RPGSite4SlashGear4Tom'sGuide4Variety4WindowsCentral4WSB-TV4404Media380Level3Aftermath3AndroidCentral3AndroidPolice3BellofLostSouls3DigitalFoundry3DroidLife3EventHubs3HollywoodReporter3HuffPost3Motor13MP1st3PlayStationLifeStyle3PokeBeach3SouthChinaMorningPost3SeattleTimes3TimeExtension36abcPhiladelphia2ABC7LosAngeles2Benzinga2CanonRumors2Currently2DCRainmaker2DigitalCameraWorld2FratelloWatches2Futurism2GamesIndustry.biz2GearPatrol2HouseDigest2InsiderGaming2LosAngelesTimes2Lifehacker2Nature2PCMag2qz2SFGATE2SimsCommunity2Register2TweakTown2YahooFinanceUK2WhatHi-Fi?224/7WallSt.1BusinessInsiderAfrica1Alternet1AndroidHeadlines1AOL1ArizonaSports1AZFamily1BikeRadar1BloodyDisgusting1Boston1BostonGlobe1BusinessTimes1BuzzFeed1CTech1CalMatters1CarBuzz1cbn1ChromeUnboxed1Chron1ClaimDepot1ColoradoSun1ChristianScienceMonitor1CyberSecurityNews1Cyclingnews1DailyDownforce1DailyKos1DaringFireball1DarkHorizons1Decrypt1Defector1Defense1denver71DenverPost1Designboom1DirtonDirt1Draftsim1DSOGaming1DW1empireonline1erictopol.substack1Euronews1Fangoria1FOX191DetroitFreePress1GameDeveloper1GameRant1GameWorldObserver1GAMINGbible1GamingOnLinux1AAAGasPrices1GeekyGadgets1Global1Gothamist1Hackaday1Hackster.io1Hodinkee1HoustonChronicle1Independent1InterestingEngineering1Invezz1Jalopnik1KITCO1MacObserver1Magic:Gathering1MakeUseOf1Mashed1Maxroll1Mercury1MLive1MonochromeWatches1MorningBrew1MortgageDaily1Motorsport1MyNintendo1BloombergLaw1SemiAnalysis1Newsshooter1Newsweek1NintendoWire1nrn1CrudeOilPricesToday1OneMileataTime1OregonPublicBroadcasting1PageSix1politico.eu1QuantaMagazine1Realtor1Road&Track1RockstarINTEL1Salon1CultureMapSanAntonio1SeattleRed1Semafor1Slate1SlippedDisc1Space1SpaceNews1YahooTech1the5krunner1DailyBeast1DailyMeal1Drive1Hacker1Hindu1Intercept1Times1TimesofIndia1TimesUnion1TMZ1TODAY1UploadVR1VisualCapitalist1WHYY1WindowsLatest1WKYT1WOWT1YGOrganization1YourTango1
  1. 001Hacker NewsSEP · 23English

    Trained KV cache bank turns any LLM into Jev like Model

    The content appears to be a technical interface or dashboard showing inference metrics and latency measurements, but lacks substantive information to analyze.

    By faangguyindia
  2. 002X 主题热门SEP · 23English

    DeFi · X 热门 · 2026-09-23 06:47 UTC

    A collection of DeFi discussions from September 23, 2026, covering validator latency challenges in blockchain networks, the NectarFi platform for global money management, Bitcoin integration into Starknet for yield opportunities, and analysis of DeFi protocol sustainability beyond advertised APY rates.

  3. 003Hacker NewsSEP · 23English

    Mastering LLM Inference Optimization

    This article explains LLM inference optimization techniques for production deployment. It covers the two-phase inference process (prefill and decode), memory management strategies like KV caching and PagedAttention, and methods including model compression and speculative decoding to improve speed, cost, and reliability without retraining.

    By Bala Priya C
  4. 004Hacker NewsSEP · 22English

    Claude Opus 5.5 (High Effort) Intelligence, Performance and Price Analysis

    Claude Opus 5.5 is a high-intelligence reasoning model by Anthropic with a score of 54 on the Artificial Analysis Intelligence Index, placing it well above average. It offers faster-than-average speed at 91 tokens per second but carries somewhat elevated pricing at $4.00 per 1M input tokens and $20.00 per 1M output tokens, with a 1M token context window.

    By Topfi
  5. 005Hacker NewsSEP · 22English

    We Rebuilt Jev's API on an Open Model and Used It to Play Doom

    Researchers rebuilt Jev's API using an open base model (Gemma4) to replicate TypeSafe's System One Model design, which performs zero-shot classification for fast decisions. They demonstrated the replica playing Doom and Flappy Bird with 100-124ms latency per decision, validating that open models can match Jev's published performance without needing TypeSafe's proprietary training methods.

    By Stephen Blum
  6. 006Hacker NewsSEP · 22Chinese

    Show HN: Ego-jev – 0.4s typed decisions for browser agents

    Ego-jev is a browser agent skill that makes typed decisions in ~0.4 seconds per DOM step using TypeSafe's System One API, replacing full LLM calls. It numbers interactive elements, makes one API call to pick an operation and target together, then executes via ego-browser, escalating complex tasks like logins and payments back to the planner.

    By ZephyrDeng
  7. 007X 主题热门SEP · 22English

    DeFi · X 热门 · 2026-09-22 16:00 UTC

    A collection of DeFi-focused X posts discussing privacy token infrastructure on Beldex, price performance of Kaspa cryptocurrency, latency optimization solutions for decentralized finance, and ecosystem mechanics of Clutch Markets NFT and token platform.

  8. 008Hacker NewsSEP · 22English

    Show HN: Gemma 3 4B as a typed decision function in Rust (47 ms/decision)

    A Rust implementation runs Google's Gemma 3 4B model as a typed decision function, achieving 47 ms per decision (21.2 decisions/sec) on Apple M1 Pro. Instead of token generation and parsing, the system directly extracts logits for legal labels and applies softmax to produce typed answers, with benchmarks showing 58.9% accuracy on JevBench's 231 public decisions.

    By Zozo
  9. 009Hacker NewsSEP · 22English

    GPU is starving – LLM host dispatch at 191k steps/s on 1 vCPU

    Floria is a high-throughput serving engine for large language models that eliminates GPU starvation by replacing Python-based scheduling with native hardware dispatching. Running on a single vCPU, it achieves 191k tokens/sec and 100% GPU utilization, compared to conventional systems like vLLM that leave GPUs idle 25-40% of the time due to host scheduling latency.

    By Cortexlab