2026년 11월 9일

홈 타임라인 게시물 99개, 인기순

1위Mark Zuckerberg@finkdX에서 보기

Meta는 Muse Ember를 출시하며 같은 등급 모델 중 달러당 더 높은 지능을 제공하고, 도구 호출과 문서 파싱 등 토큰 집약적 에이전트 작업에 적합하다고 소개했다.

좋아요 7.2K리포스트 545답글 424노출 336.1K북마크 707

Mark Zuckerberg@finkd
Releasing Muse Ember today -- our smallest model that delivers higher intelligence per dollar than others in its class. Built for token-heavy work that agents repeat all day: tool calls, coding subtasks, routing, extraction, and long document parsing. Available today on the Meta Model API and in Muse Code.

2위Anthropic@AnthropicAIX에서 보기

Anthropic은 Claude Fable 5.2와 Claude Mythos 5.2를 출시했으며 코딩, 추론, 비전 성능이 향상되었다고 알린다.

좋아요 11.5K리포스트 744답글 402노출 506.4K북마크 770

Anthropic@AnthropicAI
Claude Fable 5.2 and Claude Mythos 5.2 are available now:
Claude@claudeai
Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges. They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.

3위The White House@WhiteHouseX에서 보기

트럼프 행정부 아래에서 야망과 추진력, 리더십이 되살아났고 NASA도 다시 돌아왔다고 강조하는 백악관 게시글이다.

좋아요 8.7K리포스트 1.5K답글 1.3K노출 353.9K북마크 216

4위Caleb@calebfahlgrenX에서 보기

누군가 Hugging Face에 자막이 달린 16×16 픽셀 아트 스프라이트 140만 개를 무료로 공개했다고 알린다.

좋아요 4.9K리포스트 367답글 39노출 292.5K북마크 6.1K

5위Martin Pestana@martin_pestana_X에서 보기

YC와 Afore의 투자를 받은 Volaren은 계좌 포트폴리오 변동을 설명하고 거래를 대신 실행하는 AI 에이전트를 공개했으며, 투자 위험과 면책 조항을 함께 안내했다.

좋아요 1.5K리포스트 443답글 280노출 2.5M북마크 1.5K

Martin Pestana@martin_pestana_
We raised $1M from YC & Afore to leave Goldman Sachs & build an agent that trades for you. The top 1% gets a “money guy.” Today we’re introducing Volaren so you can have one too. Volaren is designed to plug into your brokerage accounts and explain why your portfolio moved, suggest what to do next, and trade it for you. • Web, app, or text message • Copy trade top investors/politicians • Deep financial analysis • Agents that reason like humans to trade your best ideas Try it now: volaren.com Investing involves risk, and past performance does not guarantee future results. For informational purposes only, not a recommendation or investment advice. Other fees may apply, please review Volaren’s website and disclosures in full.

6위Soham Chakraborty@sochak03X에서 보기

STASIS 스마트 안경은 눈 깜빡임과 찡그림을 읽어 피로와 수면을 파악하며 11월 출시를 앞두고 창립 멤버 주문을 받는다고 소개한다.

좋아요 527리포스트 95답글 52노출 575.1K북마크 348

Soham Chakraborty@sochak03
A first look at what STASIS (@ycombinator F26) has been up to. Your watch knows your heart rate. Your ring tracks your sleep. But after 9+ hours on a screen, you guess when it comes to your eyes. STASIS is the first smart glasses that read every blink, every squint, & every stare to understand your fatigue, cognitive performance, & sleep like never before. Smart lenses tint right when your eyes need it. Building the wearable for 16 billion eyes. Launching this November. Founding Member orders now open.

7위Ricursive Intelligence@RicursiveAIX에서 보기

자체 사전학습한 650억 매개변수 밀집 LLM인 Ripple-65B Preview를 공개하며 에이전트 코딩 과제에서 미국 오픈 웨이트 모델 중 최고 성능을 주장하는 발표이다.

좋아요 4.5K리포스트 537답글 393노출 765K북마크 2.6K

8위Google AI Studio@GoogleAIStudioX에서 보기

Google AI Studio에 Google Maps Platform이 통합되어, 사용자가 아이디어를 설명하면 Gemini가 지도 앱을 만들어 주는 기능이 추가되었다.

좋아요 2.3K리포스트 201답글 96노출 174.4K북마크 1.2K

10위a16z@a16zX에서 보기

a16z는 Roblox에서 AI 게임 개발이 급증하고 있다고 전하며, 상위 게임 출시 증가, MCP 일일 사용자 급증, 에셋 업로드 증가를 근거로 제시했다.

좋아요 616리포스트 52답글 48노출 63K북마크 297

a16z@a16z
AI game development is exploding on Roblox: - An growing share of top games were launched in the last 90 days - Roblox MCP daily users are exploding - An increasing fraction of game development time is spent using AI - Asset uploads are shooting up In other words, people are using AI to make games, and other people are actually playing them. More charts: a16z.news/p/charts-of-th…

12위Salim Ali@Salim_AIiX에서 보기

오퍼스 5.5 에이전트로 가우스, 리만 등의 노트 약 1만 2천 쪽을 분석해 리만의 미출간 자료에서 랭킨-위고니오 조건을 찾았다고 주장하며, 위대한 수학자들의 연구를 AI로 공개·검토하자고 제안한다.

좋아요 2.6K리포스트 247답글 59노출 149.4K북마크 876

Salim Ali@Salim_AIi
Been running Opus 5.5 agents over the notebooks of Gauss, Riemann, Hilbert and Grothendieck. around 12,000 pages so far. Heres the best finding so far: Riemann's 1860 shock wave paper is famous for getting the energy part wrong however his unpublished sheets at Göttingen have it right the Rankine Hugoniot conditions a decade or two before Rankine and Hugoniot plus results usually dated 19061937, all in his own notation. A second sheet in another folder has the same physics. There are probably thousands of potential undiscovered stuff in the research papers of the smartest ever to live imo all of them should be made public and reviewed by ai years of work can probably be transcribed by agents in days

13위OpenAI@OpenAIX에서 보기

OpenAI가 연구 인프라 전반의 모델 작업을 추적하고 안전 우려 시 35초 안에 출력을 중단할 수 있는 Control Room 시스템을 소개한다.

좋아요 2.3K리포스트 147답글 388노출 297.6K북마크 598

OpenAI@OpenAI
We built Control Room, a central system to track model workloads across our research infrastructure and stop them quickly when safety or security concerns arise. It connects every research inference workload, including agentic sampling for model training and evaluation, data processing, and more. This works together with our misalignment monitoring systems, and halts model output within 35 seconds of a decision to intervene to stop or pause a job. Control Room is just one part of our holistic approach for safe frontier model training and deployment.
openai.comControl Room: A central system to track and stop model workloads across our research infrastructure

14위Patrick Collison@patrickcX에서 보기

patrickc가 한 기술 억만장자가 샌프란시스코 공원에 1억 달러를 기부하기로 했다는 기사를 공유하며 bchesky를 칭찬했다.

좋아요 961리포스트 25답글 22노출 188.3K북마크 231

15위OpenAI Developers@OpenAIDevsX에서 보기

OpenAI Devs는 여러 로컬 컴퓨터를 같은 도구에 연결해 작업마다 사용할 기기를 고를 수 있게 되었으며, 전용 기기에서 앱을 빌드하고 테스트할 수 있다고 알린다.

좋아요 4.2K리포스트 258답글 426노출 300.5K북마크 957

16위Greg Brockman@gdbX에서 보기

오픈테이블이 오픈AI와 네이티브로 연동해 9월 한 달에만 19만 명을 식당으로 보내는 성과를 냈다는 내용이다.

좋아요 635리포스트 29답글 86노출 52.2K북마크 138

Greg Brockman@gdb
integrating with OpenAI can drive a lot of business:
Juozas Kaziukėnas@juokaz
OpenTable's native integration with OpenAI drove 190,000 people to restaurants in just the month of September. It has to be one of the biggest agentic commerce success stories - a number so high I'm triple checking if the CFO could have meant something different.

17위AI at Meta@AIatMetaX에서 보기

Meta는 복합 추론과 에이전트 업무에 맞춘 소형 모델 Muse Ember를 소개하며, 앱 간 워크플로 수행, 도구 순차 호출, 코드 수정 및 1M 토큰 컨텍스트와 다중 모달 이해 기능을 강조했다.

좋아요 2.4K리포스트 255답글 155노출 648.8K북마크 450

AI at Meta@AIatMeta
Today we're introducing Muse Ember, our smallest and most efficient model, built for complex reasoning execution and everyday agent work. Muse Ember can carry workflows across apps, call tools in sequence, run tests and make targeted code fixes. With a 1M-token context window and native understanding of text, images, audio and video, it can also work across large document sets and varied inputs.

18위Amin Vahdat@AminVahdatX에서 보기

구글이 8세대 TPU 제품군으로 TPU 8i Max와 8tx를 내놓고, 파이토치 실행과 KV 전송 지연 감소 등 개발자용 도구를 함께 공개했다는 소식이다.

좋아요 3.2K리포스트 320답글 89노출 408K북마크 562

Amin Vahdat@AminVahdat
Thrilled to share a big milestone from our AI & Infrastructure team at @Google. Today we’re expanding our 8th generation TPU portfolio with: • TPU 8i Max: Built for massive context windows and the highest compute density per rack • TPU 8tx: Purpose-built for secure, confidential inference This marks the first time we’ve introduced four distinct TPU systems in a single generation, in one year. For developers, hardware is only as good as the software stack. And we want you to be able to build with TPUs as easily and efficiently as we do at Google. You can now: ↳ Run native PyTorch & Hugging Face models via TorchTPU (zero code rewrites) ↳ Cut KV transfer latency >90% with open-source TPU Sync ↳ Eliminate idle compute waste with Agent Compute Engine Read more about our newest TPUs on our blog: I look forward to connecting with builders, researchers, engineers and organizations pushing the boundaries of AI and computing!
blog.googleOur newest TPUs: 8tx and 8i Max, built for the agentic era

19위Alexandr Wang@alexandr_wangX에서 보기

Muse Ember는 Haiku, Luna 등 다른 플래시 모델보다 달러당 더 높은 지능을 제공하며 비용은 3분의 1 이하라고 소개하는 소형 효율 모델 발표 글이다.

좋아요 4.6K리포스트 249답글 222노출 247.4K북마크 752

Alexandr Wang@alexandr_wang
1/ announcing our smallest, most efficient little model is here: muse ember 🔥 built for work that agents repeat all day, it gets you more intelligence per dollar than Haiku, Luna, and all other flash models at any reasoning setting and does it at a third of the cost or less!

20위DHH@dhhX에서 보기

맥북에서 Omarchy 가상머신을 이용해 macOS를 실행하는 데 성공했다는 글이다.

좋아요 735리포스트 23답글 60노출 30.4K북마크 89

21위DHH@dhhX에서 보기

NVIDIA에서 DGX Spark 두 대를 받아 Omarchy 설치를 마쳤고 로컬 AI 기능을 조정 중이며 4.6 버전에 포함될 예정이라는 감사 글이다.

좋아요 3.8K리포스트 103답글 270노출 164.5K북마크 303

DHH@dhh
Many thanks to @nvidia for sending me a pair of DGX Sparks! Not only are the machines amazing, but they look absolutely incredible too. Omarchy is fully up and running now! Working to tune the out-of-box local AI offerings. All shipping in 4.6 shortly 🤘

22위Sundar Pichai@sundarpichaiX에서 보기

Sundar Pichai는 Google이 한 세대에 네 가지 TPU 시스템을 1년 안에 선보였다고 발표하며, TPU 8i Max와 8tx 등 새 제품과 PyTorch 호환 등 개발자 도구를 소개한다.

좋아요 2.9K리포스트 202답글 150노출 174.7K북마크 202

Sundar Pichai@sundarpichai
Four distinct TPU systems in a single generation in one year! Instead of waiting years for a traditional semiconductor cycle, hardware engineers co-designing systems alongside Google Deepmind means we can keep innovating quickly for customers.
Amin Vahdat@AminVahdat
Thrilled to share a big milestone from our AI & Infrastructure team at @Google. Today we’re expanding our 8th generation TPU portfolio with: • TPU 8i Max: Built for massive context windows and the highest compute density per rack • TPU 8tx: Purpose-built for secure, confidential inference This marks the first time we’ve introduced four distinct TPU systems in a single generation, in one year. For developers, hardware is only as good as the software stack. And we want you to be able to build with TPUs as easily and efficiently as we do at Google. You can now: ↳ Run native PyTorch & Hugging Face models via TorchTPU (zero code rewrites) ↳ Cut KV transfer latency >90% with open-source TPU Sync ↳ Eliminate idle compute waste with Agent Compute Engine Read more about our newest TPUs on our blog: bit.ly/3VfETPR I look forward to connecting with builders, researchers, engineers and organizations pushing the boundaries of AI and computing!

23위NASA@NASAX에서 보기

주말이 끝나도 임무는 계속된다며 미국의 야망을 강조하는 NASA의 게시물이다.

좋아요 1.5K리포스트 242답글 63노출 484.4K북마크 101

24위NASA@NASAX에서 보기

NASA는 MAX POWER 행사에서 새로운 우주비행사 모집, 달 기지 진전, 100만 달러 규모 학생 설계 대회 등 여러 발표를 했다고 정리했다.

좋아요 885리포스트 126답글 86노출 255.6K북마크 44

26위Aritra 🤗@ariG23498X에서 보기

Hugging Face의 이번 주 소식으로 MoE 학습 핸드북, 학습 하네스 어댑터, 하네스 리더보드가 공개되었다.

좋아요 633리포스트 73답글 11노출 28.7K북마크 739

27위Caleb@calebfahlgrenX에서 보기

Hugging Face는 .glb, .ply, .stl, 스플랫 파일을 올리면 3D 갤러리가 자동으로 만들어지는 3D 에셋 공유 공간이 되었다고 소개한다.

좋아요 162리포스트 16답글 9노출 13.7K북마크 148

28위Nick Bushak@nbushakX에서 보기

Boxes는 기존 구독을 활용해 클라우드에서 Claude Code와 Codex를 여러 개벼 개발 환경으로 병렬 실행할 수 있는 YC F26 기업으로, 프로 요금제 3개월 무료 혜택을 제공한다.

좋아요 337리포스트 19답글 77노출 24.4K북마크 302

29위Fabian Zbinden@Bartonaut55X에서 보기

YC F26 기업 0cloud는 코드 배포 전에 취약점을 찾아내고 원클릭 수정안을 제공하는 보안 제품을 소개했다.

좋아요 161리포스트 13답글 26노출 15.9K북마크 138

Fabian Zbinden@Bartonaut55
Software can now secure itself Introducing 0.security (YC F26) We found 30+ Linux kernel 0-days, + 500+ CVEs incl. government code 🇨🇭 🇪🇺 🇮🇹 and worked with EY & defense organizations. Then we got backed by @ycombinator Now we're bringing frontier security to everyone. With 0cloud, we find vulnerabilities before your code ships and deliver the 1-click fix.

30위younes@yb2698X에서 보기

리퀴드AI 및 웹AI 모델용 하네스 어댑터 학습을 마치고 도구 효율과 정확도가 향상된 9개 어댑터와 GGUF 파일을 공개했다는 내용이다.

좋아요 163리포스트 20답글 11노출 11.9K북마크 128

younes@yb2698
Harness adapters training has now finished! @liquidai LFM2.5-24B, LFM3 + @thewebAI Twil-8B optimized for pi, hermes and opencode (up to 43% tool efficiency + 15% accuracy gain on our held-out data-analysis test set for LFM3 🤯 / up to 37% tool call efficiency on Twil-8B model). More than 1900 RL steps trained using @huggingface jobs. 9 total adapters with GGUFs which are ready to use! Reward hacking attempts blocked for all models with safeagents (and we opensource the attempts). Let's explore together the outcome of this open experiment 👇 huggingface.co/collections/ha… huggingface.co/collections/ha… huggingface.co/collections/ha…

31위Emma Scharfman@EmmaScharfmannX에서 보기

Hugging Face의 Science on the Hub 스페이스에 따르면 2026년 들어 과학 모델이 1만 8천 개 공개되어 같은 기간 2025년의 6,100개보다 크게 늘었다.

좋아요 68리포스트 14답글 5노출 7.9K북마크 45

Emma Scharfman@EmmaScharfmann
I see more and more AI for Science releases on @huggingface, but I was not sure how much they had really increased (and I thought I might be biased) 😇 Then, I found this space which shows how much science releases have increased over the last year on the Hub. So far in 2026, there have been 18k new science models against 6.1k in the same period of 2025! I thought I might be biased, but in fact, I was underestimating it 🚀 🤗 Science on the Hub space: Thanks @evijit for making this space available and @cgeorgiaw for creating such an amazing AI for Science community!
huggingface.coScience on the Hub - a Hugging Face Space by evijit

32위Randall Balestriero@randall_balestrX에서 보기

학회 첫 주요 업데이트로 맨해튼 중심 장소가 확정되었고, 12월 22일이 논문 마감이며, 단일 블라인드 자유 형식으로 AI 사용이 허용되고, OpenReview에 공개되며 심사위원당 논문은 3편 미만이라고 알린다.

좋아요 74리포스트 15답글 1노출 7K북마크 33

33위a16z@a16zX에서 보기

a16z 패널은 Roblox에서 AI 게임 개발이 폭발적으로 늘고 실제 플레이어가 있다는 점이 많은 바이브코딩 앱과 다르다고 논의한다.

좋아요 70리포스트 11답글 12노출 13.6K북마크 31

a16z@a16z
a16z's Alex Danco and Moses Sternstein on how Roblox is unlocking AI game development: "We're seeing this explosion of app development with AI, but in a lot of the cases, the apps don't really have any users. They seem to be, creating a lot of vibecoding stuff, but to what end?" "But a big outlier here for this has been gaming, and the standout within the outlier has been Roblox." "Across the board people are clearly using the various AI game development tools across the ecosystem." "Maybe it's just, again, another indication that it's very early. You saw a year of people vibecoding stuff and without much effect, but maybe people are getting better at it, or maybe the tools are getting better at it."
a16z@a16z
AI game development is exploding on Roblox: - An growing share of top games were launched in the last 90 days - Roblox MCP daily users are exploding - An increasing fraction of game development time is spent using AI - Asset uploads are shooting up In other words, people are using AI to make games, and other people are actually playing them. More charts: a16z.news/p/charts-of-th…

34위Alexander Belanger@abelanger5X에서 보기

Hatchet v2와 570만 달러 시드 투자 유치를 발표하며, 내구성 있는 실행과 큐, 스트림을 제공하는 범용 오케스트레이션 계층을 소개한다.

좋아요 91리포스트 25답글 32노출 9.7K북마크 39

Alexander Belanger@abelanger5
despite coding agents, modern apps are fragile. without proper architecture, you can spit out as much code as you’d like, but you’re just adding levels to a house of cards. introducing Hatchet v2 & our $5.7M seed round. our bet is that great orchestration is the difference between a reliable app and a broken one. Hatchet v2 is the universal orchestration layer: durable execution, durable queues and durable streams, sandbox integrations, serverless support, and operators. this is the architecture used by high-growth companies like @greptile, @linear, @getenter_ai, and many others, who collectively process over 5 billion tasks/month on Hatchet. we’ve spent over two years obsessing over making our engine performant and scalable. this release extends the existing @hatchet_dev engine to support any sandbox provider, serverless environment, edge function, or container.

35위AirCaps AI@tryaircapsX에서 보기

화자 구분 음성 인식 모델 A5S Async V1이 VoiceArena 1위를 차지했고 경쟁 모델보다 오류가 41% 적다며 벤치마크를 공개했다고 밝힌다.

좋아요 123리포스트 34답글 30노출 122.5K북마크 51

AirCaps AI@tryaircaps
Introducing A5S Async V1: the world's most accurate speech recognition model for speaker-attributed transcription. #1 on VoiceArena, ahead of ElevenLabs, AssemblyAI, Deepgram, NVIDIA, Meta & more. The next-best model makes 41% more errors across 5 public benchmarks. And yes, every lab says they're #1. So we open-sourced our benchmark. Try it now👇

36위Joël Niklaus@joelniklausX에서 보기

OpenCode 2는 Hugging Face 하네스 리더보드에서 OpenCode 1.19를 크게 앞서 상위권에 올랐고, 해결 과제당 비용도 3.35달러에서 0.45달러로 줄었다.

좋아요 611리포스트 35답글 61노출 71.6K북마크 246

Joël Niklaus@joelniklaus
OpenCode 2 far outscores OpenCode 1.19: from 8th of 9 harnesses to tied for first on all three models. It replaces 1.19 in the @huggingface Harness Leaderboard, which compares coding agents on the same models. Like most harnesses there, it now uses the Chat Completions API directly, so it's more comparable. At high effort it solves 65.7% of tasks instead of 43.3% on GLM-5.3-Flash and 69.4% instead of 57.0% on DeepSeek-V4.1-Flash. On Flint, 22.9% instead of 7.1%. Cost per solved task fell from $3.35 to $0.45. Part of the old gap was the setup. Harbor's standard route for OpenCode 1.19 is OpenAI's Responses API, which I translated to Chat Completions so it works with Hugging Face Inference Providers. It ran without errors but silently dropped the model's earlier reasoning. Thanks to @thdxr and @rekram11 from @opencode for spotting it! Leaderboard: huggingface.co/spaces/joelnik…

37위Paige ✈️ CoRL@paigeinsfX에서 보기

로봇 RL 환경을 구축하는 se3labs가 Y Combinator의 지원을 받아 출범했으며, 실제 작업에서 정책이 실패하는 지점을 찾아 그 데이터를 기반으로 학습을 개선하는 서비스를 제공한다고 밝혔다.

좋아요 203리포스트 23답글 50노출 14.3K북마크 79

Paige ✈️ CoRL@paigeinsf
Today we're launching @se3labs, backed by @ycombinator. Robot models are getting better fast, but there's still a big gap between a great demo in sim or the lab and a robot that works reliably on a real task. We're building real-world RL environments for robots. We start with tasks from industry partners, run policies on real robots, and find exactly where they fail. Those failures decide what data we collect and curate for the next policy update. Next, we're bringing this to partner sites with physical sandboxes for real-world RL, so policies can train on the exact tasks and conditions they'll face in practice. If you’re training robot policies or have a real task you want robots to do, we’d love to talk: se3labs.ai or DM me directly! See you at CoRL 🤖👋

38위Google for Developers@googledevsX에서 보기

Google Antigravity에 원격 에이전트가 출시되어, 노트북을 닫은 뒤에도 클라우드 환경에서 작업과 풀 리퀘스트 준비를 계속할 수 있다고 알린다.

좋아요 129리포스트 19답글 7노출 10.1K북마크 48

39위Mario Rodriguez@mariorod1X에서 보기

슬랙과 팀즈 기능이 정식 출시되었고 주간 릴리스 주기로 제공된다는 안내다.

좋아요 47리포스트 13답글 6노출 14.9K북마크 15

40위Greptile@greptileX에서 보기

Greptile은 엔지니어들이 여러 에이전트로 수백 개의 PR을 만드는 상황에서 이를 빠르고 안전하게 이해하고 병합하도록 돕는 새 코드 리뷰 웹 앱을 공개했다.

좋아요 101리포스트 9답글 8노출 11.3K북마크 32

Greptile@greptile
Introducing a new way to review code, faster than ever. Engineers today might kick off dozens of agents that produce hundreds of PRs within hours. Greptile’s new web app is designed to help you understand and safely merge them as fast as possible.

41위Google Cloud@googlecloudX에서 보기

Google Cloud는 AI 인프라 최고 기술 책임자와 함께 새로운 TPU에 대해 설명하는 생중계를 진행한다고 알렸다.

좋아요 73리포스트 11답글 0노출 12.1K북마크 18

42위Google for Developers@googledevsX에서 보기

Google 개발자 계정이 Antigravity의 음성 모드를 소개하며 음성으로 코드 작성, 파일 생성, 명령 승인과 작업 중 요구사항 변경이 가능하다고 설명했다.

좋아요 128리포스트 13답글 10노출 11.5K북마크 31

Google for Developers@googledevs
🎙️ Start building hands-free with voice mode in @Antigravity Built with our Gemini Live models, voice mode lets you pair-program, brainstorm, and guide multi-step builds in your spoken language: • Use your voice to write code, generate local files, and spin up interactive UI previews • Approve terminal commands and permission prompts out loud or interrupt mid-task to change requirements • Drop in files, such as PDFs or screenshots, mid-conversation Watch Antigravity engineer Tawsif localize and style a website component in different languages using just his voice ⤵️

43위Tavus@tavusX에서 보기

Tavus가 커피 키오스크에 PAL을 설치해 주문을 가로채려는 시도를 음성 주체 추적 모델로 막았다고 밝히고 키오스크 모드 파일럿 참여자를 모집했다.

좋아요 54리포스트 8답글 7노출 6.5K북마크 11

Tavus@tavus
We put a Tavus PAL on a coffee kiosk and had someone try to hijack every order. No directional mics. Our audiovisual isolation model tracks who’s speaking to it, so he (the faded captions) never got an answer. We’re opening a kiosk mode pilot. Want in? tavus.io/demo

44위Google Gemma@googlegemmaX에서 보기

젬마 기반 모델을 로봇에 탑재해 대화, 도구 호출, 비전 처리를 8GB 젯슨 한 대에서 구현한 엔비디아의 사례를 소개한다.

좋아요 215리포스트 17답글 14노출 12.8K북마크 40

45위Vercel@vercelX에서 보기

Vercel이 2026년 11월 AI 게이트웨이 지수를 공개하며 신규 유료 팀이 OpenAI보다 Anthropic 모델을 약 2배 가까이 선택했다고 밝혔다.

좋아요 64리포스트 5답글 11노출 9K북마크 11

46위GitHub@githubX에서 보기

Claude Fable 5.2가 GitHub Copilot에서 정식 제공되며, 이전 버전보다 적은 단계와 도구 호출로 더 많은 작업을 해결했다고 밝혔다.

좋아요 76리포스트 10답글 18노출 10.9K북마크 13

GitHub@github
It's here ➡️ @AnthropicAI's Claude Fable 5.2 is now generally available in GitHub Copilot. Our early testing shows Fable 5.2: • solved more tasks than Fable 5.1 while using fewer steps and tool calls • completed difficult tasks that earlier models missed Try Fable 5.2 for complex, multistep coding jobs across large codebases. Find it in the GitHub Copilot app, CLI, or @code.
github.blogClaude Fable 5.2 in GitHub Copilot - GitHub Changelog

47위Vals AI@ValsAIX에서 보기

ValsAI는 Meta의 Ember가 Vals Index에서 테스트당 0.15달러로 가장 저렴하고 약 15분 만에 테스트를 끝내 상위 12개 모델 중 가장 빠르다고 전한다.

좋아요 181리포스트 18답글 9노출 40.9K북마크 28

Vals AI@ValsAI
Meta's Ember is the cheapest model on the Vals Index at just $0.15 per test outperforming both Haiku 5.5 and GPT-6 Luna. It also finishes tests in about 15 minutes, making it the fastest model in the top 12.

48위a16z@a16zX에서 보기

금문교는 예정보다 빨리 예산 내에 완공됐지만 지금은 공중화장실 하나 짓기도 어려운 현실을 지적하며, 결과보다 절차를 중시하는 제도를 바꿔 인프라를 빠르게 건설해야 한다고 주장한다.

좋아요 245리포스트 22답글 15노출 25.7K북마크 37

a16z@a16z
In 1937, San Francisco finished the Golden Gate Bridge, initially derided as "the bridge that couldn't be built," ahead of schedule and under budget. Nearly a century later, the same city couldn't build a single public toilet without exorbidant costs and years of delay. Across America, communities face the same dysfunction: stalled factories, delayed energy projects, and crumbling infrastructure. We should be building faster and more safely than ever before. Instead, we struggle to deliver the basics. The problem: our system rewards process over results. A growing bipartisan consensus offers a chance to change that. a16z's Matt Cronin on the 5 priorities bipartisan legislation must address to get America building again: a16z.news/p/america-can-…

49위Viswa Kotra@viswa_kotraX에서 보기

Nodus Compute는 GPU가 평균 5%만 사용된다는 점을 지적하며, 유휴 시간을 채워 카드 하나에 더 많은 모델과 작업을 돌릴 수 있는 플랫폼을 소개한다.

좋아요 69리포스트 11답글 32노출 6.1K북마크 9

Viswa Kotra@viswa_kotra
The world is borrowing hundreds of billions of dollars to build GPUs, and the average GPU is busy 5% of the time. Nodus Compute gets more AI out of the GPUs that already exist. We fill idle hours with useful work, pack more models and jobs onto every card, and run training, RL, agents and inference on one platform.

50위Michelle Ko@michelleekokokoX에서 보기

Meta의 Muse Ember는 반복적인 에이전트 작업을 위해 만들어진 가장 작고 효율적인 모델로, 도구 호출과 코딩 하위 작업, 문서 파싱 등 대규모 작업에 적합하다고 소개하고 있다.

좋아요 131리포스트 14답글 7노출 7K북마크 16

51위Ordo Labs (YC F26)@OrdoLabs_X에서 보기

Ordo Labs는 규제 실험실의 데이터 분석과 보고서 작업을 자동화하는 서비스를 출시했으며, 유료 파일럿에서 분석 후 작업 시간을 90% 이상 줄였다고 밝혔다.

좋아요 113리포스트 14답글 52노출 6.3K북마크 13

52위DHH@dhhX에서 보기

초기 윈도우 출시 실패로 저렴해진 스냅드래곤 X1 PC가 여전히 성능과 에너지 효율이 좋으며, Omarchy로 전 라인업을 지원할 예정이라는 글이다.

좋아요 465리포스트 9답글 28노출 18.4K북마크 45

DHH@dhh
The original X1 @Snapdragon PCs are still very capable and shockingly energy efficient. We'll soon have the entire range covered with Omarchy. There are some stellar bargains to be had on these devices because Windows flopped so hard with the initial rollout. This is the A14 X1.

53위Vishnu Krishnaprasad@an_engineer_logX에서 보기

YC 참여 후 두 사업 중 하나에 집중하기로 결정했고, 에이전트용 비용 최적화 샌드박스 매출 런레이트가 약 15배 성장했다는 회고이다.

좋아요 94리포스트 3답글 12노출 4.8K북마크 9

Vishnu Krishnaprasad@an_engineer_log
Getting into YC has pushed me harder than I ever imagined. Watching my batch mates do incredible things is the best kind of peer pressure. When we started YC we had two revenue streams. One was most of our revenue but growth rate had plateaued. The other bet was to build cost optimized sandboxes for agents. This was smaller but growing. For a while I tried to do both and had no focus. But after a lot of conversations with our YC partner, it became clear that we had to double down on one. So here we are... our sandbox revenue run rate has grown nearly 15X since and I'm grateful to YC.

54위William Shen@shenbokuiX에서 보기

Muse Ember의 첫 버전을 공개하며 GW당 추론 지능을 극대화하고 사용자 피드백을 받아 개선하겠다고 밝힌다.

좋아요 223리포스트 27답글 17노출 15.6K북마크 20

William Shen@shenbokui
Our first version of Muse Ember 🔥 Ember aims to maximize intelligence per GW of inference to help bring personal superintelligence to everyone. We would love all the feedback to keep hill-climbing. Please send it our way. We will keep hill-climbing on efficiency, user experience, and better approach to train with internal frontier! More to come 🫡
Alexandr Wang@alexandr_wang
1/ announcing our smallest, most efficient little model is here: muse ember 🔥 built for work that agents repeat all day, it gets you more intelligence per dollar than Haiku, Luna, and all other flash models at any reasoning setting and does it at a third of the cost or less!

55위Kevin McKernan@Kevin_McKernanX에서 보기

화이트헤드 연구소가 Sabatini 소송 사건을 중립적으로 평가하지 않고 Knouse와 협력해 법률비용을 지원했다며, 공개된 합의 문서를 근거로 마녀사냥이라고 비판한다.

좋아요 187리포스트 37답글 6노출 43.3K북마크 16

Kevin McKernan@Kevin_McKernan
WTF? @WhiteheadInst didn’t run a neutral evaluation of the event. They teamed up with Knouse and paid her legal bills. An absolute Cotton Mather style DEI witch-hunt @BillAckman
ScienceGuardians@SciGuardians
🚨 BOMBSHELL COURT EVIDENCE EXPOSES A SCANDAL OF SECRECY, DECEPTION CONCERNS, AND DOCUMENTED JOINT-DEFENSE COORDINATION INVOLVING KRISTIN KNOUSE, RUTH LEHMANN, THE WHITEHEAD INSTITUTE, AND HHMI IN THE DAVID M. SABATINI ORDEAL. A confidential financial settlement. Payments to Kristin Knouse and her lawyers. Legal protections extending to Ruth Lehmann. Documented joint-defense coordination. And an explicit provision restricting disclosure of the agreement to David Sabatini. 🔱 The ScienceGuardians™ Legal and Investigative Team has obtained the six-page settlement agreement filed in Massachusetts Superior Court on February 6, 2025. The document reveals consequential arrangements involving Kristin Knouse, the Whitehead Institute (@WhiteheadInst), the Howard Hughes Medical Institute (HHMI, @hhmi_science), and Whitehead Director and President Ruth Lehmann (@REMLehmann). These are not speculations. The provisions appear in the court-filed agreement itself. 🔴 1. Whitehead and HHMI agreed to pay Kristin Knouse and contribute to her legal defense. Section 1 establishes that Whitehead and HHMI agreed to jointly fund a settlement payment, including compensation to Knouse and a contribution toward her attorneys' fees and costs in connection with the Sabatini litigation. The financial amounts are redacted in the court-filed agreement, although we believe the total settlement may have reached SEVEN figures. How much money did Whitehead and HHMI actually commit to Knouse and her legal defense? The scientific community deserves full transparency regarding the financial scale of these arrangements and the institutional justification for such payments. 🔴 2. Ruth Lehmann's leadership was expressly covered by the legal releases. Section 2 establishes broad mutual releases between Knouse, Whitehead, and HHMI. Whitehead's release expressly includes its directors and senior administration, specifically naming Ruth Lehmann. These provisions were not limited to an informal understanding between Knouse and her former employer. They created formal contractual protections extending across the participating institutions and their leadership. 🔴 3. Knouse, Whitehead, and Lehmann agreed to continue their joint defense. Section 4 is particularly revealing. It expressly provides for continuing cooperation between Knouse and Whitehead, including its senior administration and Ruth Lehmann, concerning matters related to Sabatini's lawsuit. The agreement further states: “Whitehead and Dr. Knouse further agree to continue to work pursuant to their common interest and joint defense privilege in the context of the Sabatini Lawsuit.” Their lawyers were also expected to cooperate and share work where practicable. The document therefore establishes not only financial arrangements but also an explicit commitment to continued legal cooperation between Knouse and Whitehead's leadership. 🔴 4. The agreement expressly restricted disclosure to David Sabatini. Section 5 contains one of the document's most consequential provisions: “This Agreement shall not be provided to Dr. Sabatini unless required by law.” The same section establishes procedures allowing parties to seek protective orders in response to compulsory disclosure requests. A financial settlement and continuing joint-defense agreement, expressly restricted from disclosure to the very scientist who was litigating against these parties. The existence of this restriction demands serious examination of what was disclosed during litigation, when it was disclosed, and whether Sabatini received the information to which he was legally entitled. 🔴 5. Whitehead and HHMI expressly denied liability while agreeing to the payments and releases. Section 6 characterizes the settlement as a compromise of disputed claims, intended to avoid further expense, and expressly disclaims any admission of liability. The absence of an admission does not erase the financial payments, mutual releases, continuing joint defense, or disclosure restriction documented in this agreement. ⚠️ Ruth Lehmann and Whitehead's leadership must answer for these arrangements. This court-filed document raises substantial questions about the financial and legal relationships among the individuals and institutions involved in Professor @DMSabatini's ordeal. 📑 EXAMINE THE ORIGINAL COURT EVIDENCE: All SIX pages of the settlement agreement are provided individually in high resolution in the thread below. Scroll down to examine the documented provisions firsthand. 🔱 ScienceGuardians™ Legal and Investigative Team MORE COURT EVIDENCE. MORE REVELATIONS. STAY TUNED. ⚠️ @SciGuardians 🔱

56위Thomas Kurian@ThomasOrTKX에서 보기

Google Cloud는 보안 추론용 TPU 8tx와 대규모 컨텍스트 처리용 TPU 8i Max를 발표하며, 올해 공개한 네 가지 새 TPU 시스템으로 제품군을 넓혔다.

좋아요 265리포스트 25답글 15노출 9.2K북마크 18

Thomas Kurian@ThomasOrTK
At Google Cloud Next in April, we introduced two new specialized TPUs, 8t for large-scale training and 8i for real-time reasoning. Today, we’re expanding our TPU portfolio with two additional systems: TPU 8tx for secure and confidential inference and TPU 8i Max for addressing massive context windows. The pace of AI adoption combined with the unique requirements of agentic AI necessitates more specialization for our customers, which is why we have unveiled four new systems in just one year.

57위AI at Meta@AIatMetaX에서 보기

메타가 가장 작고 효율적인 모델 Muse Ember를 구글 클라우드에 제공하며 Muse Wonder와 Muse Spark 1.3도 함께 공개한다고 알린다.

좋아요 177리포스트 12답글 15노출 12.9K북마크 12

AI at Meta@AIatMeta
Today we're also bringing Muse Ember to @GoogleCloud along with Muse Wonder and Muse Spark 1.3.
AI at Meta@AIatMeta
Today we're introducing Muse Ember, our smallest and most efficient model, built for complex reasoning execution and everyday agent work. Muse Ember can carry workflows across apps, call tools in sequence, run tests and make targeted code fixes. With a 1M-token context window and native understanding of text, images, audio and video, it can also work across large document sets and varied inputs.

58위BijanBowen@bijanbowenX에서 보기

작성자는 Muse Ember가 최고 추론 강도에서 2분 안에 작업을 끝냈다며 이 모델이 매우 인상적일 수 있다고 평가한다.

좋아요 167리포스트 8답글 11노출 13.6K북마크 11

59위AI at Meta@AIatMetaX에서 보기

Muse Ember는 Artificial Analysis 지능 지수에서 Claude Haiku 5.5와 비슷한 성능을 약 3분의 1 비용으로 내며, 모든 추론 설정의 GPT-6 Luna보다 높은 점수를 기록했다고 밝혔다.

좋아요 155리포스트 12답글 5노출 16.9K북마크 6

AI at Meta@AIatMeta

@AIatMeta 님에게 보내는 답글

For agents that perform hundreds of small tasks, efficiency matters at every step. Muse Ember is built to deliver more intelligence for the cost. On the @ArtificialAnlys Intelligence Index, it achieves performance comparable to Claude Haiku 5.5 at roughly one-third of the cost per task, while scoring higher than GPT-6 Luna across all of Luna's reasoning settings.

60위DHH@dhhX에서 보기

2008년형 ThinkPad X200에 Omarchy 4.5rc2를 2분 32초 만에 설치해 오래된 노트북이 되살아난 것을 기뻐하는 글이다.

좋아요 420리포스트 4답글 21노출 20K북마크 16

62위Sudipta Biswas@sudiptabiswas93X에서 보기

웹 데이터를 대규모로 추출해 출처가 명시된 검증 데이터를 제공하는 API 이니악을 출시했으며, LLM이 기억에 의존해 데이터를 지어내는 문제를 해결한다고 소개한다.

좋아요 4리포스트 1답글 1노출 3K북마크 5

Sudipta Biswas@sudiptabiswas93
The web was built for humans: one question, one answer, ten links. It was never built for large scale data extraction. So, today we are launching Eniac. Eniac (eniac.floworks.ai) is an API to query the web for extracting verifiable and cited data at the largest possible scale. Sales lists. Market research. Candidate search. Financial research. Journalism. Buyer and supplier discovery. And much more. Three years of building AI agents taught us one thing: agents rarely fail at acting. They fail because the data they act on is incomplete, or made up. Today, extracting large scale web data involves wiring together search APIs and agentic LLM loops- an architecture that fails since LLMs keep pulling data from its training memory (instead of current web) and often fill in inferred (often made up) details where it finds the gap. Eniac changes that. Build with Eniac today: eniac.floworks.ai

63위Mooncast Productions@MooncastOnlineX에서 보기

Muse Ember의 일본 차 정원 벤치 제품을 소개하며, 저렴하면서도 똑똑한 제품이라 업그레이드된 Muse Spark Open 같다는 평가를 전한다.

좋아요 2리포스트 1답글 1노출 345북마크 1

64위Hera Video@heravideoappX에서 보기

Hera로 브랜드 키트와 참고 자료를 활용해 한 캔버스에서 여러 변형을 만들고 정적 및 동적 광고를 내보내는 광고 제작 방법을 소개한다.

좋아요 21리포스트 4답글 5노출 2.6K북마크 10

Hera Video@heravideoapp
You can now make ads like Lovable with Hera. Here's how we did it (full guide in first comment). 1. Start with your brand kit 2. Add your references 3. Create lots of variations on one canvas 4. Export static and motion ads Try Hera & create all the ad creatives you need.

65위Diana@sdianahuX에서 보기

sdianahu는 소프트웨어 설계는 회로 설계처럼 자동화될 수 있지만, 설계 가능한 범위가 실제 제조 가능한 범위보다 넓어 물리적 제조가 병목으로 남는다고 주장한다.

좋아요 17리포스트 3답글 5노출 3.7K북마크 6

66위Mo Gharabat 🦇@ikbenbashaX에서 보기

“클로드 같은 제품 전용 AI”가 나왔으니 2027년까지 에이전트형 제품이 아니면 이미 늦었다는 주장이다.

좋아요 17리포스트 3답글 2노출 3.9K북마크 6

67위Google Cloud Tech@GoogleCloudTechX에서 보기

Meta의 Muse Wonder, Muse Spark 1.3, Muse Ember 모델이 Google Cloud 모델 가든에서 공개되어 서버리스 방식으로 배포할 수 있게 되었다.

좋아요 10리포스트 2답글 0노출 2.3K북마크 3

68위Y Combinator@ycombinatorX에서 보기

YC는 정부가 AI 개발을 늦춰야 하는지를 주제로 11월 16일 월요일 첫 토론을 열며, Jacob Coxon이 찬성, Robin Hanson이 반대 입장을 맡는다고 알린다.

좋아요 35리포스트 3답글 9노출 10.6K북마크 10

69위Visual Studio Code@codeX에서 보기

VS Code에서 PostgreSQL 성능 대시보드와 Azure Advisor 권장 사항으로 데이터베이스 성능 문제를 진단하는 방법을 소개하는 영상 홍보 글이다.

좋아요 65리포스트 5답글 4노출 10.2K북마크 18

70위Vivian Midha Shen@vivianmshenX에서 보기

YC가 AI 개발 속도를 정부가 늦춰야 하는지를 주제로 로빈 핸슨과 제이콥 콕슨의 첫 공개 토론을 연다고 알리고 초대 신청을 안내한다.

좋아요 16리포스트 2답글 0노출 2.9K북마크 4

Vivian Midha Shen@vivianmshen
We're excited to host Robin Hanson and Jacob Coxon for our inaugural YC Debate at @ycombinator Safe AI development is one of the defining topics of our era. Request an invite at the link below @robinhanson @hilbertspaess
Y Combinator@ycombinator
Should governments intervene to slow down AI development? Join our inaugural YC Debate on Monday, 11/16 at 6pm: Jacob Coxon (ex-Anthropic, OpenAI) will argue the affirmative. Robin Hanson (GMU) will argue the negative. events.ycombinator.com/yc-debates-gov…

71위Garyk Brixi@garykbrixiX에서 보기

garykbrixi가 유전체 언어 모델 연구와 생물학적 추론의 방향, 그리고 DNA부터 시작하는 이유에 대해 대담한 내용을 이야기했다.

좋아요 41리포스트 3답글 5노출 3.8K북마크 10

Garyk Brixi@garykbrixi
It was great talking about our research on genome language models and where they are headed What are the paths to biological reasoning, and why start with DNA? Thanks @FrancoisChauba1 @ycombinator for hosting, and the community for great chats after
Y Combinator@ycombinator
This week’s Paper Club is all about AI for biology. Sequencing costs have collapsed, biology is generating more data than researchers can make sense of, and the models are beginning to catch up. The opportunity now is to move from reading biological data to modeling and designing living systems. We brought together researchers and builders working across cancer genomics, virtual cells, genome language models, and biological design to show what that shift looks like in practice. We start with a single patient, asking how AI can build a research program around their cancer, then zoom out to whole cells and genomes. Along the way, we look inside a language model trained on DNA and explore how in-context learning emerged even though the model was never explicitly trained to do it. 00:00 - Is the Turing test solved? A live experiment 03:22 - Why now is the best time for AI Bio 08:26 - Alfredo Gonzalez: cancer as a data problem 29:50 - Yusuf Roohani: virtual cells and autonomous research 51:47 - Garyk Brixi: what a model learns from DNA 1:09:57 - Divya Nori: few-shot prompting for biological design

72위Simon Willison@simonwX에서 보기

방 안 상자 더미에서 들리는 간헐적인 삐 소리의 근원을 상자를 하나씩 열어 찾을지, 노트북과 태블릿, 휴대폰으로 소리를 삼각측량하는 소프트웨어를 만들지 선택을 묻는 글이다.

좋아요 23리포스트 1답글 14노출 2.1K북마크 5

73위Pi Dash@pidashaiX에서 보기

Pi Dash는 Muse Ember 기반의 에이전트 10개가 협업해 약 17만 8천 줄의 코드를 하루 만에 SeaORM으로 다시 작성했다며 성과를 자랑한다.

좋아요 16리포스트 4답글 2노출 5K북마크 3

Pi Dash@pidashai
Pi Dash + Muse Ember == Coding beast! 🔥 Pi Dash is honored to get early access to Muse Ember over the past weekend! 🤖 10 Muse Code agents with Muse Ember working together under Pi Dash's Coordination to evolve itself:⚡ Raw SQL → SeaORM, ~178,000 lines of production code rewrite in a day Watch it in action 👇 pidash.airepublic.com

74위Meta for Developers@MetaforDevsX에서 보기

Meta 개발자 계정이 가장 작고 효율적인 모델인 Muse Ember를 Muse Code와 Meta Model API에서 사용할 수 있다고 알리고 개발 블로그 링크를 공유했다.

좋아요 62리포스트 11답글 3노출 8.2K북마크 11

Meta for Developers@MetaforDevs
Ready to get started with Muse Ember? Read the dev blog for details on the model, how to access, and recipes:
dev.meta.aiStart building with Muse Ember, our smallest, most efficient model
Meta for Developers@MetaforDevs
Muse Ember, our smallest, most efficient model, is now available in Muse Code and on Meta Model API. Built for agentic execution at scale, Muse Ember excels at coding subtasks, tool calls and document analysis, with support for audio and video inputs alongside text and images for your multimodal workflows. Pair it with Muse Wonder or Muse Spark to combine efficient execution with deeper planning, reasoning and open-ended work. We’re excited to continue building out the Muse series of models to give developers more choice over how they tackle their toughest challenges. s Start building: bit.ly/4z7XqLS

75위BrendanEich@BrendanEichX에서 보기

이츠하크 퍼만이 연주한 차이콥스키 바이올린 협주곡 영상을 공유하며, 어려운 곡을 땀 한 방울 없이 연주하는 그를 한 번뿐인 전설적 바이올리니스트라고 극찬하는 평을 인용한다.

좋아요 18리포스트 2답글 2노출 1.8K북마크 3

76위Visual Studio Code@codeX에서 보기

VS Code와 GitHub Copilot의 주요 업데이트를 제품팀 라이브 데모로 소개하는 VS Code Live 행사가 11월 12일 목요일 오전 8시(PST)에 열린다는 안내이다.

좋아요 12리포스트 2답글 1노출 5.1K북마크 2

77위Visual Studio Code@codeX에서 보기

Claude Fable 5.2가 GitHub Copilot에서 정식 출시되었으며, 대규모 코드베이스의 복잡한 다단계 코딩 작업에서 Fable 5.1보다 적은 단계와 도구 호출로 더 많은 작업을 해결했다고 알린다.

좋아요 45리포스트 2답글 9노출 5K북마크 7

78위AI at Meta@AIatMetaX에서 보기

Muse Ember가 Meta Model API에서 백만 입력 토큰당 0.10달러, 출력 토큰당 0.50달러에 제공되며 자세한 내용은 연구 블로그에서 확인할 수 있다고 안내한다.

좋아요 121리포스트 8답글 4노출 12.8K북마크 15

79위Basic Attention Token (BAT)@AttentionTokenX에서 보기

Brave와 Solana가 런던 Breakpoint 컨퍼런스에 한정판 Brave와 BAT 머천다이즈를 선보이며, Brave Wallet으로 결제하면 10% 할인된다고 알린다.

좋아요 22리포스트 5답글 1노출 2.2K북마크 2

Basic Attention Token (BAT)@AttentionToken
We're teaming up with @solanaspaces and to bring limited-edition @brave + $BAT merch to the @solana Breakpoint conference in London this week. 👀 Use your @brave Wallet to pay and get a 10% discount at checkout. 🤑🔥
Solana Spaces@solanaspaces
no extension to install, so no fake extension to get phished by 🦁 Brave Wallet is built into the @brave browser, and it gets you 10% off at our checkout Internet Capital Supermarket | 15–17 NOV | Olympia

80위BrendanEich@BrendanEichX에서 보기

Brendan Eich는 영화 NOPE의 시작에 인용된 나훔 3:6 구절을 언급하며 트럼프가 누구도 서둘러 비난하지 않는 점을 꼬집고, 오픈 소스와 오픈 웨이트가 독점에 맞서 이긴다고 주장한다.

좋아요 11리포스트 1답글 1노출 2K북마크 1

BrendanEich@BrendanEich
Nahum 3:6 (quoted at start of the excellent film NOPE): "I will cast abominable filth upon you, make you vile, and make you a spectacle." Good to see Trump not rush to condemn anyone. Open tends to win, even in face of duo/monopoly, especially with software. Open weights FTW!
Aprajita Nafs Nefes 🦋 Ancient Believer@aprajitanefes
🎭|The mainstream narrative wants you to believe the Nvidia scandal is about an affair and a spy. That is the theater. The dark truth is that Washington is using this personal drama as a smokescreen to hide a catastrophic strategic defeat in the very chip war it started. The speculation isn't the story; it's the distraction from the story: America's economic war on China has failed, and the architects of that failure need a scapegoat. The Distraction: Why Washington Needs a Spectacle The timing of the leaked memo is the tell. It was drafted in the spring and circulated just as Jensen Huang was successfully lobbying the Trump administration to reopen the Chinese market for Nvidia's H200 chips. This wasn't a national security emergency; it was a policy dispute. Huang was winning the argument, and the national security establishment needed to neutralize him. The "affair with a spy" narrative does exactly what a cheap magic trick is designed to do: it keeps your eyes on the scandal while the real larceny happens elsewhere. The larceny here is the collapse of America's "Pax Silica" strategy. Washington's plan to keep China dependent on US technology through export controls has backfired spectacularly. China has accelerated its push for self-sufficiency, with companies like SMIC achieving volume production of a 5nm-class node and Huawei leading a domestic push for chipmaking equipment. The US strategy, as one analyst put it, has been a "gift" that accelerated China's independence. 💰 The Real Target: A 25% Shakedown, Not a Security Strategy The truth behind the H200 chip deal is far more cynical than a security concern. It was a revenue grab. The Trump administration approved the sale of H200 chips to China on the condition that the US government receive 25% of the revenue from every sale. This "Trump Surcharge" was designed to extract money from Nvidia's foreign sales to fund the US Treasury. But here is the part that exposes the true desperation: China refused to buy the chips. As of May 2026, not a single unit of the H200 was shipped to Chinese buyers, despite the licenses being approved. Beijing directed its companies to invest in domestic alternatives like Huawei, and Nvidia's market share in China effectively collapsed to zero. The shakedown failed because the target walked away. The US was left holding 750,000 high-end chips with no Chinese buyer. The "spy" scandal is designed to bury the humiliation of this failed extortion attempt. 🌑 The Darker Truth: Washington Is Hiding Its Own Decline While the media obsesses over Huang's private life, the structural foundation of American power is cracking. The chip war was the centerpiece of Washington's strategy to contain China's rise. Its failure is a geopolitical catastrophe that the establishment cannot admit. ◾The Military-Industrial Complex Needs an Enemy: The "China threat" narrative justifies the massive budgets of the Pentagon and the surveillance state. A CEO who proves the threat is overblown is a threat to the entire ecosystem. The scandal is a weapon to discredit him and protect the narrative. ◾The "Pax Silica" Illusion: The US launched the "Pax Silica" initiative to build a coalition of countries to de-Sinicize supply chains. But forcing supply chains out of optimal economic zones is choosing inflation over integration. The chip war is hurting the US economy, driving up costs, and cannibalizing other sectors. ◾ The Trump Pattern: Trump, facing a stalled agenda in the Middle East and a losing trade war, is resorting to diversionary tactics. When he cannot deliver a victory, he creates a spectacle. The Nvidia scandal is a classic "wag the dog" operation, a foreign policy crisis manufactured to distract from domestic failure and electoral weakness. The Bottom Line The "Chinese spy" is a ghost. The real target is Jensen Huang, the CEO who proved Washington's chip war was a failure. The real story is that the United States is losing the technology war, its strategy of economic containment has backfired, and its political leadership is using a personal scandal to hide the collapse of its own geopolitical strategy. The theater is loud, but the silence from Washington about its strategic defeat is deafening. The bill for this failure is being paid by the American economy, and the receipt is a chip scandal designed to keep you from reading the balance sheet.

81위BrendanEich@BrendanEichX에서 보기

스타트렉의 체코프와 장뤼크 배우들의 별세를 애도하며 알츠하이머 연구와 간병인 지원 단체에 기부를 권하는 글이다.

좋아요 63리포스트 3답글 5노출 4.4K북마크 3

83위BrendanEich@BrendanEichX에서 보기

Brendan Eich는 로비 회사 관련 논란에서 대통령을 포함해 모두에게 책임이 있으며, 1달러 1표는 올리가르히의 지배를 뜻한다고 Lex Fridman에게 말했다고 밝힌다.

좋아요 19리포스트 1답글 3노출 2.1K북마크 0

BrendanEich@BrendanEich
Hang 'em all -- I see some trying to blame others and excuse POTUS as if buck doesn't stop there, but they all stink. One dollar one vote means rule by Oligarchs, as I told Lex Fridman.
Josh Walkos@JoshWalkos
Thread 🧵 Brian Ballard has issued a statement denying allegations that his lobbying firm improperly influenced the Trump administration on glyphosate and flavored vapes. Unfortunately for Ballard, his own federal lobbying disclosures paint a rather inconvenient picture.

84위Uncle Bob Martin@unclebobmartinX에서 보기

오빗 슬링어 레벨 30에서 30,139점으로 최고 기록을 세웠다는 게임 홍보 글이다.

좋아요 12리포스트 0답글 2노출 4.5K북마크 0

86위BrendanEich@BrendanEichX에서 보기

칼뱅주의가 전적 타락을 믿는다는 이유로 개종했지만 상상만큼 재미있지 않았다는 농담으로, 앞선 칼뱅주의 논쟁에 답한다.

좋아요 3리포스트 0답글 0노출 440북마크 1

87위Google for Developers@googledevsX에서 보기

Antigravity 음성 모드에 대한 공식 문서 링크를 공유했다.

좋아요 13리포스트 1답글 1노출 2.8K북마크 4

88위Ankit Jain@ankitxgX에서 보기

ankitxg는 QCon SF의 DevEx 트랙을 진행하며, Kent Beck의 'Features With Futures' 세션을 소개하고 Kent가 인간 전용 워크플로를 재현하거나 기존 지식을 버리는 데 노력이 낭비되고 있다고 말했다고 전했다.

좋아요 4리포스트 1답글 0노출 1.5K북마크 1

89위Visual Studio Code@codeX에서 보기

Claude Fable 5.2는 기본적으로 데이터 보관이 필요하며, 제로 데이터 보관 접근은 일부 자격을 갖춘 엔터프라이즈 고객에게만 제공된다.

좋아요 4리포스트 0답글 0노출 2.1K북마크 1

90위BrendanEich@BrendanEichX에서 보기

Brendan Eich는 트위터가 Apache Lucene을 포크해 성능을 개선했으나 팀이 이를 떠나는 중이라 버그가 예상되며, 빠른 수정과 AI 지원 회귀 테스트 확장이 필요하다고 말한다.

좋아요 6리포스트 3답글 1노출 1.1K북마크 1

BrendanEich@BrendanEich

@DefNotJC603 님에게 보내는 답글

Twitter (when it was called that) forked Apache Lucene and made perf fixes. It mostly worked for me until 2nd half this year, but the team is moving off it, so I expect some bugs (and require both fast fixes and a growing, AI-assisted regression test suite).

91위BrendanEich@BrendanEichX에서 보기

국회의사당과 시멘트 자루를 언급하는 노래 가사를 인용하며 불멸의 가사라고 칭찬한다.

좋아요 6리포스트 2답글 1노출 2.3K북마크 1

92위GitHub@githubX에서 보기

Claude Fable 5.2는 안전 분류기 운영을 위해 기본 데이터 보관이 필요하고, 제로 데이터 보관은 자격 있는 일부 엔터프라이즈 고객만 이용할 수 있다는 안내이다.

좋아요 6리포스트 1답글 0노출 3.6K북마크 1

93위Vercel@vercelX에서 보기

Vercel이 언어 모델 요청의 절반 이상이 60일 이하 된 모델로 향하며, 8월의 26%에서 늘어났다고 밝혔다.

좋아요 2리포스트 0답글 3노출 2.2K북마크 0

94위BrendanEich@BrendanEichX에서 보기

힘과 마음과 머리를 다해 일어서라는 내용의 짧은 시 구절을 옮긴 글이다.

좋아요 8리포스트 2답글 0노출 0북마크 0

95위BrendanEich@BrendanEichX에서 보기

Kevin McKernan의 글에 답하며, Cotton Mather와 그의 아버지 Increase가 청교도였다는 점을 들어 그런 왜곡된 철학이 오늘날 다시 나타난 것이 이상하다고 말한다.

좋아요 6리포스트 0답글 1노출 702북마크 0

96위Naymur Rahman@naymur_devX에서 보기

가벼운 모델이 Command Code의 /design 기능으로 3D 장면을 한 번에 만들어냈고, Muse Ember가 GPT-6-Luna보다 한 번의 프롬프트로 더 나은 결과를 냈다고 밝힌다.

좋아요 2리포스트 1답글 0노출 350북마크 0

Naymur Rahman@naymur_dev
Muse Ember surprised me with the output 👀 A lightweight model created this 3D scene using the /design feature from Command Code. Muse Ember nailed it in one shot, while GPT-6-Luna needed two prompts to get it perfect.
Command Code@CommandCodeAI
Muse Ember is now live in Command Code on all plans. · $10/mo GOAT gets $20 usage · 1M token context window · GPT-6 Luna & Haiku 5.5 level price Try it out!

97위BrendanEich@BrendanEichX에서 보기

패트릭 스튜어트가 젊은 시절 카를라 역으로 출연한 스파이 드라마 장면을 공유하며 인물들의 대사와 연기에 대한 감상을 전한다.

좋아요 1리포스트 0답글 1노출 709북마크 0

98위Vercel@vercelX에서 보기

Vercel은 결정 모델이 요청의 26%를 차지하고 Anthropic이 신규 팀을 2대1로 더 많이 확보하고 있다는 내용의 전체 보고서를 소개한다.

좋아요 1리포스트 0답글 0노출 2.1K북마크 2

99위BrendanEich@BrendanEichX에서 보기

패트릭 스튜어트가 우주 흡혈귀 소녀를 연기한 1985년 영화 라이프포스를 기묘하고 과장된 SF 호러의 즐거움으로 소개한다.

좋아요 0리포스트 0답글 0노출 893북마크 0

BrendanEich@BrendanEich

@BrendanEich 님에게 보내는 답글

Also can't forget Patrick channeling space vampire girl
youtube.comLifeforce (1985) - Her True Form Scene (6/10) | Movieclips
BrendanEich@BrendanEich
A wild film, over the top male-hysteria score by Henry Mancini. A guilty pleasure; truly strange, one of a kind SF/Horror movie. Harlan Ellison roasted Star Trek the motion sickness (late 1979) cosmic space f*** ending, but LIFEFORCE (1985) did it better. Dan O'Bannon screenplay!