微博微博今日头条今日头条百度百度快手快手哔哩哔哩哔哩哔哩知乎知乎腾讯新闻腾讯新闻网易新闻网易新闻澎湃新闻澎湃新闻新浪新闻新浪新闻新浪网新浪网豆瓣电影豆瓣电影GitHubGitHubCSDNCSDNIT之家IT之家36氪36氪宝庆银楼宝庆银楼中国黄金中国黄金周生生周生生英皇珠宝英皇珠宝老凤祥（广东）老凤祥（广东）六福珠宝六福珠宝周大福周大福周六福周六福Adafruit BlogAdafruit BlogAIbaseAIbaseAnt Bailing Developer BlogAnt Bailing Developer BlogAnthropicAnthropic小众软件小众软件AppleAppleApple PodcastsApple PodcastsAppleInsiderAppleInsiderArs TechnicaArs TechnicaarXivarXivAxiosAxiosBerkeley AI ResearchBerkeley AI ResearchBarchartBarchartBBC NewsBBC NewsBD Tech TalksBD Tech TalksBerkeley RDIBerkeley RDIBig ThinkBig Think哔哩哔哩热搜哔哩哔哩热搜哔哩哔哩热门视频哔哩哔哩热门视频BinanceBinanceBloombergBloombergBlueskyBlueskyBoston Dynamics BlogBoston Dynamics BlogBusiness InsiderBusiness InsiderByteDance Seed ResearchByteDance Seed Research参考消息参考消息虫部落虫部落Claude BlogClaude BlogClaude Code ReleasesClaude Code ReleasesCloudflare BlogCloudflare BlogCMU Machine Learning BlogCMU Machine Learning BlogCoinGeckoCoinGeckoCointelegraphCointelegraphcrates.iocrates.ioCrowd SupplyCrowd SupplyCSS-TricksCSS-Tricks51CTO51CTOCult of MacCult of MacCursor BlogCursor Blogdaily.dev · Populardaily.dev · PopularDaring FireballDaring FireballDario AmodeiDario AmodeiDeepLearning.AI · The BatchDeepLearning.AI · The BatchGoogle DeepMind BlogGoogle DeepMind BlogDeepSeek GitHubDeepSeek GitHubDefiLlamaDefiLlamaDescript BlogDescript BlogDEV.toDEV.to数字尾巴数字尾巴Docker HubDocker Hub豆瓣读书豆瓣读书豆瓣讨论豆瓣讨论Dwarkesh PatelDwarkesh PatelThe EconomistThe EconomistEleutherAI BlogEleutherAI BlogEngadgetEngadgetFinancial TimesFinancial TimesFlathubFlathubfreeCodeCampfreeCodeCampFrontiersFrontiersGary MarcusGary Marcus极客公园极客公园原神原神果核剥壳果核剥壳GoogleGoogleGoogle TrendsGoogle Trends果壳果壳HackadayHackadayHacker NewsHacker NewsHelloGitHubHelloGitHub历史上的今天历史上的今天HomebrewHomebrew崩坏3崩坏3Hugging FaceHugging Face活动行活动行虎扑虎扑虎嗅虎嗅IEEE SpectrumIEEE Spectrum爱范儿爱范儿凤凰网 · 热点资讯凤凰网 · 热点资讯inclusionAIinclusionAIIndie HackersIndie HackersInfoQInfoQInterconnectsInterconnects爱奇艺热播榜爱奇艺热播榜IT之家「喜加一」IT之家「喜加一」简书简书稀土掘金稀土掘金Andrej KarpathyAndrej KarpathyMoonshot AI KimiMoonshot AI KimiLatent SpaceLatent SpaceLessWrongLessWrongLichessLichessLil'Log (Lilian Weng)Lil'Log (Lilian Weng)Linux.doLinux.doLMSYS BlogLMSYS BlogLobstersLobsters英雄联盟英雄联盟LWN.netLWN.netMacRumorsMacRumorsMake: MagazineMake: MagazineMaven CentralMaven CentralMediumMediumMeituan LongCatMeituan LongCatMeta AI BlogMeta AI BlogMeta EngineeringMeta EngineeringMidjourney UpdatesMidjourney UpdatesMiniMaxMiniMaxMIT News · RoboticsMIT News · RoboticsMIT Technology ReviewMIT Technology Review米游社米游社Mozilla.ai BlogMozilla.ai BlogNatureNatureNeuroscience NewsNeuroscience NewsNew Atlas · RoboticsNew Atlas · RoboticsThe New YorkerThe New Yorker水木社区水木社区NGANGA9to5Mac9to5MacNodeSeekNodeSeek牛客牛客NuGetNuGetNVIDIANVIDIA纽约时报纽约时报OEISOEISOne Useful ThingOne Useful ThingOpen Robotics BlogOpen Robotics BlogOpenAIOpenAIOpenAlexOpenAlexOpenReviewOpenReviewOpenRouterOpenRouterPackagistPackagist远景论坛远景论坛PhoronixPhoronixPhys.orgPhys.orgPixivPixivPlanet ROSPlanet ROS吾爱破解吾爱破解PoliticoPolitico中国电建阳光采购网中国电建阳光采购网Product HuntProduct HuntPubMedPubMedPyPIPyPIQoderQoder腾讯视频热搜榜腾讯视频热搜榜Quanta MagazineQuanta MagazineQwenQwenReutersReutersRFC EditorRFC EditorRobohubRobohubRobotics & Automation NewsRobotics & Automation NewsRobotics TomorrowRobotics TomorrowROS DiscourseROS Discourse阮一峰的网络日志阮一峰的网络日志RubyGemsRubyGemsRunwayRunwaySam AltmanSam AltmanScienceAlertScienceAlertAhead of AIAhead of AISecurityOnlineSecurityOnlineServeTheHomeServeTheHomeServiceNow AIServiceNow AISimon WillisonSimon WillisonSingularity HubSingularity HubSky NewsSky NewsSlashdotSlashdotSmashing MagazineSmashing Magazine什么值得买什么值得买SolidotSolidotSpotifySpotify俄罗斯卫星通讯社俄罗斯卫星通讯社少数派少数派Stack OverflowStack OverflowStack Overflow BlogStack Overflow Blog崩坏：星穹铁道崩坏：星穹铁道SteamSteamSubstackSubstackSunoSunoSuno BlogSuno BlogSynced ReviewSynced Review淘宝逛一逛淘宝逛一逛TechCrunchTechCrunchTech Xplore · RoboticsTech Xplore · Robotics腾讯热点腾讯热点The AtlanticThe AtlanticThe DecoderThe DecoderThe GradientThe GradientThe GuardianThe GuardianThe RegisterThe RegisterThe VergeThe Verge百度贴吧百度贴吧TikTokTikTokTindie BlogTindie BlogTomer TunguzTomer TunguzTom's HardwareTom's HardwareTransformer CircuitsTransformer CircuitsTVmazeTVmaze优设网优设网UiverseUiverseV2EXV2EXVentureBeat · AIVentureBeat · AI万方数据万方数据中央气象台中央气象台微信读书微信读书WikidataWikidataWikipediaWikipediaWiredWiredThe Wall Street JournalThe Wall Street JournalxAI NewsxAI News小鹅通小鹅通YahooYahooYahoo FinanceYahoo FinanceYollomiYollomi有道精品课有道精品课YouTubeYouTube游研社游研社知乎日报知乎日报Zhipu AI ResearchZhipu AI Research财新数据通财新数据通央视新闻联播央视新闻联播CelesTrakCelesTrakCISA KEVCISA KEV财联社财联社CoinDeskCoinDesk东方财富东方财富法布财经法布财经富途牛牛富途牛牛格隆汇格隆汇HDXHDX和讯网和讯网Investing.comInvesting.com界面新闻界面新闻金十数据金十数据21财经21财经金融界金融界Launch Library 2Launch Library 2MKTNews · 快讯MKTNews · 快讯NASA EONETNASA EONETNOAA/NWSNOAA/NWSNVDNVDopenFDAopenFDAOSV.devOSV.dev量子位 · 具身智能量子位 · 具身智能新浪财经新浪财经Spaceflight NewsSpaceflight NewsTelegram OSINTTelegram OSINT腾讯混元研究腾讯混元研究同花顺同花顺USAspendingUSAspendingUSGS EarthquakesUSGS Earthquakes华尔街见闻华尔街见闻WHO 疫情通报WHO 疫情通报选股宝选股宝第一财经第一财经

实时热搜

731 个榜单

Adafruit Blog

实时热榜
6分钟前更新
  • 01
    Is this the year of the Linux… television?
    WayDroid-ATV has published a set of Waydroid Android TV builds on GitHub. The images run Android TV on top of an existing Linux install rather than replacing it, so a desktop or server keeps doing its normal work while the TV interface runs alongside. The point is resource use. Android TV x86 ROMs like LineageOS […]Anne Barela
  • 02
    What it took to triple software engineering output in 18 months
    A technology leader writing in InfoWorld describes what changed at a legal software company after it rebuilt its product development life cycle around AI over 18 months. Output per engineer roughly tripled, deployments went from 82 to more than 155, and customer reported defects fell 65% per million lines of code. Progress was tracked with […]Anne Barela
  • 03
    The ill-fated HP-Compaq merger
    Dave Farquhar looks back at the HP-Compaq merger on his blog, The Silicon Underground. HP announced the $25 billion acquisition on September 6, 2001. At the time it was the largest technology merger in history. HP retired the Compaq brand in May 2012. Farquhar traces Compaq’s weakness to its own 1997 purchase of Digital Equipment […]Anne Barela
  • 04
    Making a Python interpreter in 1024 bytes
    Austin Z. Henley set himself a weekend challenge on his blog at austinhenley.com: write a Python interpreter in 1024 bytes of C, with no macro tricks and no libraries. The original target was 512 bytes. A basic calculator that handled x = 1 + 2 * 3 and a single if statement already blew past that, so […]Anne Barela
  • 05
    The CircuitPython Show Episode 64 – CircuitPython Day 2026
    Episode 64 of The CircuitPython Show is a replay of The Bootloader livestream that ran on Adafruit’s YouTube channel for CircuitPython Day. Host Paul Cutler and co-host Tod Kurt look back at the past year of CircuitPython releases and community projects. The episode uses screen sharing, so the video version is the one to watch. […]Anne Barela
  • 06
    Transmit unsigned 8-bit audio samples as analog FM using a CC1101 module
    ktauchathuranga has published a library on GitHub called cc1101-analogfm that transmits unsigned 8-bit audio samples as analog FM using a CC1101 radio module. The stream() function takes bytes from any Arduino Stream and treats each one as an audio sample: 0 is the negative peak, 128 is the carrier center, and 255 is the positive […]Anne Barela
  • 07
    CHC5 open Camera System Image Sensor Specification comparison
    Gaurav Singh has posted the first part of a series on the CHC5 open camera system at circuitvalley.com, and it starts with picking a sensor. The post is a side by side technical reference for nine candidates: the Sony IMX477, IMX678, IMX485, IMX585, IMX283, IMX294, IMX568 and IMX565, plus the OmniVision OX08B40. The comparison covers […]Anne Barela
  • 08
    Raspberry Pi CM4 bindings for JDI LT070ME05000 1200×1920 IPS display
    harlab has published a project that connects the JDI LT070ME05000 display to the Raspberry Pi Compute Module. The panel is the 7 inch 1200×1920 IPS screen from the 2013 Asus/Google Nexus 7 tablet. Some suppliers still have them for around $20 shipped. The display uses a 4-lane DSI interface, so it needs a Compute Module […]Anne Barela
  • 09
    nes-radar: Live ADS-B air traffic radar for the original Nintendo Entertainment System
    Developer k6lcm has published NES Radar which turns an original Nintendo Entertainment System into a live air traffic scope. Enter any airport’s four-letter ICAO code on the NES controller, press Start, and aircraft flying near that airport show up as moving targets on the TV. It works for airports worldwide, so a quiet local field […]Anne Barela
  • 10
    wordstar: The original, unmodified MicroPro WordStar Professional Release 4 from 1987
    GitHub user nampara-ai has posted a GitHub project that runs MicroPro WordStar Professional Release 4 from 1987 on modern computers. The files in the repo are the original, unmodified binaries, not a clone or a rewrite. WordStar is 16-bit MS-DOS code, so the desktop version wraps it in DOSBox Staging, which is already bundled for […]Anne Barela
  • 11
    T3CAD: An agent-driven electronics workspace with KiCad PCB, schematic, Gerber, 3D, library, and more
    Developer i2cjak has published T3CAD, an agent-driven electronics workspace forked from T3 Code. It pairs coding-agent conversations with KiCad PCB, schematic, Gerber, and 3D inspection. It runs on web, desktop, and mobile. Inside the KiCad viewer there are optional tabs for BOM, footprint, symbol, and EMerge / Analysis. The BOM follows the project’s saved KiCad […]Anne Barela
  • 12
    Stopping the unstoppable
    Practical Engineering looks at how railroads try to stop a train that has run out of track. The starting point is the 2016 New Jersey Transit crash at Hoboken Terminal, where a commuter train hit the terminal wall at about 21 miles per hour, killing one person and injuring 110. A Long Island Railroad train […]Anne Barela
  • 13
    Flipper One Alternative? Free Wili 2 Multitool packs dual RP2350 chips and FPGA
    Julian Horsey writing at Geeky Gadgets covers the Free Wili 2, an open source handheld multi tool that Talking Sasquatch put through its paces on video. It runs dual RP2350 processors with FPGA support, and it is aimed at hardware development, security testing and prototyping work. The front has a 3.5-inch capacitive touchscreen, a D-pad, […]Anne Barela
  • 14
    ESP32-Bus-Expander: Expand the radio capabilities of the ESP32 Bit Pirate with 5GHz Wi-Fi and 802.15.4 protocols
    ESP32 Bus Expander is a companion firmware designed to extend the capabilities of the ESP32 Bit Pirate. It runs on an ESP32-C5 or ESP32-C6 and connects to the main Bit Pirate device via UART, adding hardware features that are not available on the primary board. The first goal of this expansion module is to extend […]Anne Barela
  • 15
    The Python on Microcontrollers Newsletter: subscribe for free
    The Python for Microcontrollers Newsletter is the place for the latest news involving Python on hardware (microcontrollers AND single board computers like Raspberry Pi). This ad-free, spam-free weekly email is filled with CircuitPython, MicroPython, and Python information that you may have missed, all in one place! You get a summary of all the software, events, projects, and the latest hardware worldwide once a week, no […]Anne Barela
  • 16
    New gpclk overlays for Raspberry Pis
    Raspberry Pi has added a pair of device tree overlays that turn one of the general purpose clock outputs, GPCLK, into a clock signal on a GPIO pin. The pint can stand in for a crystal in a project. The trick was possible years ago with a custom dt-blob.bin, which is no longer supported. One […]Anne Barela
  • 17
    NOMAD: a Cyberpunk2077 inspired guitar project
    Pio has spent over a year building NOMAD, a Cyberpunk 2077 inspired electric guitar posted on the PJRC forum. The base is a Harley Benton JA DIY kit, rebuilt around custom electronics. The pickguard is itself a large PCB running a Cypress PSoC5LP, handling pickup buffers, active hum cancelling, capacitive touch buttons, an X-Y touch […]Anne Barela
  • 18
    The 2026 Ig Nobel Awards awards nods to some different achievements
    Adam Kovac, writing for Scientific American, covered the 2026 Ig Nobel Prizes, handed out at a ceremony in Zurich on September 3. The awards go to research that sounds absurd on first hearing. The chemistry prize went to Leonard Chavas of DECTRIS Japan and Subramanian Ramaswamy of Purdue University for measuring the calorie content of […]Anne Barela
  • 19
    Rebuilding a 1995 GPS Time Server with Raspberry Pi and Python
    A 1995 TrueTime XL-AK GPS time server has been rebuilt around a Raspberry Pi 5, and the writeup was published on jeffgeerling.com. The original unit was bought in June to learn about the history of GPS-based time. Sixteen days later, a similar GPS time server took down Australia’s cell service for 12 hours. The restomod […]Anne Barela
  • 20
    Decompiler Explorer: try 13 decompilers on code
    Decompiler Explorer is a web tool at dogbolt.org that runs one binary through many decompilers at once and shows the C-like output side by side. Upload a file, pick which decompilers to run, and compare how each one reads the same code. The current lineup includes angr, Binary Ninja, Boomerang, dewolf, Ghidra, Hex-Rays, Kuna, RecStudio, […]Anne Barela
  • 21
    Building an interactive instrument for a one-of-a-kind festival
    Ben Holmen built a playable chime instrument for the Eaux Claires festival in Eau Claire, Wisconsin. He calls it Halfmoon Chimes, named for nearby Halfmoon Lake, and the 37 chimes are arranged in a half moon around the player. The notes are aluminum pipe, cut long and then shaved down millimeter by millimeter to tune […]Anne Barela
  • 22
    Microsoft 6502 Basic released as open source a year ago
    Dave Farquhar writes at The Silicon Underground that Microsoft released version 1.1 of its 6502 Basic as open source on September 3, 2025. The code went up on GitHub under an MIT license, so it is free to modify and reuse. It is 6,955 lines, written between 1976 and 1978. This is the code behind […]Anne Barela
  • 23
    Ubisoft decides Linux isn’t worth the effort
    Sourav Rudra reports at It’s FOSS that Ubisoft is dropping Linux support for its melee fighting game For Honor. Starting September 10, the game will no longer be playable on Linux desktops or the Steam Deck. The cutoff lands the same day the game’s Ranked Dominion and Ranked Duel modes go live. Ubisoft says the […]Anne Barela
  • 24
    1960s Theory That Stonehenge Was a Prehistoric Computer
    One of the earliest near-accurate depictions of Stonehenge by Lucas de Heere, image via Wikipedia Commons Though now debunked, in 1966 Gerald Hawkins put forth the theory that Stonehenge was actually an ancient computer and the ring of circles could have been used to follow the Moon, and predict eclipses. Even though future research […]Ben
  • 25
    Surprisingly STEM: Telescope Engineer
    In this episode of Surprisingly STEM from Learn with NASA, Mechanical Engineer Caroline Griffin gives viewers an inside look at the Nancy Grace Roman Space Telescope. Set your path to the stars with the Adafruit Learning System. In this project we’ll build the iconic SpaceX helmet! This helmet features an articulating top half that allows […]Stephanie
  • 26
    Developing Safer Alternatives to Lithium Batteries
    Researchers Andong Liu (left) and Dian-Zhao Lin perform experiments in Yayuan Liu’s battery lab. CREDIT: WILL KIRK / JOHNS HOPKINS UNIVERSITY via Johns Hopkins University Johns Hopkins University and the Toyota Research Institute have teamed up to research and develop better batteries. Utilizing new AI tools, the researchers tested combinations to determine what will work […]Takara
  • 27
    New Products 09/05/2026
    We’ve got the New nEw NEW for you right here This week we debuted 1 New Product. Adafruit TMF8806 Time of Flight Distance Sensor – 10mm to 5m – STEMMA QT / Qwiic Visit www.adafruit.com/new for more info.Jessie Mae
  • 28
    The Business Card That Lights Up Without Batteries
    Kevin Tang made this circuit board business card with an NFC chip and LEDs. One nifty trick of the card is that it’s battery-less; when tapped to a phone, in addition to sharing your info, the card will draw power and light up the LEDs! Here’s what’s actually happening, and it took me a […]Ben
  • 29
    Refurbishing a Tektronix TDS7104 oscilloscope
    A Tektronix TDS7104 oscilloscope picked up for $300 at the Silicon Valley Flea Market was the starting point for a detailed refurbishment. The scope is a 4-channel, 1 GHz instrument with a 10 Gs/s maximum sample rate on one channel, running Windows 2000 Pro Embedded on a RadiSys NLX motherboard with an 850 MHz Celeron. […]Anne Barela
  • 30
    Halloween Costume Detector #Raspberry Pi
    Build a wise cracking skull to MC your Halloween costume party. This project uses a Raspberry Pi 5 paired with a NoIR PiCamera and speaker bonnet. The Pi runs a Python script with OpenCV person detection. When a person crosses paths with the skull, a picture is sent to Claude with a prompt requesting a […]Takara
  • 31
    Shipping Alert: Rosh Hashanah – Friday, September 11, 2026
    Double holiday week! Rosh Hashanah begins next Friday 9/11/2026. This is an official holiday for the Adafruit team but orders will still be shipping. Please be aware! In conjunction with Labor Day at the start of the week, this makes for a tight shipping window and there may be delays. Please plan accordingly.Ben
  • 32
    Semi-Permanent E-Paper Camera with ‘Digital Film’ #piday #raspberrypi
    Camera with Digital Film, image from Strange Inventions Lab Strange Inventions created this…strange invention: a cartridge-based Raspberry Pi camera that ‘prints’ to e-paper. The 3D printed camera houses a Raspberry Pi 3 A+, camera module, battery, and microSD card. Each cartridge contains a 1.54″ 4 color module. Snap a shot! The semi-permanent stills can […]Ben
  • 33
    John Park’s CircuitPython Parsec: Fruit Jam SD Card Contents
    #circuitpythonparsec List SD card contents easily on the Fruit Jam using CircuitPython. Try out the code example To learn about CircuitPython: https://circuitpython.orgJohn Park
  • 34
    New Learn Guide: Hacking The Teenage Engineering SP-1 #Hacks #AdafruitLearningSystem @Adafruit
    The Teenage Engineering SP-1 is a hardware stem player device that was never officially released. It was manufactured in a partnership that fell apart before official release. The devices that were manufactured have ended up available in “as-is” condition through liquidation and other online stores. There is no official support or documentation for the device, […]Adafruit Learning System
  • 35
    New Learn Guide: CircuitPython Watchdog Module #Microcontrollers #AdafruitLearningSystem @Adafruit
    The CircuitPython watchdog module implements a watchdog timer that can be used to detect and recover from locked or stalled processes running on the microcontroller. The core principal is: configure the watchdog for a specified timeout period and activate it, then the code must feed the dog at least once within every timeout interval. Letting […]Adafruit Learning System
  • 36
    NOAA scientists spot an ‘outrageously tiny’ fish
    Laura Baisas writing for Popular Science reports that a NOAA Ocean Exploration team spotted a fish estimated at five millimeters (0.19 inches) to one centimeter (0.4 inches) long. It was seen in the Northeast Canyons and Seamounts Marine National Monument in the Atlantic off New England, during the Deep Connections 2019 expedition. The species is […]Anne Barela
  • 37
    OWASP launches OASIS to fix Open Source bugs at scale
    Apurba Sen, writing at Open Source For You, reports that OWASP has launched the Open Automated Security Initiative for Software, or OASIS. It was announced on August 26, 2026, in San Francisco. The goal is to move open source vulnerability work from “find and report” to “find, validate and fix.” OASIS uses AI to scan […]Anne Barela
  • 38
    NASA selects Blue Origin as Mars telecommunications network provider
    Jessica Taveau writing for NASA reports that the agency has awarded Blue Origin a contract to build the Mars Telecommunications Network. The firm-fixed-price contract has a maximum potential value of about $700 million, and calls for delivery of a high-performance Mars telecommunications orbiter no later than December 31, 2028. Blue Origin will design, develop, integrate, […]Anne Barela
  • 39
    The Python on Microcontrollers Newsletter: subscribe for free
    The Python for Microcontrollers Newsletter is the place for the latest news involving Python on hardware (microcontrollers AND single board computers like Raspberry Pi). This ad-free, spam-free weekly email is filled with CircuitPython, MicroPython, and Python information that you may have missed, all in one place! You get a summary of all the software, events, projects, and the latest hardware worldwide once a week, no […]Anne Barela
  • 40
    datasette-mcp 0.2 connects Datasette to MCP clients
    Simon Willison has released datasette-mcp 0.2, announced on his weblog. It is the first non-alpha release of the plugin, which exposes Datasette to Model Context Protocol clients. Datasette is a tool for exploring and publishing data. It helps people take data of any shape or size and publish that as an interactive, explorable website and […]Anne Barela
  • 41
    Cursed Skull Desk Lamp – Bambu LED Kit 001 with Golden Flame Cutout #3DThursday #3DPrinting
    user_3081978295 shares: A macabre yet elegant skull-shaped desk lamp perfectly sized for the Bambu Lab LED Lamp Kit 001. Internal flame-shaped cutout creates dramatic light play — the captivating golden glow is the reflection of the warm-white LED on matte black PLA. Assembles entirely without glue. Integrated cable hole at the back of the skull […]Pedro
  • 42
    ESP32-CAM Case Kit – Snap-Fit with Ball Joint + Magnetic Mount #3DThursday #3DPrinting
    Shared by i-Boxit on Makerworld: A comprehensive ESP32-CAM case kit — micro-USB and USB-C variants, snap-fit closure, optional USB mount, boot/reset buttons, LED hole, lens plate for Classic and M12 lenses, and ball mount stand (optional 10×2mm magnetic). Wall mount available (screw or 8×2mm magnetic). Fix the ball joint to the case with a piece […]Ben
  • 43
    John Park’s Workshop — LIVE TODAY 9/3/26
    It’s JOHN PARK’S WORKSHOP — LIVE! — Coming up at 4pm ET / 1pm PT Today (the usual time, no change to the Workshop show schedule)! LIVE TEXT CHAT IS HERE in the Adafruit Discord chat! Projects: Arcade Ports & Fruit Jam Arcade Pad design CircuitPython Parsec: SD Content Reader Learn Guide Recap Tool Tips: […]John Park
  • 44
    Raspberry Pi 5 Cyberdeck – DIY Handheld with Powerbank #3DThursday #3DPrinting
    MadHias shares: A DIY handheld cyberdeck designed by the maker specifically for their own Raspberry Pi 5 (not based on Compute Modules, which are more expensive). High-resolution display for actual work capability. Powerbank battery (5V 3A) with the option to swap in the official Pi power supply for heavy loads (no unscrewing required). Free access […]Pedro
  • 45
    ESP32 38-Pin WLED Case – Snap-Fit for ESP32-WROOM DevKit #3DThursday #3DPrinting
    Shared by drathbone on MakerWorld: Snug snap-fit case for an ESP32-WROOM 38-Pin devkit USB-C. Hole at the back for a cable (can easily fit a 5mm wide 3-core or anything similar). … Case comprises of a base, with room for pins and cables, a tray for the ESP32 USB-C to sit in, and a click-lock […]Ben
  • 46
    Hex LED Desk Lamp – WLED Honeycomb Geometry with FCOB RGBW Strip #3DThursday #3DPrinting
    printspire3d shares: A sleek modern lamp inspired by honeycomb geometry — interconnected hexagonal frames form a minimalist eye-catching pattern. Central structure houses an LED strip for vibrant multi-color effects. Circular base with a built-in groove for an additional LED ring for underlighting. Uses FCOB RGBW IC LEDs (WS2814/SK6812, 784 LEDs, 10mm, 24V) with WLED. Four […]Pedro
  • 47
    Haunted House – A Halloween Scene with LED Integration #3DPrinting #3DThursday
    A layered display piece featuring a haunted house glowing from within, surrounded by pumpkins, lanterns, and a gargoyle keeping watch at the gate. Each part prints separately for perfect color control and easy assembly. Three lighting options: LED Lamp Kit from MakerSupply for simple setup, or components from the BOM for full illumination (lanterns, pumpkins, […]Jessie Mae
  • 48
    PiDeck – 7-Inch Cyberdeck for Raspberry Pi 5 with Waveshare IO + NVMe HAT #3DThursday #3DPrinting
    Screase3D shares: A rugged, functional, hacky-looking Raspberry Pi 5 cyberdeck built around a Waveshare IO Board and NVMe HAT. Fully bolted design (M2.5 hardware). 7″ display. Designer notes this is ‘not for beginners’ — specifically designed around parts on hand, so tolerances are strict. CAD (.step) files provided so you can modify the housing to […]Pedro
  • 49
    Cyber Loop Lamp – Futuristic Sci-Fi WLED Light with 300 Pixels
    A dramatic futuristic sci-fi lamp powered by ESP32 + WLED, controlling nearly 300 LED pixels of a 144-LED/m high-resolution strip. Three independent light areas (each with its own data output for easier wiring). Funnel with 12 identical light segments (8 pixels each in parallel). Print the housing in darker filament and inner diffusers in PLA […]Jessie Mae
  • 50
    The Core Lamp – Modular LED Cube with PC Unplugged + Levitating Variants #3DThursday #3DPrinting
    ScarfMeadow shares: A modular LED core lamp with three variants: standard ‘The Core’, ‘PC Unplugged’ (same design in a different format), and a levitating cube version. Featured in the RGB variant with larger, cleaner design. Multiple builds from the same file library — pick what fits your desk aesthetic. By ScarfMeadow download the files on: […]Pedro
共 50 条热搜查看完整榜单

AIbase

实时热榜
1小时前更新
共 8 条热搜查看完整榜单
11分钟前更新
共 9 条热搜查看完整榜单
3小时前更新
  • 01
    Developing Enterprise Frontier Safeguards with our customers
    Today we’re announcing Enterprise Frontier Safeguards (EFS), a solution that combines the privacy of zero data retention (ZDR) with state-of-the-art safeguards for detecting misuse. EFS works by storing data in cloud infrastructure controlled by the customer, not Anthropic. EFS will be rolling out to customers in phases, starting later this fall. To make the transition smooth, eligible customers will receive ZDR on Fable 5 and Fable 5.1 until EFS is ready. We developed EFS in close collaboration
  • 02
    Improving our alignment and security efforts
    On July 30, we reported three incidents in which Claude models gained unauthorized access to real computer systems. The models—intentionally running without cyber safeguards for evaluation purposes—accessed the internet due to a misconfiguration inside a third-party evaluation environment. Separately, on August 4, the UK AI Security Institute reported an incident from its own cybersecurity testing, in which Claude Mythos 5 took a series of unauthorized actions on the live internet. In that case,
  • 03
    Previewing the Model Hardware Standard
    We’re opening a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to safely operate physical devices, to a first group of scientific research labs and advanced manufacturers. MHS enables AI agents to operate multiple lab and manufacturing instruments, such as microscopes, liquid handlers, and robotic arms, in parallel, and perform intricate tasks ranging from routine drug discovery experiments to laser calibration on a quantum computer. The development o
  • 04
    Expanding our support for scientists
    As Claude becomes increasingly capable at scientific research, we are focusing on building products and programs to support the research community. In June, we launched Claude Science , a product that integrates the tools that researchers most commonly use, produces auditable artifacts, and provides flexible access to computing resources. We have also continued to broaden our AI for Science program , which provides free credits to researchers working on high-impact scientific projects. Starting
  • 05
    Funding better evaluations of AI’s impact on wellbeing
    We’re launching a $5 million grant program to fund independent research into how AI impacts users’ wellbeing. The program will provide direct funding, access to our models, and technical support to grantees building open-source evaluations that help the AI industry measure how our models affect those who use them. Grantees will work fully independently, and will publish their work as open-source projects that any developer can make use of. AI systems have become central to how many people work,
  • 06
    How Claude’s text watermark works
    Future Claude models will generate text that contains a watermark. This is a way of determining the likelihood that Claude was involved in writing the text, and we, along with several other major AI providers, are implementing this change to comply with the EU AI Act. In this article, we share answers to some of the questions we’ve received about how our chosen watermarking method works, whether it affects Claude’s outputs, and why we’re making this change. To summarize: We use a method of water
  • 07
    Improving Fable 5's biology safeguards
    We’re making updates to Claude Fable 5’s biology safeguards in a way that substantially reduces false positives. Fable 5 users will now experience many fewer “fallbacks”—where the system switches to a less capable model after they make a biology-related query. In our testing, this update reduced biology-related fallbacks by about 85% across our product surfaces. 1 Fable 5 will thus be able to assist with a wider range of biology tasks. In practice, users should see far fewer fallbacks on everyda
  • 08
    Mariano-Florentino (Tino) Cuéllar to join Anthropic as Chief Global Affairs Officer
    Mariano-Florentino (Tino) Cuéllar will join Anthropic as its first Chief Global Affairs Officer, leading the company’s work on policy, strategic international engagement, and government relationships worldwide. Tino’s career spans law, technology, international security, and public institutions at the international, national, and state levels. He recently stepped down as President of the Carnegie Endowment for International Peace, a leading independent global policy research institution with sch
  • 09
    Investigating three real-world incidents in our cybersecurity evaluations
    In a review of our cybersecurity evaluation transcripts, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations. Below we describe what happened, how it happened, and what we’re changing. We encourage other AI labs to perform similar reviews. This post reflects our current understanding; we'll update it if any details
  • 10
    Our position on open-weights models
    A post by Dario Amodei, Anthropic CEO Over the last few days there has been a lot of discussion about open-weights models, especially those from China. Reports suggest that some US officials are considering banning the use of Chinese open-weights models by US companies. In response, many tech companies have signed a letter supporting open-weights models, and some people have even accused Anthropic of wanting to ban open-weights models as a means of protecting our business. Anyone who has read my
共 10 条热搜查看完整榜单

小众软件

实时热榜
15分钟前更新
  • 01
    给别人发密码、API 密钥，不想留在聊天记录里？试试 binthere
    临时把密码或 API 密钥发给别人，你一般怎么发？直接贴进微信、邮件，事情办完了，但密码却还留在聊天记录里。 另一种办法是使用 Pastebin 类文本分享服务：粘贴内容，生成链接，再把链接发出去。 但如果是密码、API 密钥、恢复代码，就会有几个问题： 尤其在 API 密钥本身就是现金的情况下&#青小蛙
  • 02
    压箱底办公神器：一键搞定Word、Excel、PDF、PPT、图片的批处理工具，效率翻倍！
    你是否也曾被这样的场景折磨到崩溃： 一个简单的PDF转word，找的工具要么收费，要么限制超多；100份Word合同，都要统一修改公司信息，手动改到崩溃；上百份图片裁剪、压缩，头都大了…… 每天处理一堆不同格式的文档，重命名、转格式、去水印、压缩，批量替换内容，手动操作真心累青小蛙
  • 03
    Pushin – 没有美国参与的 Git 托管服务
    还记得前几天介绍的：Radicle – 免费去中心化代码托管，GitHub、GitLab 的替代品，作为一款去中心化的 Git 服务，青小蛙表示也不敢用啊。@Appinn 今天又看到一个选择： Pushin.eu 一个来自欧洲的 Git 代码托管平台 我看到它把没有美国参与列到第一特性，就想笑： 想青小蛙
  • 04
    发现频道：10款大家发现的好评软件[2026年第36期]
    最近10日，来自小众软件论坛的发现频道的热门排行榜，由系统自动生成，直接列出来： 序号 主题 1️⃣ 【开发者自荐】macOS 一边打字一边翻译来学习英语 2️⃣ 【开发者自荐】BongoCat – 开源桌宠实时响应鼠标和键盘事件，使用纯C彻底重新实现 3️⃣ 【开发者自荐】Snippe青小蛙
  • 05
    WinToastRelay：将 Windows 系统通知转发到手机｜Codex 跑完，手机会叫你
    在 Windows 用 Codex 时，会遇到一个小问题： 有些任务几分钟就结束，有些却要跑很久。一直守在电脑前没必要，但去做别的事情，又总忍不住隔一会儿回来看看： “跑完了吗？” 既然 Codex 在任务完成或出错时都会发送 Windows 系统通知，能不能把这些通知直接推送到手机上？ WinTo青小蛙
  • 06
    uBlock Origin 回来了！第三方开发者将其完整功能移植到 Manifest V3
    uBlock Origin 回来了！ 在 Chrome/Edge 相继淘汰了 Manifest V2 标准的老旧插件之后，首当其冲就是强大的广告过滤插件 uBlock Origin 被淘汰。原作者也不太愿意折腾，但终于有人出手了！@Appinn 一位名叫 r58Playz 的开发者，在原 uBloc青小蛙
  • 07
    本周赛博领鸡蛋（9.4~9.10）：《与你独处》《寻找埃文》
    本周赛博领鸡蛋 Epic Games 商城共有 1 款电脑游戏、1 款手机游戏限时免费领取，分别是电脑端的《与你独处》，以及手机端的《寻找埃文》。其中《寻找埃文》同时提供 Android 和 iOS 版本。截止 9 月 10 日 09:00。 《与你独处》/ Alone With You 《与你独处青小蛙
  • 08
    Radicle – 免费去中心化代码托管，GitHub、GitLab 的替代品
    代码一定要保存在 GitHub、GitLab 上？或者托管在某一家商业公司的服务器上吗？ 无论是 GitHub 还是 GitLab，都曾出现因政策、制裁或地区限制导致用户封号的情况发生。甚至青小蛙自己就遇到过：有开发者上午刚在发现频道提交自己的项目，下午再点进去，GitHub 项目已经 404 了。青小蛙
  • 09
    高性价比（廉价）SSD VPS 提供商 CloudCone 补货，最低 96 元人民币/年
    CloudCone 发来了最新邮件，提醒最近的 Turns 9 Sale 活动补货，最低档只需要年付96元（14.24美元），支持支付宝。 推荐两个套餐 （均包含1个IPv4和3个IPv6） 套餐（可点击购买） 配置（CPU / 内存 / SSD / 流量） 价格（美元 / 人民币） 👉 CCTUR青小蛙
  • 10
    Windows 11 任务栏位置终于移动到屏幕四周了，开始菜单大小也可以更改
    Windows 11 8月份可选更新 KB5120998 带来了两个喜闻乐见、大家都期待的功能：任务栏位置可调上下左右、开始菜单可调大小以及丰富的自定义功能。@Appinn 如果你不使用自定义鼠标样式，可以重新设置一次壁纸，那么还是推荐更新的，因为有不少大家想要很久的功能。 （是的，请注意这个更新会青小蛙
共 10 条热搜查看完整榜单

Apple

Newsroom
7小时前更新
共 20 条热搜查看完整榜单

Apple Podcasts

中国区 Top 100
12分钟前更新
共 100 条热搜查看完整榜单

AppleInsider

实时热榜
6分钟前更新
  • 01
    Apple planned for $1,999 iPhone Ultra, but then the RAM crisis hit
    The iPhone Ultra was originally intended to be a premium model costing consumers $2,000. That was still the case until the memory and flash crisis became a bigger problem. A render of the iPhone Ultra - Image Credit: AppleInsider There's no question that the iPhone Ultra will be an expensive device. With a starting price thought to be at $2,099 by analysts and with upper-specification models hitting $3,000, it will be the most premium-priced iPhone ever. While it seems expensive, and even more sMalcolm Owen <malcolm@appleinsider.com>
  • 02
    A folding iPhone was Tim Cook's decade-long passion project
    For a man known to not be a product guy, it was Tim Cook who pushed hard for 10 years to perfect a foldable iPhone before he stepped down. Tim Cook during his 2019 Asia tour where he reportedly saw many people with foldable smartphones - image credit: Apple. It was the biggest criticism Steve Jobs ever had of Tim Cook, that he didn't share the same passion for products that either Jobs or Jony Ive did. Cook certainly never got as involved with any Apple designs as they did, or so it seemed. NowWilliam Gallagher <wg@williamgallagher.com>
  • 03
    15-inch MacBook Air M5 plunges to $1,349 ahead of Apple Event
    Amazon's $1,349 15-inch MacBook Air deal delivers the lowest price of the season. Upgraded models are $150 off as well. Save $150 on several M5 MacBook Air models ahead of the Apple Event - Image credit: Apple Save $150 on Apple's 2026 15-inch MacBook Air, bringing the price of the M5 model with 16GB of RAM and 512GB of storage down to $1,349. Buy M5 15" MacBook Air for $1,349 Continue Reading on AppleInsider | Discuss on our ForumsChristine McKee <christine@appleinsider.com>
  • 04
    OWC's new Thunderbolt 5 dock connects two HDMI displays without adapters
    OWC is expanding its Thunderbolt 5 lineup with a $350 dock offering dual HDMI and 140W charging. OWC Thunderbolt 5 Dock with Dual HDMI On Tuesday, OWC announced an upcoming Thunderbolt 5 Dock with Dual HDMI, compatible with both Mac and PC. The dock is aimed at creative and business professionals looking to maximize performance and streamline workflows. The new dock includes two HDMI 2.1 ports for connecting displays without separate adapters. It supports two displays at up to 4K 240Hz or 8K 60HAmber Neely <amber@appleinsider.com>
  • 05
    iPhone Ultra release in China faces delay over eSIM issues
    Apple users in China wanting to buy the iPhone Ultra may have a longer wait than the rest of the world, all apparently due to issues with carriers. A render of Apple's foldable iPhone - Image Credit: AppleInsider The release of new Apple products tends to bring them out to major markets first before expanding into others. However, while China is an important market, it may not be among the first to get the latest device. According to Weibo leaker Instant Digital on Tuesday , Apple's shipping timMalcolm Owen <malcolm@appleinsider.com>
  • 06
    iPhone Air line may replace the entry-level iPhone, but not soon
    A leaker with no real track record to speak of is claiming that Apple is going to drop the expected iPhone 18 and instead give that name to what was the iPhone Air 2 . A move like this may happen someday, but not in 2027. iPhone Air 2 is predicted to have two cameras It's previously been reported that the iPhone 18 will closely resemble the performance and hardware of the iPhone 18e. But now one of the leakers who claimed that is now saying that Apple is dropping its iPhone 18 design. Instead, aWilliam Gallagher <wg@williamgallagher.com>
  • 07
    Apple acquires brain imaging firm for health and accessibility features
    Apple has bought Sonera, a California company that makes non-invasive sensors that can measure neural signals from the brain, with health benefits ranging from muscle monitoring to active thought control of devices. Nishita Deka (left) and Dominic Labanowski, co-founders of Sonera - image credit: Sonera. It's already been proven possible to control an Apple Vision Pro by thought, but it involved an implant in a user's head. Now Apple has acquired Sonera, whose technology detects magnetic muscleWilliam Gallagher <wg@williamgallagher.com>
  • 08
    Code in iOS 27 seems to prove iPhone Ultra will use Touch ID
    It's been widely predicted that the iPhone Ultra 's chassis will be too thin for Face ID, but now code references in the iOS 27 beta back up the presumption that Touch ID will be used instead. iPhone Ultra render Face ID was introduced with the iPhone X in 2017 and Apple has been steadily adding it to devices ever since. There has been one notable exception with the iPad Air sticking to Touch ID , and now it appears confirmed that the iPhone Ultra will be the same. According to prolific code exaWilliam Gallagher <wg@williamgallagher.com>
  • 09
    'Cupertino' legal drama doesn't represent region demographics, landmarks, or reality
    The trailer for an upcoming legal drama for CBS called "Cupertino" suggests it knows where the show takes place, but doesn't offer much else that represents the region where Apple calls home. You won't spot Apple Park in the 'Cupertino' trailer No, Cupertino isn't coming to Apple TV , but it does appear to be a drama hoping to showcase legal battles between tech giants and their employees. However, reports suggest that the trailer leaves a lot to be desired in representing the city. According toWesley Hilliard <wes@appleinsider.com>
  • 10
    High prices and no base iPhone 18 could hurt Apple but only in the short term
    Keybanc has historically had an Apple stock price target below reality, and its latest commentary about a lack of a non-Pro iPhone debut on Wednesday ignores a few key factors. Mockup of a possible iPhone 18 Pro Max, which is rumored to come in a dark cherry red. When Apple launches its new iPhones on September 9, 2026 , it's expected to make two key changes. As well as the predicted announcement of a foldable iPhone Ultra, Apple is not believed to be launching its usual base iPhone models untilWilliam Gallagher and Mike Wuerthele <news@appleinsider.com>
  • 11
    Apple TV's dystopian drama 'Silo' could live on with a rumored spinoff
    A rumored "Silo" spinoff could keep Apple TV's dystopian franchise alive beyond its planned finale, but the books' author is pushing back on the claim. Image Credit: Apple TV "Silo" is a dystopian science-fiction drama about a society living underground after humanity has been driven from the surface. The Apple TV series is based on the bestselling "Silo" trilogy, "Wool," "Shift," and "Dust," by Hugh Howey. While it was well liked, Apple TV had told the creators that the series would end after tAmber Neely <amber@appleinsider.com>
  • 12
    watchOS 27 will still have surprises, even if Apple Watch Series 12 won't
    When Apple unveils its Apple Watch Series 12, only minor improvements are expected. Newly discovered code suggests that the surprises will be in watchOS 27 instead of the hardware. Apple Watch Series 11 health sensors won't change much in the Series 12, but watchOS will hold the surprises Apple has been steadily releasing new betas since WWDC , and they have included over 50 improvements in watchOS 27. Now, however, it appears that the most significant updates have remained hidden. In hardware,William Gallagher <wg@williamgallagher.com>
  • 13
    Disconnect your LG television from the internet, now
    If you've got an Apple TV set-top box, that's not enough to protect your privacy if you have an LG television. Unplug Ethernet and tell it to forget your Wi-Fi information. An LG TV playing Apple Music Smart televisions have a history of being a privacy problem for users. Repeated reports have surfaced over the years about them logging how you use the television, as well as misusing sensors intended to monitor the viewers. In a September investigation by Gamers Nexus posted to YouTube, LG's smarMalcolm Owen <malcolm@appleinsider.com>
  • 14
    Apple is packing more pro photography tools into iOS 27
    Apple appears to be preparing a slate of more advanced photography tools for the iPhone , including exposure scopes, focus peaking, and expanded manual focus controls. The camera plateau on the current iPhone 17 Pro Max We already know that Apple is improving the Camera app in iOS 27 . Now, new beta code gives us a sneak peek into what we can expect in the fall. The references appeared in iOS 27 beta code, first spotted by pdfu, a MacRumors forum regular. Five new features could appear at launchAmber Neely <amber@appleinsider.com>
  • 15
    Amazon knocks $100 off AirPods Max 2 ahead of Apple Event
    Amazon's $100 discount on AirPods Max 2 over-ear headphones brings the price down to $449. Save $100 on AirPods Max 2 during Amazon's Labor Day sale - Image credit: Apple AirPods Max 2 are $100 off at Amazon for Labor Day, with four of the five colors eligible for the discount at press time. Buy AirPods Max 2 for $449 Continue Reading on AppleInsider | Discuss on our ForumsChristine McKee <christine@appleinsider.com>
  • 16
    iPhone 18 Pro rumors, new Sonos gear, & more on Smart Home Insider
    On this week's episode of the Smart Home Insider podcast, Matic co-founder Mehul Nariyawala joins to round up the Apple Event rumors and talk about why Matic's robot is so radically different. Smart Home Insider Podcast This is a jam-packed episode of the show as it coincides with not only Apple's iPhone event, but it is the end of IFA. Mehul brings a wealth of knowledge as we break it all down, including what we want to see Apple announce. After the Apple event coverage, we pivoted to discussinAndrew O'Hara <andrew@appleinsider.com>
  • 17
    Apple takes home 20 wins in first two nights of the 2026 Emmy Awards
    Apple TV 's comedy-horror hit "Widows Bay" took home eight wins, the most Emmy Awards of any program, helping Apple secure a record 20 wins at the 78th Emmy Awards. "Widow's Bay" wins eight Emmy Awards The awards keep on rolling in as Apple TV leads as the most-awarded network on the second night of the 78th Emmy Awards. Unsurprisingly, the network's new darling, "Widows Bay," brought in nearly half the wins. Also making strong showings were Apple's post-apocalyptic black comedy "Pluribus," andAmber Neely <amber@appleinsider.com>
  • 18
    Inside iMovie: Getting started with video editing in iOS 27 and iPadOS 27
    Take your first steps to becoming the next Mr Beast, or make your vacation recap more bearable to watch, by using iMovie. Here's how to get started using it on iPhone and iPad. iMovie is free in iOS 27 and iPadOS 27 Video is a type of media that most of the world enjoys every day. It's a fantastic tool for sharing ideas and experiences with others, and in a way that can be much easier to digest than text or photographs alone. Social Media has a massive appetite for video, with almost all platforMalcolm Owen <malcolm@appleinsider.com>
  • 19
    Viral in-store iPhone 18 Pro Max video is fake
    A leaker with a mixed but usually better than average track record, is now making the absurd claim of having already handled the iPhone 18 Pro Max in a store, and has made a video to back it up. The problem is, demo units aren't in stores yet, and won't be for some time. A mock-up of a Dark Cherry iPhone With such a short time to go before Apple unveils its new iPhones , there's already been a carrier posting optimistic details of the lineup. Now leaker yeux1122 has posted a video purporting toWilliam Gallagher <wg@williamgallagher.com>
  • 20
    Best of IFA for Apple users: DJI, Belkin, Lexar, Aqara, Philips Hue, & more
    The 2026 IFA is over, and unlike CES, nearly all of this stuff will ship soon. Here's the best of the announcements that we saw, that will appeal to Apple users. IFA 2026 took place in Berlin and was full of new gadgets and accessories for Apple users Taking place in Berlin, IFA is an annual tech show that includes countless new product advancements. Unlike CES, it's open to the public and takes place over the course of a week. Also unlike CES which has a reputation for vaporware, many of the prAndrew O'Hara <andrew@appleinsider.com>
  • 21
    Desktop Mac buyer's guide: Which Mac to buy in fall 2026
    Apple's unexpected August update to the Mac mini and Mac Studio has changed the pricing of the models in a big way. Here's the best desktop Mac you should buy for your budget in fall 2026. Apple's current desktop Mac range. In August, while the world was preparing for the expected launch of new iPhone models in September , Apple had a trick up its sleeve. Without any warning, it launched changes to the Mac mini and Mac Studio near the end of August. Chiefly, the change was to introduce a new M6Malcolm Owen <malcolm@appleinsider.com>
  • 22
    Trying to squeeze more profit out of the App Store is a bad idea
    A new report says that Phil Schiller left the App Store because Apple aims to make more money from it. Read the room, Apple. Phil Schiller is reportedly concerned about planned changes to the App Store. Phil Schiller remains with the company as an Apple Fellow, but he recently left his roles as head of events and the App Store . Now according to Bloomberg , a reason for his leaving the App Store was that he doesn't want any part in what's coming to it. Why Apple legend Phil Schiller is moving toWilliam Gallagher <wg@williamgallagher.com>
  • 23
    Tim Cook won't be presenting the iPhone 18 Pro event
    While Tim Cook is no longer the CEO of Apple, he isn't going anywhere soon. Just don't expect him to take part in the September iPhone event video. Apple CEO John Ternus [left] and previous CEO Tim Cook [right] The CEO transition from Tim Cook to John Ternus is continuing to roll on. However, while Cook isn't leaving Apple, his change in role to the chairman of the board of directors means that he isn't going to be the usual public-facing executive this time around. Writing in Sunday's "Power OnMalcolm Owen <malcolm@appleinsider.com>
  • 24
    Before iPhone 18 Pro event, carriers don't know any more than you do
    A Korean carrier has published details of the forthcoming iPhone 18 Pro range and the expected iPhone Ultra. They're from the rumor mill. They look good, but the are AI-generated - image credit: LGT Shop. They don't tend to, but carriers could get early information about Apple launches, and it's not unknown for details to be published early. Typically such leaks are rapidly taken down, presumably after angry phone calls from Apple, but this latest one on LGT Shop continues to be live. That mightWilliam Gallagher <wg@williamgallagher.com>
  • 25
    Sunday Reboot: Job changes at the top and very safe hands
    In this week's Sunday Reboot, the sudden top-table rule changes and how John Ternus will be a good Apple CEO, whether he likes it or not. Apple CEO John Ternus. Sunday Reboot is a weekly column covering some of the lighter stories within the Apple reality distortion field from the past seven days. All to get next week underway with a good first step. Continue Reading on AppleInsider | Discuss on our ForumsMalcolm Owen <malcolm@appleinsider.com>
  • 26
    Play iPhone 18 Pro Apple Keynote Bingo with this app
    Use this Bingo Card Generator to play along with Apple's September iPhone event. Drinking is entirely optional. Make the Apple event more fun with the Bingo Card Generator Apple is well known for the format of its special events. From the WWDC keynote to the usual September iPhone events, viewers are familiar with the usual tropes, such as Tim Cook 's "Good morning" at the start, and the earlier famous "One more thing" at the end of some videos. While it would be easy for people to turn it intoMalcolm Owen <malcolm@appleinsider.com>
  • 27
    Asahi Linux rolls out support for M3 Apple Silicon
    The Asahi Linux project now officially supports Macs with an M3, M3 Pro, or M3 Max chip, bringing the operating system to more Apple Silicon desktops. Asahi Linux now supports M3. Just not M3 Ultra. In late August , a blog post for the Asahi Linux project, which intends to bring Linux to Apple Silicon and run natively on Apple's ARM chips, said M3 support was on the way. While it said that it was a matter of weeks, it seems to have arrived in a much quicker timeframe. A blog post on September 6Malcolm Owen <malcolm@appleinsider.com>
  • 28
    Crime blotter: Last two defendants sentenced in counterfeiting scheme
    A woman is robbed after visiting the Apple Store; a fake NFL swindler used apps and FaceTime calls, and an AirTag signal led to the rescue of 20 cats, all in this week's Apple Crime Blotter. The Memorial City Apple Store The latest in an occasional AppleInsider series, looking at the world of Apple-related crime. Continue Reading on AppleInsider | Discuss on our ForumsStephen Silver <stephen@appleinsider.com>
  • 29
    Best Labor Day deals: Save up to $500 on Macs, AirPods, Apple Studio Display
    Labor Day savings across Apple's product lines are in full swing this weekend, with savings of up to $500 off current MacBook Pros and even the brand-new Mac mini. Save up to $500 on Apple products this Labor Day weekend - Image credit: Amazon, Apple Save up to $500 on Apple products for the Labor Day weekend at Amazon. Highlights include preorder discounts on the 2026 Mac mini, along with steep markdowns on Apple's latest MacBook Pros. You can also save on AirTags and even the Apple Studio DispChristine McKee <christine@appleinsider.com>
  • 30
    Apple may have two very different game controllers in the works
    Apple already supports nearly every major game controller, but new evidence suggests that the company may want to add its own hardware to the mix. Game controllers with a MacBook Pro Apple may be looking to expand its ambitions beyond software and platform support. Newly discovered code references two unreleased controllers with notably different feature sets. The references appeared within code briefly included in macOS 26.7 , first spotted by pdfu, a MacRumors forum regular. The controllers inAmber Neely <amber@appleinsider.com>
  • 31
    This is the folding iPhone, if case manufacturers are right
    Ahead every iPhone release event, case manufacturers generate dummy units to see if their cases will fit properly. We've got our hands on these faked iPhone 18 Pro and folding iPhone models, so let's take a look. Fake dummy units of Apple's rumored upcoming iPhones The dummy unit "industry" is bigger than ever this year, as you'd expect when there are three flagship models coming out, not just two. In recent weeks we've seen 3D prints and metal blanks, of what case manufacturers think that iPhonMike Wuerthele <mike@appleinsider.com>
  • 32
    Apple still on track for a revolutionary iPhone 20 in 2027
    We're approaching the dawn of the iPhone 18 Pro, but a new report is addressing what to expect from the iPhone 20 . Here's what to expect, so far. The 20th anniversary iPhone will have more glass From the earliest days of Apple's iPhone there has been that graphic showing the sea change it brought to smartphones. Before it, nothing looked like the iPhone, and after it, everything looked like it. It wasn't as clear cut when the iPhone celebrated its tenth anniversary with the iPhone X. But when yWilliam Gallagher <wg@williamgallagher.com>
  • 33
    Apple's home security devices may now launch in 2027
    Summarizing existing rumors about Apple launching an AI home security camera, a new report says it will be announced in 2027 alongside a monitoring service. Apple is expected to release a home security system that includes a home hub (left) and cameras such as this HomeKit-compatible Amazon one (right). Apple has not had the best luck so far with security devices, having effectively abandoned HomeKit Secure Routers after few vendors supported it. But it was one of several signs that Apple was aiWilliam Gallagher <wg@williamgallagher.com>
  • 34
    CEO John Ternus, legal battles, & iPhone Fold on the AppleInsider Podcast
    Apple's CEO transition is complete, OpenAI may have been caught red-handed, and Apple Maps had a controversial update on the AppleInsider Podcast. John Ternus is now CEO and the AppleInsider podcast has to bring back Controversy Corner - original image credits: Apple. It's been a busy week for Apple fans and the iPhone 18 event hasn't even happened yet. Tim Cook is now Executive Chairman while John Ternus is the new CEO. His new era has begun with a lot of controversial things that aren't exactlWesley Hilliard <wes@appleinsider.com>
  • 35
    Alogic's newest monitors include a 6K touchscreen and 5K ultrawide
    Alogic has announced six new monitors, including new 5K and 6K additions to its Aspekt Touch line that work with the Mac mini docking stand, as well as a new 5K ultrawide model with touchscreen support. Some Alogic monitors are perfect for Mac mini owners The new monitors also include models aimed squarely at the professional and business markets, including a 34-inch ultrawide curved option. Availability of the newly announced monitors varies. Some will be available soon, others will not go on sOliver Haslam <oliver@monkeymanmedia.com>
  • 36
    Wokyis G7 Thunderbolt 5 dock - Retro styled, speedy connectivity
    While the Thunderbolt 5 Wokyis G7 dock certainly goes after the retro gaming nostalgia vote, it also happens to be a well-thought-out add-on for your Mac Studio or older Apple Silicon Mac mini. Wokyis G7 80Gbps I am not a stranger to Mac docks. I suspect between myself and the rest of the staff, we've done the most reviews of them out of anybody facing the internet over the last decade. And, I'm not a stranger to nostalgia. Continue Reading on AppleInsider | Discuss on our ForumsMike Wuerthele <mike@appleinsider.com>
  • 37
    Apple's suppliers only made a few hundred iPhone Ultra units per day in August
    Mass production of the iPhone Ultra has been underway for at least some weeks, but quality control inspections are reportedly rejecting all but a few. The expected iPhone Ultra is still believed to be announced at the September 9 event, but supplies may be extremely constrained. Reports have varied for months over whether the iPhone Ultra is delayed because of manufacturing problems , or whether those have been resolved . Now according to Nikkei Asia , however, production has been very slow. RepWilliam Gallagher <wg@williamgallagher.com>
  • 38
    HomePods may finally get surround sound through four connected speakers
    Code discovered in audioOS 27, the HomePod operating system, shows references to a four-speaker setup, new calibrations, and a "Party Mode." A four HomePod surround system may be coming soon HomePod owners have been clamoring for a more robust sound system since the original model debuted in 2017. Stereo pairs are phenomenal, but lack the robust audio capabilities of full surround sound systems. According to a post on the MacRumors forums from user pdfu, the code is available in audioOS 27. It cWesley Hilliard <wes@appleinsider.com>
  • 39
    Global chemical giant BASF sues Apple over Face ID tech
    Nearly a decade after Face ID debuted in the iPhone X, BASF is bringing a lawsuit against Apple for violating face authentication technology patents it holds. Apple must face yet another patent infringement lawsuit Many versions of the iPad and nearly every iPhone since the iPhone X has used Face ID. For whatever reason, BASF chose today to pursue a patent infringement lawsuit. According to a report from Reuters , the German-based BASF company has started a lawsuit against Apple. They allege patWesley Hilliard <wes@appleinsider.com>
  • 40
    Apple TV picks up sci-fi thriller 'Waffle House Index' in seven-figure deal
    Apple TV is betting on the Waffle House Index for a sci-fi disaster flick that takes the restaurant's reputation for staying open to its logical extreme. Apple lands disaster film 'Waffle House Index' "Waffle House Index" follows a Waffle House waitress as a series of mounting disasters unfold during her night shift. First, a man with a gun, then a flood, followed by killer crocodiles, nuclear explosions, and yes: even aliens. Inde Navarrette, known for her role as Nikki Freeman in the horror fiAmber Neely <amber@appleinsider.com>
  • 41
    M4 Macs can share local AI work with PCs using Nvidia PAIR
    Got an M4 Mac ? Nvidia's new free tool lets it offload local AI tasks to your PC, turning your home network into a shared supercomputer. M5 MacBook Pro The Personal AI Router (PAIR) isn't hardware. Instead, it's software that connects supported Apple Silicon Macs, Nvidia RTX PCs, and DGX Spark systems. PAIR works with Ollama and LM Studio, two tools for running AI models locally. It discovers participating computers and gives AI apps a single connection for sending requests, so users don't haveAndrew Orr <andreworr@appleinsider.com>
  • 42
    Apple Studio Display drops to $1,499 with Labor Day deals
    Labor Day monitor sales at B&H and Amazon include discounts on the 2026 Apple Studio Display, with prices falling to as low as $1,499. Grab the lowest prices of the season on the Apple Studio Display. The Labor Day deals at Amazon and B&H deliver the lowest prices of the season on the Apple Studio Display that was released in March 2026. Grab Apple Studio Display deals Continue Reading on AppleInsider | Discuss on our ForumsChristine McKee <christine@appleinsider.com>
  • 43
    Pfizer launches immersive Apple Vision Pro experience to explain cancer research
    Pharmaceutical firm Pfizer has turned to Apple Vision Pro to make an experience that's meant to help cancer patients understand how the company conducts its research into the disease, but looks more like an infomercial. The captions explaining how drug research works are interesting, but this is an immersive experience that feels more like an advert - image credit: Groove Jones/Pfizer. Among its many other medical uses, Apple Vision Pro has been used to help with cancer detection . But now pharmWilliam Gallagher <wg@williamgallagher.com>
  • 44
    Apple AirPods Pro 3 are down to $199 during Amazon's Labor Day Sale
    September Apple sales are heating up, and AirPods Pro 3 have returned to $199 at Amazon after a $50 discount. Amazon's $199 AirPods Pro 3 deal returns for September. Amazon has dropped AirPods Pro 3 back down to $199 for September, marking the lowest post-Prime Day price on Apple's flagship earbuds. Buy AirPods Pro 3 for $199 Continue Reading on AppleInsider | Discuss on our ForumsChristine McKee <christine@appleinsider.com>
  • 45
    Amazon launches first M6 Mac mini deals, grab prices from $879
    Preorder deals are in effect on the newly-announced M6 Mac mini at Amazon. Pick up the 2026 desktop for as low as $879.99. Grab preorder deals on Apple's brand-new M6 Mac mini. Save $20 on Apple's brand-new M6 Mac mini that was announced this week, bringing the price of the standard model down to $879.99 . This configuration has the new M6 chip with a 12-core CPU and 12-core GPU, along with 16GB of memory and 256GB of storage. Buy M6 Mac mini 256GB for $879.99 Continue Reading on AppleInsider |Christine McKee <christine@appleinsider.com>
  • 46
    What to expect from Apple's 'Surprise and shine' iPhone 18 Pro event on September 9
    Apple has confirmed its September special event will be held on September 9. It won't just be the iPhone 18 Pro or folding iPhone launch, but also the first event with John Ternus as CEO. Here's what to expect. The Apple logo accompanying the 2026 event invite Regular as clockwork, Apple holds an event in early September to launch new iPhone models and other products. Apple has just confirmed that this will be the case for 2026's presentation. Titled "Surprise and shine," the event will be takinMalcolm Owen <malcolm@appleinsider.com>
  • 47
    How to use NameDrop to exchange iPhone contacts in seconds
    NameDrop lets iPhone users exchange contact information by bringing two devices together. Here's how to use the feature, customize what it shares, and fix the most common problems. NameDrop causes a light and vibration to show between the two iPhones Apple introduced NameDrop in iOS 17 as part of a larger contact-sharing system built around Contact Posters. Instead of typing phone numbers manually, users can tap their iPhones together and choose whether to share or receive contact information. IAndrew Orr <andreworr@appleinsider.com>
  • 48
    Ikea Matter-over-Thread review: Amazing smart home tech, when they work
    Ikea has an impressive new lineup of Matter-over-Thread devices works with Apple Home , but after months of testing, we think that lingering connection issues with bulbs, controls, and sensors are the main problem. Ikea Matter-over-Thread lineup review: Part of Ikea's new smart home lineup with Matter-over-Thread Unlike Tradfri, Ikea's last smart home push, the fresh batch of smart home wares doesn't have a unifying name. Instead, it is made up of over 20 individual products. The launch made bigAndrew O'Hara <andrew@appleinsider.com>
  • 49
    Boox Palma 2 Pro review: Adds color at a significant cost
    The Boox Palma 2 Pro turns the idea of a pocket e-reader into a smarter, more capable daily carry without losing the minimalist charm that made the original so good. Boox Palma 2 Pro In 2023, I reviewed the Boox Palma , a pint-sized, pocket-friendly e-ink device. I praised it for its portability and ability to offer distraction-free reading pretty much anywhere you go. Now, Boox has just released the Palma 2 Pro . It has a few new tricks up its sleeve, but I wanted to see how it stacks up to itsAmber Neely <amber@appleinsider.com>
  • 50
    How to back up your Mac under macOS Tahoe
    Backing up your Mac is not just something you do when testing betas like macOS Tahoe ; it is an everyday essential if you want to avoid risking losing irreplaceable work. You may never glance at your Mac's drive, but you need to make a copy of it. Maybe you think you don't need backups because your work is like throughput in a larger workflow. You get sent a task, you work on it, then you send it on to someone else as if in an assembly line, and never, ever go back to anything you did before. BuWilliam Gallagher <wg@williamgallagher.com>
共 50 条热搜查看完整榜单

Ars Technica

实时热榜
15分钟前更新
共 20 条热搜查看完整榜单

arXiv

cs.AI
9小时前更新
  • 01
    EXAONE Forecast for Finance
    arXiv:2609.04239v1 Announce Type: new Abstract: This technical report presents EXAONE Forecast for Finance (EXAONE Finance), a financial time series (TS) foundation model (TSFM) tailored to financial forecasting. Recent TSFMs achieve strong zero-shot performance through large-scale pretraining. However, they are primarily developed for general-domain TS and largely rely on self-attention backbones whose computational cost grows quadratically with sequence length and variate count. Moreover, theySeunghan Lee, Jaehoon Lee, Jun Seo, Tae Yoon Lim, Dongwan Kang, Hwanil Choi, Minjae Kim, Sungdong Yoo, Junhyeok Kang, Sangjun Han, Soonyoung Lee, Wonbin Ahn
  • 02
    From Matching Models to Recruiting Agents: A Systematized Narrative Review of AI Recruitment Systems, Evaluation, and Governance
    arXiv:2609.04286v1 Announce Type: new Abstract: Artificial intelligence in recruitment has shifted the object being automated from profile pairs and ranked lists to multi-stage workflows that retrieve evidence, compare candidates, and support or execute actions. This systematized narrative review traces that development from bilateral retrieval and behavioral ranking through neural person--job matching, large language model (LLM) components, and tool-using recruiting agents. Using a purposive seZiyi Zhao, Guanzheng Wei
  • 03
    Harbor Adapters and Harbor-Index: Infrastructure and a Curated Meta-Dataset for Large-Scale Agentic Evaluation
    arXiv:2609.04298v1 Announce Type: new Abstract: Evaluating agents on the growing number of agentic benchmarks is challenging because they often require complex environments and agent integrations. We introduce Harbor Adapters, a unified evaluation infrastructure for agentic benchmarks. Our work makes three contributions. First, we develop benchmark adapters that port more than 80 benchmarks to evaluate arbitrary agents, and validate them through rigorous code review and parity experiments. SeconLin Shi (Audrey), Haowei Lin (Audrey), Zixuan Zhu (Audrey), Xiaoyue Zhou (Audrey), Xiang Li (Audrey), Xiangning Lin (Audrey), Yaxuan Deng (Audrey), Han Xu (Audrey), Yuangang Li (Audrey), Shanda Li (Audrey), Zizhao Chen (Audrey), Hanwen Xing (Audrey), Harsh Raj (Audrey), Bo Chen (Audrey), Quan Shi (Audrey), Steven Dillmann (Audrey), Yipeng Gao (Audrey), Puneesh Khanna (Audrey), Ruofan Lu (Audrey), Chao Beyond Zhou (Audrey), Michael Yang (Audrey), Robert Zhang (Audrey), Siyuan Chai (Audrey), Jiayu Chang (Audrey), Yizhao Chen (Audrey), Xiaokun Chen (Audrey), Yiwei Dai (Audrey), Wenting Yang (Audrey), Hange Liu (Audrey), Minghao Liu (Audrey), Zihan Wang (Audrey), Adnan El Assadi (Audrey), Benedikt Stroebl (Audrey), E. Kelly Buchanan (Audrey), Han Meng (Audrey), Junwei He (Audrey), Longxuan Yu (Audrey), Radin Shayanfar (Audrey), Yukyung Lee (Audrey), Zhikang Dong (Audrey), Allen G Hart (Audrey), Anjiang Wei (Audrey), Anurag Kashyap (Audrey), Arpandeep Khatua (Audrey), Audrey Jixin Zheng (Audrey), Chengrui Ma (Audrey), David Heineman (Audrey), Dubing Chen (Audrey), Hai-Anh Trinh (Audrey), Haishuo Fang (Audrey), Hefan Zhang (Audrey), Hui Shen (Audrey), Issa Sugiura (Audrey), Jiankai Sun (Audrey), Jiechao Gao (Audrey), Junhong Lin (Audrey), Junnan Li (Audrey), Kai Yang (Audrey), Lei Hsiung (Audrey), Maoyu Wang (Audrey), Mengze Tang (Audrey), Nabil Omi (Audrey), Negin Raoof (Audrey), Nicholas Edwards (Audrey), Octavia Guo (Audrey), Orfeas Menis Mastromichalakis (Audrey), Pengliang Ji (Audrey), Przemys{\l}aw Hejman (Audrey), Qi Qi (Audrey), Qunshu Lin (Audrey), Richard Zhuang (Audrey), Rui Yang (Audrey), Ruichen Zheng (Audrey), Ryan Marten (Audrey), Shaghayegh Fazliani (Audrey), Shizheng Hou (Audrey), Sicong Jiang (Audrey), Sijie Li (Audrey), Song Bian (Audrey), Terry Yue Zhuo (Audrey), Tianqing Wu (Audrey), Tom Tang (Audrey), Wanjia Zhao (Audrey), Weihao Xuan (Audrey), Wenhua Liang (Audrey), Xian Liu (Audrey), Xin Lan (Audrey), Xuan Zhang (Audrey), Xuandong Zhao (Audrey), Yanchuan Tang (Audrey), Yifan Jiang (Audrey), Yijiang Li (Audrey), Yitong Guan (Audrey), Yizhi Li (Audrey), Yonghui Liu (Audrey), Yuheng Tang (Audrey), Yujun (Audrey), Mao, Yunfei Zhao, Yuxin Wang, Yuxuan Tang, Zhenheng Tang, Zhifei Li, Ziruo Wang, Ziyu She, Kaiyuan Liu, Iheb Chaabane, Yuxin Tang, Xiangyi Li, Andy Konwinski, Boxuan Li, Leon Liangyu Chen, Alex Dimakis, Nicholas Carlini, Soroush Vosoughi, Di He, Etash Guha, Benjamin Feuer, Mike Merrill, Ludwig Schmidt, Alex Shaw
  • 04
    Data-Optimized Contingency Screening: A Machine Learning Approach to Power System Security
    arXiv:2609.04300v1 Announce Type: new Abstract: Ensuring the security of the power system is essential for stability and reliability, especially in the event of disruption. Effective classification of contingency in power systems enables proactive decision-making and mitigates large-scale breakdowns and failures. This study explores the use of machine learning algorithms to classify security levels of contingencies in power systems into safe, moderate or severe classes. For this approach, NewtonJoshua Salako, Folajimi Osikomaiya, Olakorede Olamiju
  • 05
    Iris: Climbing to the Search Frontier
    arXiv:2609.04304v1 Announce Type: new Abstract: We present Iris-mini and Iris-pro, two search agents trained at the 35B-A3B and 397B-A17B scales, together with the data pipeline and training recipe behind them. Tasks are reverse-constructed from the hyperlink structure of a web corpus: we author multi-hop chains over an entity graph distilled from a seed page and its out-links, rewrite every non-answer entity into a descriptive reference so that no clue can be resolved by string matching, and adZiyuan Liu, Hengqi Liu, Zichuan Wang, Yang Qin, Jiachen Liang, Xu Chu, Shaowei Chen, Yuantao Gu, Mu Chuan
  • 06
    A Removal Based Approach to Improve LLM Faithfulness at Test-Time
    arXiv:2609.04343v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used for consequential decisions, making their explanations an important tool for auditing model behavior. Unfortunately, these explanations can be unfaithful, failing to reflect the actual reasoning underlying the model's decisions. We consider a setting in which an LLM provides both an answer and an explanation in response to a question. We identify two distinct dimensions of unfaithful explanations:Qinglan Luo, S M A Nahian, John Guttag, S. Mazdak Abulnaga, Katie Matton
  • 07
    Why Better Models Can Create Riskier Systems: Evidence from LLM Agents in Financial Markets
    arXiv:2609.04373v1 Announce Type: new Abstract: Large language models (LLMs) are being deployed at scale in consequential real-world systems, from financial markets to content moderation to hiring. We show that improving individual model capability can degrade rather than improve system-level outcomes. We hypothesize that shared training and architectures can lead more capable LLMs to behave more similarly, creating correlated actions that do not diversify away. We develop a general framework shJillian Ross, Eric So, Zoe De Simone, Charles Pozniak, Andrew W. Lo
  • 08
    Corporate Language Model (CLM): Transforming Tacit and Fragmented Enterprise Knowledge into a Sovereign, Auditable, and Executable Corporate Intelligence Layer
    arXiv:2609.04377v1 Announce Type: new Abstract: Enterprise AI deployments fail not from model inadequacy, but because organizations lack a structured substrate encoding how they decide, negotiate, and execute. Generic LLMs carry no firm-specific ontological priors; RAG remains brittle, with no path to executable action; static playbooks encode logic but cannot reason or adapt. This demands an architecture treating tacit-knowledge capture, ontological grounding, sovereign deployment, and auditablFabricio C. Avini, Guilherme Trez
  • 09
    HarvestBench: Measuring Whether LLM Agents Will Pay to Avoid Killing Animals
    arXiv:2609.04444v1 Announce Type: new Abstract: Benchmarks for the side effects an agent causes on the way to a goal already exist, but HarvestBench is the first to put a price on avoiding the side effect and to name that side effect as a living creature. It is a farm simulation: LLM sub-agents drive a crew of two tractors through a cooperative corn harvest, with animals in the field. The environment is a reinforcement learning gridworld, every decision is made without memory, and the harm is neJasmine Brazilek, Miles Tidmarsh, Matthias Endres, Anshuman Singh, Jeremiah Miller
  • 10
    PerfReasoning: How Well Do LLMs Reason on Hardware Performance?
    arXiv:2609.04476v1 Announce Type: new Abstract: Performance modeling is central to hardware design and software optimization, yet constructing these models requires structured reasoning about computation, data reuse, storage, and movement. We introduce PerfReasoning, a benchmark that evaluates LLMs both as direct performance reasoners and as generators of analytical performance-model code. Given workload, architecture, and mapping specifications, models compare mappings and predict off-chip trafDan Zhao, Karthikeyan Sankaralingam, Christos Kozyrakis, Qijing Huang
  • 11
    When Quantization Breaks Memory: Recurrent-State Write-Back in Low-Precision Temporal Inference
    arXiv:2609.04490v1 Announce Type: new Abstract: Quantization is widely used to reduce the computational and memory demands of neural-network inference. In recurrent networks, however, the quantized state is stored and returned at the next time step, so the rule used to store that state can alter subsequent computations. Here, we introduce recurrent-state write-back to denote this rule and isolate its effect in a compact GRU encoder--decoder for fluorescence lifetime imaging, a molecular imagingIsmail Erbas, Xavier Intes, Vikas Pandey
  • 12
    ResLearn-XR: Residual Learning for Network Traffic and Quality-of-Experience-Aware Modeling in Extended Reality
    arXiv:2609.04493v1 Announce Type: new Abstract: We present ResLearn-XR, a residual learning framework for predicting eXtended Reality (XR) network traffic and estimating Quality-of-Experience (QoE) risk. ResLearn-XR adopts a two-stage temporal learning structure comprising a base sequence prediction model augmented with task-specific residual learning components to improve adaptability to bursty, non-stationary XR traffic dynamics. The residual learning stages operate in the value space for contYoga Suhas Kuruba Manjunath, Jie Gao, Lian Zhao
  • 13
    Rethinking Indirect Prompt Injection as a Test-Time Search Problem
    arXiv:2609.04495v1 Announce Type: new Abstract: We formulate indirect prompt injection as a test-time search over a task-dependent attack surface induced by the environment, user task, and injection task. To operationalize this formulation, we introduce an agentic attacker with a dedicated search harness that performs environment reconnaissance, structured reasoning over attack strategies, and adaptive evaluation using victim-agent feedback. Across heterogeneous tasks, we find that increasing atDuong M. Nguyen, Joon Sik Kim, Blazej Manczak, Vaikkunth Mugunthan
  • 14
    BioSync: Transformer-Based Cross-Modal Fusion for a Multimodal Physiological Digital Biomarker
    arXiv:2609.04504v1 Announce Type: new Abstract: Cardiac, neural, behavioral, and speech measurements from wearable and mobile devices provide partial, noise-sensitive views of physiological state. BioSync combines these measurements into the \textbf{BioSync Index (BSI)}, a continuous composite digital biomarker defined under the BEST framework. The model applies multi-head self-attention to modality tokens and adds a linear branch whose hypothesis class includes standard feature concatenation. TSeyed Mahmoud Sajjadi Mohammadabadi
  • 15
    What Does Multi-Harness RL Learn? Credit Assignment and Portability in Coding Agents
    arXiv:2609.04518v1 Announce Type: new Abstract: Agent reinforcement learning (RL) increasingly runs through full execution harnesses, and a multi-harness recipe mixes two choices: exposing the policy to several harnesses, and comparing their rewards inside one relative-advantage group. We isolate the second choice in repository-level coding. From one Qwen3-8B supervised warm start we replay the same frozen task-harness records from Aider, OpenHands, Qwen Code, and SWE-agent, with the same numberChenqian Le, Jiayi Cheng, Qijia He, Runhao Li, Yinghao Li, Xupeng Chen
  • 16
    MaxKernel: Agentic Kernel Generation for TPUs
    arXiv:2609.04523v1 Announce Type: new Abstract: Designing and authoring high-performance custom kernels for accelerators is a complex task that requires deep hardware-level expertise. Large Language Models (LLM) can be leveraged together with real-time compiler feedback to build agentic systems for kernel generation. In this work, we present MaxKernel, a multi-agent system that implements three distinct paradigms for TPU kernel development: (1) a Human-in-the-Loop (HITL) agent for collaborative,Shangkun Wang, Nina Cai, Charles Hoong, Julian Walker, Gerson Kroiz, George Vanica, Deepak Patil, Andi Gavrilescu, Hassan Sipra, Sethu Sankaran
  • 17
    Towards a universal language of concepts: A survey
    arXiv:2609.04528v1 Announce Type: new Abstract: Humans can learn and generalize novel concepts from sparse data because they express knowledge in rich structural formats. In this paper, we propose that programs are a strong candidate for universal representation of concepts. We review computational models of concept learning that use programs as their concept representation and evaluate their contribution toward a universal representational language.Aishni Parab
  • 18
    Data-Driven Discovery of Composition-Dependent Constitutive Models for Hyperelasticity and Viscoelasticity of Digital Materials
    arXiv:2609.04541v1 Announce Type: new Abstract: Digital materials fabricated by multi-material 3D printing are designed as controlled mixtures of stiff and compliant constituents, yielding effective responses that span more than an order of magnitude in apparent stiffness and exhibit strongly nonlinear, composition-dependent, and rate-dependent dissipative behavior. Classical finite-strain viscoelastic models represent such behavior with closed-form strain energy functions for equilibrium and noJosu\'e Garc\'ia-\'Avila (Department of Mechanical Engineering, Columbia University, New York City, USA), Beijun Shen (Department of Mechanical Engineering, Columbia University, New York City, USA), Manuel K. Rausch (Department of Aerospace Engineering and Engineering Mechanics, University of Texas at Austin, Austin, USA, Department of Biomedical Engineering, University of Texas at Austin, Austin, USA, Department of Mechanical Engineering, University of Texas at Austin, Austin, USA), Mary C. Boyce (Department of Mechanical Engineering, Columbia University, New York City, USA), Adri\'an Buganza-Tepole (Department of Mechanical Engineering, Columbia University, New York City, USA)
  • 19
    From Answers to Interpretations: Rethinking Ambiguity-Induced Aleatoric Uncertainty Estimation in LLMs
    arXiv:2609.04543v1 Announce Type: new Abstract: A key challenge in reliable LLM deployment is recognizing when uncertainty reflects irreducible variability in the task rather than limitations in the model's knowledge. In language tasks, a central source of such aleatoric uncertainty is input ambiguity or underspecification, where multiple interpretations remain plausible. Existing decomposition methods estimate aleatoric uncertainty by generating multiple clarifications of the input, querying thOmer Nahum, Niv Nayman, Jonathan Fhima, Alon Zolfi, Jeremy Levy, Shai Mazor, Paolo Favaro
  • 20
    IPGeoAI: Transformer-Based Geolocation with LLM Semantic Fusion
    arXiv:2609.04559v1 Announce Type: new Abstract: Accurate city-level IP Geolocation is an important enabler for the modern digital ecosystem, underpinning services ranging from local content delivery and targeting to digital rights enforcement. However, traditional heuristic and database-driven methods often struggle to resolve the complex, non-linear allocation patterns of modern network infrastructures, particularly within the exploding IPv6 address space and transient mobile networks. In thisAvinash Kadimisetty, Andy Jinqing Yu, Philip Favaloro, Wenlong Liu, Xiaolu Xiong
  • 21
    Reducing Hallucinated Transcripts in Whisper via Hallucination Space Projection
    arXiv:2609.04561v1 Announce Type: new Abstract: Whisper is a widely used foundation model for automatic speech recognition (ASR), but its generative decoder can produce fluent hallucinated transcripts for inputs containing little or no speech. We propose a training-free, inference-time method to reduce these hallucinations using low-rank projection of decoder activations. A compact hallucination-associated subspace is estimated from non-speech calibration data, and decoder hidden states are projMaryam Abbasihafshejani, Murtuza Jadliwala
  • 22
    La Agente \'Optima: Towards Agentic Self-Driving Laboratories
    arXiv:2609.04564v1 Announce Type: new Abstract: Self-driving laboratories (SDLs) combine automated experimentation with adaptive decision-making to accelerate scientific discovery. Their operation nevertheless often depends on human specialists who translate scientific objectives into executable closed-loop campaigns. Specialists adjust them as data and operating conditions change. Here, we present La Agente \'Optima, an agentic framework that constructs and supervises Bayesian optimization campMarcel M\"uller, Jiaru Bai, Willi Gottstein, Abhijoy Mandal, Mohammad Nazeri, Elia Savino, Yanlin Fang, Sujoy Das, Sergio Pablo Garc\'ia Carrillo, Yeonghun Kang, Juan B. P\'erez-S\'anchez, Simone Pilon, Martin Fitzner, Timothy No\"el, Frank Gu, Varinia Bernales, Al\'an Aspuru-Guzik
  • 23
    Extremely Sparse Supervision Incentivizes Reasoning Ability
    arXiv:2609.04565v1 Announce Type: new Abstract: Large language models demonstrate increasingly strong reasoning capabilities through effective post-training. Yet, prevailing post-training methods optimize over massive numbers of tokens, implicitly assuming that effective learning must be token-intensive. We revisit this assumption in the on-policy distillation (OPD) setting, which naturally admits dense teacher supervision at every generated token. Using the Qwen3 family, we discover a counter-iZhishuai Liu, Xingzi Xu, Mehmet Saygin Seyfioglu, Pan Xu, Karim Bouyarmane
  • 24
    Does the Selected Object Reach the Reader? Auditing Identity Handoffs in Grounded Language-Model Pipelines
    arXiv:2609.04579v1 Announce Type: new Abstract: Grounded language-model pipelines can be divided into three stages: selecting an object, retrieving passages for it, and using that evidence to answer. If the selected object must reach the reader, losing it breaks the handoff. Benchmark recall checks the dataset-linked object, which can differ. We audit 600 HybridQA questions across three selector families. On 1,463 resolvable records where the selected object matches the dataset-traced passage, eSiddharth Vohra, Runmin Jiang, Xiaomo Li, Min Xu
  • 25
    $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
    arXiv:2609.04611v1 Announce Type: new Abstract: LLM agents are rapidly becoming production software, deployed to handle customer service, adjudicate disputes, and operate internal systems. Notably, the work of building them is increasingly handed to coding agents, yet existing benchmarks say little about whether an AI system can deliver one under the conditions of a real client engagement. We introduce $\tau^\tau$-bench (pronounced hyper-tau-bench), a benchmark that makes agent construction theQuan Shi, Keshav Dhandhania, Karthik Narasimhan, Victor Barres
  • 26
    Leveraging Imperfect Restoration for Data Availability Attack
    arXiv:2609.04627v1 Announce Type: new Abstract: The abundance of online data is at risk of unauthorized usage in training deep learning models. To counter this, various Data Availability Attacks (DAAs) have been devised to make data unlearnable for such models by subtly perturbing the training data. However, existing attacks often excel against either Supervised Learning (SL) or Self-Supervised Learning (SSL) scenarios. Among these, a model-free approach that generates a Convolution-based UnlearYi Huang, Jeremy Styborski, Mingzhi Lyu, Fan Wang, Adams Kong
  • 27
    SiLR: Structure-Preserving Admission and Process Reward for LLM Tool Agents
    arXiv:2609.04629v1 Announce Type: new Abstract: A runtime gate for an LLM tool agent is usually cast as a filter. In a ReAct loop a rejected proposal is followed by another at the same state, so the gate is a search operator over the proposal stream whose admission criterion shapes which trajectories are reachable. We study post-violation recovery admission, where progress must be admitted while the system is still in violation, and identify the scalar projection trap: an aggregate-score gate acChenyu Zhou, Qiliang Jiang, Shuning Wu, Xu Zhou
  • 28
    A Cost-Aware Agentic Architecture for NL-to-SQL over Nested Enterprise Schemas, with a New Benchmark
    arXiv:2609.04641v1 Announce Type: new Abstract: Natural-language-to-SQL systems have ad- vanced rapidly on academic benchmarks, yet production enterprise schemas exhibit graph- like, semi-structured, deeply nested structure that current benchmarks do not measure. We make two complementary contributions. First, we introduce the DevRev NL2SQL bench- mark: 900 execution-verified queries with nested-type and link-graph structure, accom- panied by the Semantic Depth Score (SDS), a schema-agnostic rubYoga Sri Varshan Varadharajan, Ajay Yadav, Ritesh Goru, Prateek Chaudhury, Constantine Caramanis, Prateek Jain, Divyateja Pasupuleti, Sunil Kumar Pandey
  • 29
    Continual Graph Memory for Adaptive Recommendation under Intent Drift
    arXiv:2609.04651v1 Announce Type: new Abstract: This paper studies adaptive recommendation under intent drift, where feedback from each recommendation outcome can reveal whether the relational evidence used for ranking is useful, missing, or misleading. While Knowledge Graphs (KGs) provide essential semantic structure to handle these shifts, traditional KG-enhanced systems treat the graph as a static retrieval substrate, making it brittle to evolving intents, noisy metadata, and recurring failurHao Nguyen Ngoc, Tung Nguyen, Nguyen Thi Hanh, Hoang Thai Dinh, Nguyen Xuan Tung
  • 30
    Harness-agnostic detection and immunization of reward hacking in self-evolving language models
    arXiv:2609.04665v1 Announce Type: new Abstract: Self-evolving language models improve by proposing candidate updates and keeping whatever raises a visible score. When that score is an imperfect proxy for the capability one actually wants, sustained selection widens the gap between the two. This is reward hacking. We introduce HackProbe, a monitor that attaches to an arbitrary self-evolving loop through two black-box hooks, with no access to weights or activations. It keeps a secret, distributionRongxin Yang, Yang Liu, Shang Luo, Haoxuan Jia, Chongyang Zhang, Hao Zheng, Yingguang Yang, Yulin Huang, Jianshen Zhang, Yongzhi Qi, Kefu Xu, Congjing Ran, Bin Chong
  • 31
    ERPBench: Evaluating LLM Agents for Enterprise Decision-Making Across Competitive Market Ecologies
    arXiv:2609.04667v1 Announce Type: new Abstract: Large language model (LLM) agents are increasingly proposed for enterprise workflows, yet existing evaluations rarely test whether business-decision conclusions transfer across competitive market ecologies. We introduce ERPBench, an execution-instrumented benchmark for enterprise decision agents in a six-round Enterprise Resource Planning (ERP) simulation with coupled pricing, production, procurement, inventory, finance, and shared-market competitiXinran Zhang, Pengrui Lu, Lyumanshan Ye, Pengfei Liu
  • 32
    Train What You Deploy:Token-Faithful Post-Training of a Production Coding
    arXiv:2609.04678v1 Announce Type: new Abstract: Existing post-training pipelines for coding and terminal agents suffer severe token and control fidelity errors: simplified training environments mismatch production deployments, and offline token reconstruction from agent logs distorts original prompts and conflates policy calls with background model operations. We present a fidelity-aware training coupling framework that retains trainer-side sampling over original prompts, eliminates spurious modCheng Li, Jiexiong Liu, Yixuan Chen, Chi Hong
  • 33
    Predicting Spatiotemporal Mobile Sensing-Based PM2.5 Concentrations Using Low-Rank Adapted Spatially Attentive Graph Neural Network
    arXiv:2609.04693v1 Announce Type: new Abstract: Urban air quality can vary significantly along transit corridors, necessitating high-resolution monitoring. This work introduces a novel mobile-sensing dataset from Surat, Gujarat, India, comprising PM$*{2.5}$ concentrations, meteorological variables (temperature, humidity, wind speed, wind direction), and land-use features. To represent the spatiotemporal data as a graph, two node-definition strategies were used: (i) uniform segmentation (200--400Om Chiddarwar, Priyanka Mandal, Praveen Kumar Chandaliya, Shriniwas Arkatkar
  • 34
    SQL-Zero: Self-Evolving Text-to-SQL
    arXiv:2609.04697v1 Announce Type: new Abstract: Training a competitive Text-to-SQL agent usually depends on human-annotated natural-language/SQL pairs, which are expensive, domain-specific, and a bottleneck for scaling to new databases. We show it is possible to train a competitive solver with zero annotated pairs. We introduce SQL-Zero, a proposer-solver self-play in which a challenger and a solver start from the same base LLM and the only ground truth is execution against the database itself.Daniel Machado Pedrozo, Julia Soares Dollis, Bryan Lincoln Marques de Oliveira, Vinicius Alboneti Aguiar, S\'avio Salvarino Teles de Oliveira, Telma Woerle de Lima Soares
  • 35
    Model Retirement Creates Reproducibility Risk in Biomedical AI Publications
    arXiv:2609.04699v1 Announce Type: new Abstract: Background. Large language models (LLMs) are being adopted in biomedical research at a rapid and accelerating pace, yet commercial services that host many widely used models operate under deprecation schedules that can complicate scientific reproducibility. Methods. We searched PubMed for original research articles from 2022 through March 2026 that applied a specific LLM to a biomedical task. An extraction agent identified model names from 61,077 aNathan Wolfrath, Meghan Conroy, Thomas Kosten, Dave Bell, Bhabishya Neupane, Jonah Kindel, Anjishnu Banerjee, Priya Deshpande, Bradley Taylor, Anai N. Kothari
  • 36
    FinalityBench: An Effect-Level Benchmark for Agent Decisions Under Delayed and Conflicting Financial Finality
    arXiv:2609.04706v1 Announce Type: new Abstract: A merchant's payment processor, ledger, ERP and bank feed are updated by messages that get delayed, duplicated, dropped and reordered, so for minutes at a time the four hold contradictory beliefs about the same order. An agent resolving the exception must decide whether to ship goods, re-submit a capture, refund or wait, knowing some of those cannot be undone. We present FinalityBench, an executable benchmark for that decision. It keeps a hidden caAbhishek Sharma
  • 37
    PLUME: Parameter-Efficient Personalization of Large Language Models via Low-Rank User Modulation in Shared Subspaces
    arXiv:2609.04715v1 Announce Type: new Abstract: Personalizing large language models (LLMs) is essential for delivering AI assistance that aligns with individual users' styles, intents, and preferences. While per-user fine-tuning can substantially enhance personalization quality, it introduces significant parameter and storage overhead, limiting scalability to large user populations. We propose PLUME (Personalized Low-Rank Adaptation through User Modulation and Shared Subspace), a lightweight fraXinyu Li, Hao Zhou, Jianfeng Zhu, Julina Maharjan, Ruixin Guo, Feodor Dragan, Ruoming Jin
  • 38
    Aplaud: Adaptive Personalized Low-Rank Decomposition for User-Specific LLM
    arXiv:2609.04738v1 Announce Type: new Abstract: In this paper, we study the problem of personalized survey response prediction using fine-tuned large language models (LLMs). This task poses unique challenges: limited per-user training data, scalability of model storage, and the need to exploit shared structure across survey questions. To address these issues, we propose Aplaud (Adaptive Personalized Low-rank and User-specific Nested Decomposition), a lightweight and scalable framework for LLM peXinyu Li, Ruoming Jin, Jianfeng Zhu, Ruixin Guo, Zhi Liu
  • 39
    DCFA: Dual-view Causal-inspired Attribution for Failure Reasoning in LLM-based Multi-agent Systems
    arXiv:2609.04749v1 Announce Type: new Abstract: Large language model (LLM)-based multi-agent systems have experienced rapid growth in recent years. Despite their promise, such systems remain fragile, frequently exhibiting reasoning and coordination errors that can lead to system-level failures. Failure attribution in such systems relies on tracing natural language interactions among agents to identify the decisive error, which refers to the earliest action whose correction can reverse system faiZehao Wang, Lanjun Wang, Shilong Jin, Junjie Chen, Yanghua Xiao
  • 40
    Shadow Queries for Private Retrieval in Vector Databases
    arXiv:2609.04767v1 Announce Type: new Abstract: Large language models (LLMs) increasingly rely on information retrieval (IR) systems, such as Retrieval-Augmented Generation (RAG), to incorporate domain-specific knowledge without costly re-training. These systems often store pre-computed document embeddings in cloud-based vector databases. However, such embeddings are vulnerable to embedding inversion attacks (EIAs), which can reconstruct their underlying text. Existing defenses, such as adding nXinguo Feng, Zhongkui Ma, Zihan Wang, Chuan Yan, Guowei Yang, Alsharif Abuadbba, Guangdong Bai
  • 41
    Diffusion Language Models for Mobile Edge Agentic AI: Foundations, Applications, and Challenges
    arXiv:2609.04778v1 Announce Type: new Abstract: Diffusion language models (DLMs) offer a non-autoregressive alternative for mobile edge agentic artificial intelligence (AI) by refining tokens through iterative denoising rather than left-to-right decoding. Compared with autoregressive Transformer-based large language models (LLMs), DLMs can update multiple uncertain tokens in parallel and exploit bidirectional context throughout the generation process, enabling more flexible quality-latency tradeChenqi Li, Minghui Min, Dusit Niyato, Wei Ni
  • 42
    DODR: Deterministic Operator-Driven Reasoning in Latent Space
    arXiv:2609.04782v1 Announce Type: new Abstract: Autoregressive (AR) large language models formulate reasoning as token-level probabilistic sampling, which induces three fundamental defects in complex logical reasoning: error accumulation, probability substituting necessity, and the linear-chain information bottleneck. This paper proposes the Deterministic Operator-Driven Reasoning in Latent Space architecture (DODR), which reconstructs reasoning as reasoning-graph computation in a high-dimensionWeicai Huang (Beijing MQPat Technologies, Co., Ltd.)
  • 43
    ProtLingo: Efficient Protein Language Modeling via Conditional Memory and Expert Routing
    arXiv:2609.04793v1 Announce Type: new Abstract: Proteins perform diverse cellular functions, and even single amino-acid substitutions can alter stability, activity, or molecular interactions. Protein language models (PLMs) provide a scalable approach for modeling such sequence--function relationships from unlabeled sequences, but increasing the size of dense Transformer backbones often brings substantial computational cost without consistently improving mutation-sensitive prediction. We introducMingrui Li, Sixian Shen, Minzhang Li, Ruiyi Zhang, Kexin Zhang, Jiakai Zhang, Jingyi Yu
  • 44
    Whose record is this? Diagnosing and authorizing record use in personalized multimodal models
    arXiv:2609.04801v1 Announce Type: new Abstract: Contextualized visual personalization can retrieve a true record yet apply it to the wrong visual subject. We formalize when a record may condition an answer as \emph{record authorization}: subject presence ($P$), record-edge validity ($E$), and answer support ($S$) must all hold. We call violations visual memory misbinding (VMM). We construct RecordAuth-Diag, a 3,690-case matched diagnostic suite that changes one image--record edge while holding tXinyu Mao, Junsi Li, Chenyang Liu, Haoji Zhang, Ming Sun
  • 45
    Hierarchical Possession-Aware Graph Pointer Network for Pass Receiver Selection
    arXiv:2609.04803v1 Announce Type: new Abstract: Pass receiver selection is a fundamental task in football analytics, aiming to predict the intended receiver under a given game state. This task is challenging with event-centered freeze-frame observations, a broadcast-like setting that provides only partial and variable player visibility without complete trajectories or stable player identities. The model must therefore reason over anonymous visible candidates, opponent pressure, and recent contexJingyi Wang, Da Li, Kaixin Wang, Zhangqin Huang
  • 46
    MedFlow: Class-Aware Multi-Scale Generation for Medical Time-Series Synthesis
    arXiv:2609.04804v1 Announce Type: new Abstract: Synthetic medical time-series generation can alleviate data scarcity and support the development of reliable clinical prediction models. However, existing methods mainly focus on matching the overall distribution and temporal dynamics of real data, which does not necessarily ensure strong downstream utility on imbalanced medical datasets. Clinically informative patterns often occur at heterogeneous temporal scales, while rare minority-class charactYanhao Huang, Shibo Feng, Wanjin Feng, Peilin Zhao, Chunyan Miao
  • 47
    When Financial Fine-tuning Fails: A Three-Level Detectability Analysis of Numerical Hallucination in Domain-Adapted Language Models
    arXiv:2609.04806v1 Announce Type: new Abstract: Financial large language models are increasingly deployed for summarization of reports and disclosures, where numerical hallucination poses significant practical risks. While prior work often attributes such hallucination to insufficient numerical reasoning, this assumption has not been systematically tested under controlled fine-tuning settings. In this paper, we conduct a cost-effective, controlled study of numerical hallucination in financial suXiaodong Li, Peiwei Liu
  • 48
    CPR-IE:A Compression-Prediction-Resource Intelligence Efficiency Metric
    arXiv:2609.04809v1 Announce Type: new Abstract: Comparing intelligent systems under deployment constraints requires more than predictiveaccuracy.This paper develops Compression-Prediction-Resource Intelligence Efficiency (CPR-IE) as a protocol-relative ordering by representational economy, predictive quality, and resourceburden. The analysis separates two questions-how raw resource consumption is represented, andhow the resulting attributes are aggregated. Proportional-increment composition uniqXiantao Jiang
  • 49
    Long Horizon Transformer Quantile Fault Prediction for Multi Site Industrial Predictive Maintenance
    arXiv:2609.04840v1 Announce Type: new Abstract: Long-horizon predictive maintenance requires models to distinguish slowly evolving degradation from normal operating-regime variation over planning windows measured in days rather than hours. This paper evaluates whether an explicit conditional-quantile representation provides an informative classifier interface for this problem. The proposed TQRNN30d framework combines a dual-stage quantile regression neural network (QRNN) feature extractor with aDavid J Poland, Daniele Ravi, Na Helian
  • 50
    ElderBench: Benchmarking Autonomous Mobile Agents for Older Adults
    arXiv:2609.04850v1 Announce Type: new Abstract: While autonomous mobile agents hold great potential for assisting older adults with smartphone usage, existing GUI benchmarks mainly rely on explicit, goal-oriented instructions and rarely capture the naturally occurring language patterns of older users, such as indirect speech, referential ambiguity, and under-specified requests. This mismatch between benchmark instructions and real-world elderly interactions may hinder reliable agent deployment.Weide Zhan, Qumu Shaqu, Yuanqing Liu, Peng Zhang, Jiahao Liu, Kam Him Lam, Ning Gu, Zhan Hu, Tun Lu
  • 51
    MM-IFEval-Pro: A Multilingual and Attack-Resistant Benchmark for Instruction-Following in Vision-Language Models
    arXiv:2609.04859v1 Announce Type: new Abstract: As vision-language models (VLMs) rapidly advance in image understanding, cross-modal reasoning, and complex instruction execution, instruction-following capability has become a key indicator of their reliability and practicality. However, existing multimodal instruction-following benchmarks still suffer from limited language coverage and insufficient adversarial safety scenarios, making them inadequate for evaluating real-world multilingual and safChangming Xiao, Zhenliang Ni, Jinhui He, Han Shu, Jie Hu
共 298 条热搜查看完整榜单

Axios

实时热榜
14分钟前更新
  • 01
    The body image battle is real in 2026, but the bodies may not be
    A societal spotlight on extreme thinness is stripping body diversity from screens, just as fashion magazines did in the early aughts, mental health experts tell Axios. The big picture: Fake content creators are introducing unattainable, AI-generated body images that prey on human nature and exacerbate the struggle for many to reach self-acceptance for their bodies. When shrunken celebrities "become the ideal ... because of their status, then the virtual influencers are going to be emulating thatAvery Lotz
  • 02
    Sports' grip on America sparks summer of political warfare
    The WNBA, Major League Baseball and even a golf startup spent the summer as unlikely proxies in some of the country's fiercest national conflicts. Why it matters: American culture has shattered into thousands of tiny shards of glass . But sports remains one of the few forces powerful enough to pull much of the country into the same moment — making it a prime target for political fights. That kind of shared attention has become a scarce and ultra-valuable resource for activists, politicians and cNoah Bressner
  • 03
    2026 candidates target health care "corruption"
    Campaigning on health care used to mean talking about coverage or costs . Now, in the run-up to the midterms, it's just as likely to focus on "corruption." The big picture: Candidates are accusing primary or general election opponents of being in the pockets of drugmakers, insurers and other powerful health interests, to the detriment of voters. It's a strategy that plays into affordability concerns and voter anger at the health system — but frequently avoids addressing what about the system neeAdriel Bettelheim
  • 04
    Exclusive poll: GOP's gains among Latinos fade in House battlegrounds
    Data: Hart Research/TelevisaUnivision; Table: Russell Contreras/Axios Democrats have gained sharply with Latino voters in some of the country's most competitive House districts since 2024, a new Hart Research/TelevisaUnivision poll shared exclusively with Axios shows. Why it matters: President Trump came close to erasing Democrats' longtime advantage among Latinos in the last presidential election. Now, with control of Congress at stake and Trump's approval ratings hitting new lows , that breaktRussell Contreras
  • 05
    "Voters don't trust Jeffries": Democratic leader faces impending fight with the left
    House Minority Leader Hakeem Jeffries (D-N.Y.) faces mounting behind-the-scenes frustration from progressive lawmakers and candidates that's likely to burst into public view after the midterm elections. Why it matters: Unless Democrats win a big enough House majority in November to govern without the progressive wing, a Speaker Jeffries would have to contend with the left flank's demands. An influx of Democratic Socialists of America-aligned members in January will compound latent distrust of JeAndrew Solender
  • 06
    Zelensky to Axios: U.S. exploring Russia-Ukraine winter de-escalation steps
    The U.S. is exploring whether Ukraine and Russia could take steps to de-escalate the war during the winter while restarting negotiations toward a broader peace agreement, Ukrainian President Volodymyr Zelensky told Axios in an interview Monday. Why it matters: The Ukrainian president believes the weeks before winter offer a window for diplomacy . While Zelensky said he's convinced Russian President Vladimir Putin is counting on another brutal winter to weaken Ukraine, he added that Putin doesn'tBarak Ravid
  • 07
    The grocery aisle's new double whammy: price and safety
    Plenty of Americans feel like they're forking over a fortune for groceries —only to get hit with a recall notice that forces them to throw that expensive food out. Why it matters: While grocery prices are driven by complex trade and energy costs, the safety net meant to keep them safe to eat is fraying, creating a dual crisis of affordability and safety. The average American household pays more than $10,000 on groceries every year, and two-thirds of Americans think they're spending too much. PriJosephine Walker
  • 08
    Iran war drives $100 billion in extra energy costs for U.S. consumers
    The war in Iran has now cost U.S. consumers $100 billion in higher energy prices, and the bill is rising another $1 million about every two minutes, per a real-time estimate from Brown University as of Monday morning. Why it matters : Energy inflation shows up across the entire economy, and the recent surge in diesel prices in particular threatens to have a dramatic impact on freight and travel in the weeks and months to come. By the numbers : The Iran War Energy Cost Tracker from Brown's WatsonBen Berkowitz
  • 09
    Axios interview: Altman sees revenge of "idea guy"
    OpenAI CEO Sam Altman says this era is bringing the revenge of the "idea guy" — the once-annoying aspiring entrepreneur burbling about their brainstorm with no technical chops to pull it off. Altman says it would be like saying: "I have a great idea for a song. I just need that guy with the guitar to make it for me." "It really did strike me as ridiculous," Altman told me in Chapel Hill, N.C. "But now we're watching a whole generation of people iterate on ideas faster than they ever could. The eMike Allen
  • 10
    The math behind a Democratic wave — or a GOP upset
    Data: The Downballot and Sabato's Crystal Ball at the University of Virginia Center for Politics. (Balance of power accounts for opposing party gains due to redistricting.) Map: Kavya Beheraj/Axios Redistricting has helped House Republicans build a seawall that can withstand a roughly 5-point Democratic wave. Why it matters: Early readings are pointing to something twice that size — enough to wash away the GOP House majority. Democrats have been running roughly 10 points ahead of their 2024 presHans Nichols
  • 11
    A growing movement wants AI out of schools
    America's two largest school districts both imposed sweeping AI bans this past week, hitting the brakes on a technology that's rapidly making its way into classrooms around the country. Why it matters: Research on AI's educational value is mixed. Some studies have found personalized AI tutors can improve learning, while other research has linked AI-assisted essay writing to weaker critical thinking. Driving the news: New York City's public school district blocked AI use for roughly 600,000 studeJosephine Walker
  • 12
    AfD just won 44% in a German state. Now Germany's firewall faces its biggest test.
    Germany's far-right Alternative for Germany (AfD) won a record 44% in Saxony-Anhalt on Sunday, leaving a small party with seats that could determine whether it can lead the nation's first far-right state government since World War II. Why it matters: Germany rebuilt its democracy after Nazi rule with unusually strong safeguards against extremism. AfD's breakthrough shows those barriers aren't stopping Europe's far right from becoming dominant electoral forces. The latest: AfD more than doubled iAndrew Pantazi
  • 13
    Exclusive: Latino leaders push Beshear on Texas' race for governor
    Dozens of Latino lawmakers and leaders are pressuring Kentucky Gov. Andy Beshear , who chairs the Democratic Governors Association, to invest $10 million in the Texas governor's race, according to a letter to Beshear that Axios obtained. Texas state Rep. Gina Hinojosa , who polls suggest is running a strong challenge to Gov. Greg Abbott (R), is trying to become Texas' first Latino governor. Beshear must decide soon how to invest the DGA's cash in the 36 states with governor's races in November.Alex Thompson
  • 14
    Dem governors weighing White House face the DeSantis dilemma
    Top aides for Florida Gov. Ron DeSantis ' failed 2024 presidential campaign have a warning for Democratic governors eyeing a 2028 run: Don't wait too long after November's midterms to jump into the race. Why it matters: In the coming months, Govs. Josh Shapiro of Pennsylvania, Wes Moore of Maryland, JB Pritzker of Illinois and Andy Beshear of Kentucky will face a dilemma similar to what confronted DeSantis after his 2022 reelection: Do they get in the race early — before their postelection legisAlex Thompson
  • 15
    Witkoff and Kushner, in a first, visit Ukraine for peace talks
    President Trump's envoys Steve Witkoff and Jared Kushner arrived in Kyiv on Sunday for diplomatic talks aimed at resuming negotiations between Russia and Ukraine . Why it matters: Kushner and Witkoff traveled to Kyiv right after talks with Russian President Vladimir Putin in Moscow. This is Witkoff and Kushner's first trip to Ukraine despite months of negotiating with both sides and multiple trips to Russia. Sunday also marked the highest-level visit by a U.S. official to Ukraine since Trump staBarak Ravid
  • 16
    Trump's energy boss wavers on gasoline price path
    Energy Secretary Chris Wright on Sunday cautiously predicted gasoline price relief is on the way, but stopped short of firm or precise pledges. Why it matters: Pump prices are at their highest Labor Day weekend levels on record, and the midterms loom less than two months away. His appearances on Sunday political shows also comes as diesel prices, which ripple through the economy, are at all-time highs . "I think prices headed downward is — is not an unreasonable expectation," he said on CBS's "FBen Geman
  • 17
    Witkoff and Kushner meet with Putin about ending Ukraine war
    President Trump envoys Steve Witkoff and Jared Kushner met with Russian President Vladimir Putin at the Kremlin Saturday and discussed U.S. ideas for ending the war in Ukraine. Why it matters: Witkoff and Kushner want to resume direct talks between Russia and Ukraine after negotiations were suspended six months ago and the fighting has escalated ahead of winter. Witkoff and Kushner, who are expected to travel to Kyiv on Sunday, got Russia and Ukraine to agree to a short pause in airstrikes to alBarak Ravid
  • 18
    Scoop: Republican midterms convention speakers revealed
    President Trump and his top allies will share the stage with Republican candidates in some of the nation's most competitive races at next week's GOP midterm convention in Dallas, according to a preliminary list of speakers obtained by Axios. Why it matters: The lineup underscores a central purpose of the convention: firing up Trump voters and getting them to turn out in November, even though the president isn't on the ballot. Driving the news: Trump will speak both nights of the Sept. 9-10 conveAlex Isenstadt
  • 19
    U.S. attacks 3 Iranian tankers in tit-for-tat response
    U.S. forces struck three Iranian oil tankers Saturday after Iran fired ballistic missiles at two U.S. Navy warships, U.S. Central Command said. Why it matters: The strikes continue a tit-for-tat war that just passed the six-month mark . A U.S. official told Axios this week it had a new "tanker-for-tanker" policy approved by President Trump. Driving the news: CENTCOM said it retaliated after a U.S. aircraft carrier and guided-missile destroyer evaded missiles fired by Iran. The U.S. attacks "permAndrew Pantazi
  • 20
    "We're stuck:" GOP accepts Trump's all-about-me midterms
    President Trump is hitting the campaign trail and doing the last thing many Republicans want: making the 59 days to the Nov. 3 midterms a referendum on him. At the same time, they know it's inevitable. Why it matters : Most Republicans are too scared to bluntly tell Trump how unpopular he is. And Trump wants an unprecedented convention next week, ads and events to focus on him. The big picture : Their timidity aside, Republicans need Trump's low-propensity voters to turn out. So they're going alMarc Caputo
  • 21
    Republicans sour on the economy
    Republicans' confidence in the economy has plummeted over the last six months at a pace unseen since Covid — reminiscent of the declines during the global financial crisis nearly two decades ago. Why it matters: Usually, the president's party is upbeat about the economy, while the party out of power is far more downcast. What's striking now is that Republicans have grown increasingly sour this year, even with a Republican president in office. It's another in a long list of ominous indicators forMike Zapler
  • 22
    Trump envoys Witkoff and Kushner to travel to Moscow and Kyiv this weekend
    President Trump's envoys Steve Witkoff and Jared Kushner will travel this weekend to Moscow and Kyiv for talks on ending the war, U.S. officials said. Why it matters: This will be the first visit by Witkoff and Kushner to Kyiv since President Trump returned to office. "They're bringing with them a plan to end the war," Trump told reporters in the Oval Office on Friday. State of play: Witkoff and Kushner will travel on Friday, hold talks with Russian President Vladimir Putin at the Kremlin on SatBarak Ravid
  • 23
    A costly summer sendoff approaches
    Persistent pump pain and pricey burgers mean this summer's last hurrah comes with a cost crunch. The big picture: Gasoline prices are almost certain to remain north of $4 per gallon over Labor Day weekend — a heavy travel stretch and the time when midterm election season intensifies. Driving the news: Labor Day gas prices are slated to be the highest on record, surpassing the previous one set in 2012, per market tracking firm GasBuddy and AAA . That's coming off a record-setting August, where thAvery Lotz
  • 24
    Scoop: Trump administration drafting post-war Middle East plan
    The Trump administration is working on a post-war strategy focused on a regional effort to contain Iran and expand normalization of relations between Israel and its neighbors, according to two U.S. officials and two other sources with knowledge of the issue. Why it matters: While the plan is still in the early stages of drafting, it's intended to guide the U.S. approach in the Middle East after the Iran war ends and in the final two years of President Trump's term. Two big events loom over the pBarak Ravid
  • 25
    Maria Bartiromo fired by Fox
    Fox News Media fired longtime host Maria Bartiromo for business reasons, a spokesperson confirmed to Axios on Thursday. Why it matters: Bartiromo is considered one of the network's most high profile hosts. The unexpected departure prompted questions about whether she would join the Trump administration. Bartiromo has developed a strong rapport with the president over the last decade. The White House didn't immediately respond to requests for comment. What they're saying: In a statement, Fox NewsSara Fischer
  • 26
    Scoop: Rep. Analilia Mejia launches super PAC to fund more progressive upsets
    After winning an upset primary victory over a field of better-funded, establishment-backed rivals, Rep. Analilia Mejia (D-N.J.) is launching a super PAC to help other progressive underdogs do the same, Axios has learned. Why it matters: Mejia is adding to a growing constellation of new progressive super PACs trying to compete with cash-rich groups like AIPAC that back more moderate Democrats in primaries. It is all part of a strategy that has helped the left pull off a series of high-profile priAndrew Solender
  • 27
    America's jobs scare recedes
    The labor market looks considerably healthier than it did at the start of the summer. Why it matters: Friday morning's jobs report recast the last two months as a period of modest job growth rather than of outright deterioration, capped by a gain last month that was more than triple what economists expected. The labor market is weathering the Iran conflict and an AI-driven transformation of the economy with surprisingly little damage, at least for now. For the Federal Reserve, the report puts inCourtenay Brown
  • 28
    The coming verdict on Trump's Strait of Hormuz claims
    The White House is at odds with some private sector analysts over how much oil is leaving the Strait of Hormuz — but it will take weeks until the picture of what's truly getting through becomes clear. Why it matters: How much oil exits the waterway sways the price of gasoline and other consumer products. The White House is claiming success in forcing open the world's most important energy chokepoint, which handled 20 million barrels of crude and oil products each day before the Iran war. The verBen Geman
  • 29
    AI models are becoming unknowable
    AI models may be getting safer while also getting harder to monitor. Which side of that seesaw prevails could determine whether AI is scaled safely or ruins civilization as we know it. Why it matters: Right now both options are running full steam and no one knows who's in charge of making sure the right one wins. State of play: OpenAI on Thursday released GPT-6 Astra , which president Greg Brockman said could eventually be seen as the start of artificial general intelligence, or AGI. Astra perfoMadison Mills
  • 30
    EU tech chief: AI guardrails inevitable for U.S.
    CHAPEL HILL, N.C. — The Trump administration used this week's G20 innovation summit to draw a bright line between U.S. and European approaches to AI regulation. The bloc's top tech official says the U.S. is ending up with many of the same guardrails anyway. Why it matters: The U.S. is urging countries to avoid Europe's sweeping approach to AI regulation, but the fight is shifting to which guardrails advanced AI needs and how governments put them in place. What they're saying: European CommissionMaria Curi
  • 31
    Diesel price surges to all-time high, fueled by wars
    The average price of diesel topped its 2022 record high on Friday , spiking to $5.85 per gallon, per AAA . Why it matters: Farmers, truckers and freight companies have been absorbing higher diesel costs since the Iran war began. The ballooning price threatens to further raise shipping costs while compounding a growing political problem for President Trump . It can also indirectly affect households by putting inflationary pressure on anything Americans buy that's carried by truck, which includesAvery Lotz
  • 32
    U.S. military reviewing deadly strike on Iranian wedding
    U.S. Central Command (CENTCOM) is conducting an internal review of a civilian casualty incident at a wedding during the strikes in Iran on Tuesday, U.S. officials say. Why it matters: While it still hasn't been confirmed that the incident was caused by the U.S. strike , it is one of the more significant civilian casualty incidents since the war began six months ago. The most significant civilian casualty incident happened on the first day of the war when a U.S. strike hit a school, killing 150 pBarak Ravid
  • 33
    House Democrat says his phone calls with a state legislator were wiretapped by FBI
    Rep. Suhas Subramanyam (D-Va.) said Thursday he was informed by the Department of Justice that his phone calls with a Virginia state legislator in the spring were wiretapped. Why it matters: The Democratic congressman is casting the tap as part of a pattern of the DOJ targeting President Trump's political enemies , calling the incident "deeply concerning." "They have politically investigated multiple states' Attorneys General and the former FBI director. Now they admit they have surveilled conveAndrew Solender
  • 34
    OpenAI unveils plan to protect critical services from AI cyberattacks
    OpenAI is launching a new initiative designed to provide subsidized access to its models to water systems, electricity providers, local governments and other critical services . Why it matters : Critical infrastructure organizations — which have long lacked the budget, manpower and time needed to shore up their cyber defenses — have been struggling to prepare for the anticipated wave of AI-enabled cyberattacks. Driving the news: OpenAI president Greg Brockman announced the new program while hostSam Sabin
  • 35
    "Welcome to the AGI era," OpenAI says as GPT-6 Astra debuts
    OpenAI on Thursday released GPT-6 Astra , which president Greg Brockman called a "generational leap" and said could eventually be seen as the arrival of artificial general intelligence, or AGI. Why it matters: Astra pushes AI agents closer to doing complex professional work on their own — while also raising questions about how safely they can be deployed. Driving the news: Brockman says he personally believes OpenAI has reached AGI, while leaving users to decide whether Astra meets that definitiIna Fried
  • 36
    A record 47% want stronger unions. Labor's challenge is turning support into power.
    For only the second time in a quarter century, a majority of Republicans — 52% — approve of labor unions , according to a new Gallup poll . Why it matters: Despite Americans' rising approval , union membership remains near a modern low. Congress has repeatedly blocked labor's biggest priorities, and House Republicans are now investigating unions, even as their own voters approve of them. Republicans' warmer feelings toward unions don't translate into wanting them to wield more power: 49% of RepuAndrew Pantazi
  • 37
    Not so fast on rate hikes, some Fed officials say
    After Federal Reserve chairman Kevin Warsh's speech in Jackson Hole , the markets penciled in an interest rate hike for the central bank's mid-September meeting. Comments from two influential officials now throw those expectations into question. The big picture: With August jobs and inflation data due out before a policy meeting in two weeks, the decision of whether to tighten policy appears to be on a knife-edge. A meaningful contingent of Fed leadership is eager to raise interest rates to addrNeil Irwin
  • 38
    Top Pentagon official reaffirms Anthropic blacklist despite Lutnick comments
    Pentagon official Emil Michael on Thursday said that Anthropic is still considered a " supply chain risk ," a day after Commerce Secretary Howard Lutnick said the company is "back on the right side." Why it matters: The Trump administration is not entirely aligned on one of the world's leading AI labs following months of fighting over how the military can use its models. What they're saying: "Anthropic is still a designated Supply Chain Risk at @DeptofWar and for the Defense Industrial Base. ThaMaria Curi
  • 39
    Widespread AI outage underway
    OpenAI's ChatGPT , Anthropic's Claude, and xAI's Grok are all down Thursday morning, according to the outage company Downdetector. Why it matters: Outages are common, but it's unusual for all three to go down at once, especially as the world increasingly relies on AI models. There have also been reported issues with Google's Gemini, but the company has not issued a confirmed outage notice. Driving the news: Anthropic's Claude dashboard is experiencing outages across Opus 4.8 and Opus 5, and saysJosephine Walker
  • 40
    Gloria Steinem, feminist icon, dies at 92
    Gloria Steinem, the journalist, activist and organizer who became the most recognizable face of the U.S. women's movement, died Wednesday at 92. The big picture: Steinem resisted being cast as feminism's singular face, but for six decades she carried its ideas into mainstream America across its second and third waves. Steinem made the personal political: An illegal abortion at 22 preceded decades of fighting first to legalize abortion and then to defend that right. Steinem died peacefully, surroAndrew Pantazi
  • 41
    Altman raises stakes on government scrutiny as AI advances
    CHAPEL HILL, N.C. — OpenAI CEO Sam Altman called the Trump administration's voluntary review of the company's forthcoming Astra model a "productive process" and said that closer engagement with government officials around the world will matter more as AI capabilities advance. Why it matters: OpenAI is close to releasing Astra, the first model designated at its "critical" cyber capability level that the company says requires stronger safeguards during development and before release. The Trump admMaria Curi
  • 42
    Brain implants fuel U.S.-China bioscience rivalry
    Brain implants like those being developed by Elon Musk's Neuralink are becoming the latest friction point in the intensifying U.S.-China rivalry in the life sciences. Why it matters: The devices are bringing new hope to patients with strokes, Parkinson's disease and other neurological conditions. And tech billionaires have fueled advances by pouring billions into biotech startups focused on linking the brain with computers. But a new national security analysis says U.S. development efforts are bAdriel Bettelheim
  • 43
    Markets can't ignore the war anymore
    Oil prices are heading back toward $100 a barrel, and, perhaps more crucially, the price of diesel fuel futures is now sitting at an all-time high. Why it matters: Investors are starting to recognize that higher oil prices from the Iran war aren't something they can continue to ignore. The big picture: Higher energy prices are showing up across the economy, driving up costs for companies, regular people and even governments around the world that are now facing higher borrowing rates. The risingEmily Peck
  • 44
    Historic first: Older adults now outnumber the youngest children for the first time
    Data: U.S. Census Bureau, International Database ; Chart: Noah Bressner/Axios People 65 and older now outnumber children ages 5 and younger worldwide for the first time in recorded history, according to a new report by the U.S. Census. Why it matters: A global plunge in fertility — demographers' term for the number of children being born — and increased longevity are creating a demographic inversion that's projected to deepen for decades. It's a trend that's shifting the historic balance betweenRussell Contreras
  • 45
    Alaska seeks oil revival while grappling with a warming future
    EXIT GLACIER, Alaska — America's 49th state is a study in contrasts: Rapidly melting glaciers like this one attract tourists, while the oil industry underpinning its economy is getting a fresh boost from Washington after decades of decline. Why it matters: Few places put the causes and consequences of global warming in such proximity, making Alaska a vivid window into the world's reliance on fossil fuels as climate impacts mount. Driving the news: "Alaska is a powerhouse," Energy Secretary ChrisAmy Harder
  • 46
    House Democrats erupt into infighting over leadership power play
    House Democratic leaders are trying to flex some muscle this week after two members defected on a key procedural vote. It's stoking fury among members across the political spectrum. Why it matters: The blowup offers a preview of the chaos that party leaders could face in January if they take back the House majority. Meanwhile, an effort to give House Democratic leaders more tools to keep their members in line next year is falling flat among rank-and-file members. "It'll be dead by next week," prAndrew Solender
  • 47
    Witkoff meets powerful UAE official amid crackdown on Iran
    White House Envoy Steve Witkoff met last weekend with the national security adviser of the United Arab Emirates to discuss next steps on Iran, according to two sources with knowledge of the meeting. Why it matters: The talks, which weren't announced by the White House and haven't been reported until now, took place amid the Trump administration's push to reopen the Strait of Hormuz while choking Iran economically. Witkoff met with Sheikh Tahnoon Bin Zayed al Nahyan (TBZ) on the Mediterranean islBarak Ravid
  • 48
    The $1 Trump coins are here. Why spending them is legal but tricky
    The U.S. Mint began selling $1 Trump coins Wednesday. Why it matters: The Trump coin sits in a rare middle ground: The Mint is promoting it as a collectible, but under federal law, it has a face value of $1 and can be used as real money. Driving the news: The U.S. Mint says the Trump coins are now "in circulation, so check your pocket change." The coins can be purchased in 25-coin rolls for $61 and 100-coin bags for $154.50. The U.S. Mint's website says the coins have "circulating finishes but hHerb Scribner
  • 49
    Nearly 9 in 10 Americans report widespread government corruption after Trump's return
    Data: Gallup ; Chart: Avery Lotz/Axios Americans' perceptions of government corruption have hit a 20-year high following President Trump's return to office, a new Gallup survey found. Why it matters: The results signal a deepening crisis of confidence in the federal government as trust in U.S. institutions already hovers near an all-time low. By the numbers: 89% of U.S. adults say government corruption is widespread, Gallup found . That's a 10-point increase from last year and well above the 72%Josephine Walker
  • 50
    Lutnick: Anthropic is "back on the right side" with Trump administration
    CHAPEL HILL, N.C. — Commerce Secretary Howard Lutnick said the Trump administration now trusts Anthropic following months of clashes with the AI company. Why it matters: The statement marks a remarkable turnaround in a relationship strained by a bitter fight over national security and AI safeguards . What they're saying: Asked if he trusts Anthropic CEO Dario Amodei, Lutnick told Axios' Mike Allen: "We trust Anthropic." "They've done what we asked. They're back on the right side. So the answer iMaria Curi
  • 51
    After Trump renamed Lake Ontario, Americans aren't buying his Canada fight
    President Trump's order renaming Lake Ontario to "Lake America" isn't floating with Americans, with a majority opposing the name change and his new tariffs on Washington's northern neighbor, per recent polling. The big picture: After trade talks between the longtime allies collapsed last month and triggered steep new tariffs, the Trump administration has escalated its feud with Canada. With the midterm elections creeping closer, polling suggests Americans aren't sold on the fight. Driving the neAvery Lotz
共 100 条热搜查看完整榜单
14天前更新
  • 01
    From CUDA to MLX: How K-Search Brings Decades of Kernel Expertise to Apple Silicon
    Figure 1: CUDA-to-MLX optimization translation map. CUDA optimization knowledge can be translated into architecture-native MLX strategies rather than copied instruction-for-instruction. We face a new epoch in computing. Hardware is changing rapidly — not just faster GPUs, but a growing range of chips from different vendors, each with its own architecture and often tailored to specific AI workloads. Software is changing just as fast, and AI coding tools now generate in minutes what took months of
  • 02
    Teaching LLMs to Update Beliefs for Efficient Long-Horizon Interaction
    Overview of ABBEL compared to traditional recursive summarization. Beliefs replace the full interaction history as the agent’s working context, and belief grading improves performance by supervising the contents of each belief state.. As task horizons grow, LLM contexts can’t scale forever. Self-summarization enables concise, interpretable contexts, but at a significant performance cost, especially for human assistance domains where high quality data is scarce, e.g., collaborative code generatio
  • 03
    Intelligence is Free, Now What? <br> Data Systems for, of, and by Agents
    ... government of the people, by the people, for the people ... — Abraham Lincoln, Gettysburg Address (1863) The cost of AI is dropping rapidly. GPT-4-class capabilities cost roughly $30 per million tokens in early 2023; today the same runs under $1 , and some providers are pushing costs below $0.10 . Across benchmarks, inference prices have fallen between 9x and 900x per year , with a median decline near 50x. Even frontier models are getting dramatically cheaper each generation, with open-sourc
  • 04
    2026 BAIR Graduate Showcase
    Congratulations to the Berkeley Artificial Intelligence Research (BAIR) Lab class of 2026! This year, BAIR celebrates another remarkable group of Ph.D. graduates whose curiosity, creativity, and perseverance have pushed the frontiers of artificial intelligence and machine learning. Their work spans the breadth of modern AI — robotics and embodied intelligence, large language models and reasoning, computer vision, generative modeling, AI safety, human-AI interaction, AI for science and healthcare
  • 05
    Adaptive Parallel Reasoning: The Next Paradigm in Efficient Inference Scaling
    Overview of adaptive parallel reasoning. What if a reasoning model could decide for itself when to decompose and parallelize independent subtasks, how many concurrent threads to spawn, and how to coordinate them based on the problem at hand? We provide a detailed analysis of recent progress in the field of parallel reasoning, especially Adaptive Parallel Reasoning. Disclosure: this post is part landscape survey, part perspective on adaptive parallel reasoning. One of the authors (Tony Lian) co-l
  • 06
    Gradient-based Planning for World Models at Longer Horizons
    GRASP is a new gradient-based planner for learned dynamics (a “world model”) that makes long-horizon planning practical by (1) lifting the trajectory into virtual states so optimization is parallel across time, (2) adding stochasticity directly to the state iterates for exploration, and (3) reshaping gradients so actions get clean signals while we avoid brittle “state-input” gradients through high-dimensional vision models. Large, learned world models are becoming increasingly capable. They can
  • 07
    Identifying Interactions at Scale for LLMs
    Understanding the behavior of complex machine learning systems, particularly Large Language Models (LLMs), is a critical challenge in modern artificial intelligence. Interpretability research aims to make the decision-making process more transparent to model builders and impacted humans, a step toward safer and more trustworthy AI. To gain a comprehensive understanding, we can analyze these systems through different lenses: feature attribution , which isolates the specific input features driving
  • 08
    Information-Driven Design of Imaging Systems
    An encoder (optical system) maps objects to noiseless images, which noise corrupts into measurements. Our information estimator uses only these noisy measurements and a noise model to quantify how well measurements distinguish objects. Many imaging systems produce measurements that humans never see or cannot interpret directly. Your smartphone processes raw sensor data through algorithms before producing the final photo. MRI scanners collect frequency-space measurements that require reconstructi
  • 09
    RL without TD learning
    In this post, I’ll introduce a reinforcement learning (RL) algorithm based on an “alternative” paradigm: divide and conquer . Unlike traditional methods, this algorithm is not based on temporal difference (TD) learning (which has scalability challenges ), and scales well to long-horizon tasks. We can do Reinforcement Learning (RL) based on divide and conquer, instead of temporal difference (TD) learning. Problem setting: off-policy RL Our problem setting is off-policy RL . Let’s briefly review w
  • 10
    What exactly does word2vec learn?
    What exactly does word2vec learn, and how? Answering this question amounts to understanding representation learning in a minimal yet interesting language modeling task. Despite the fact that word2vec is a well-known precursor to modern language models, for many years, researchers lacked a quantitative and predictive theory describing its learning process. In our new paper , we finally provide such a theory. We prove that there are realistic, practical regimes in which the learning problem reduce
共 10 条热搜查看完整榜单
全网热搜 - 今日热榜与实时热点聚合 - 第 3 页 | 什么火了