Latest signals
Everything, including what we didn't write about. Ordered by the field's reaction, not our score.
- 17,258Cursor is now part of @SpaceX.
- 540Every Fucking Website (2020)
- 10,835클로드 코드, 한도 걸려도 자동으로 이어서 작업한다
- 8,694We promised open weights for Qwen3.8. Now, time to meet them! 🎉
- 435Gemini 3.7 Flash
- 8,316🧩 DeepSeek Harness v0.1 is now available in Developer Preview!
- 410Count Binface receives over a quarter of votes in Clacton by-election
- 422DeepSeek Harness developer preview
- 385GLM-5.3: Frontier coding with emergent cyber capabilities
- 394Codex in ChatGPT desktop app for Linux is now in preview
- 5,413클로드 인 크롬, 대화 세션 기기 간 연동 시작
- 344Delta
- 304Why does Opus 5 feel worse to work with?
- 291Qwen3.8-27B
- 4,493Figma turns 14 today https://t.co/tHb1VKZsKD
- 285Spaghettifying DRAM
- 270Choose Boring Technology (2015)
- 279Gloomberb
- 274DeepSeek V4 Pro 0813
- 251Firefox is now the last major browser that still supports uBlock Origin
- 256Hello, me. It's been a while
- 258Accelerating GPT-5.6 Sol Ultrafast
- 261Mistral OCR 4.1
- 257Nine PBS sues Iron Mountain over blocked access to archival data
- 178deepseek-ai/DeepSeek-V4-Pro-0813 · text-generation
- 241Google is making private AI practical with homomorphic encryption
- 3,388Garry Tan says turning down Palantir for a Microsoft promotion was a $2–4 billion mistake:
- 237Principia Mathematica is modern and insightful
- 240Grok 4.6 scores 61 on the Artificial Analysis Intelligence Index
- 225Going Dark, and the era of law enforcement hacking
- 224Seven books I keep close because I love them
- 2,948"Cursor and SpaceXAI are a deep cultural fit with one another. They both work with a pace and intensity that borders on the absurd. And they’re both defined by iteration."
- 217In Australia, a home battery boom has helped cut wholesale power prices
- 223Understanding is the new bottleneck
- 229Ordinary abundance
- 222Deutsche Bank becomes first foreign yuan clearing bank in Europe
- 217ChatGPT Desktop (Codex Desktop) for Linux
- 211The other Sean Byrne doesn't exist
- 2,867Cursor, 4 years ago https://t.co/VEUuQKSxgj https://t.co/qvWb1mBghR
- 205Maximizing the value of your Claude Code sessions
- 208France's top court blocks social media ban for under-15s
- 215AI by Hand
- 207Introducing Toast 1
- 210I turned my RSS feeds into an e-ink newspaper to stop reading on my phone
- 203US conducted mass spying campaign against leftwing and anti-ICE protesters
- 203NP-overrated
- 208Single log line is 49KB+ (ext4) / 110KB+ (btrfs) of systemd-journald disk writes
- 213Where did the old web go? We followed 657,607 links to find out
- 212Donkey.bas is 45 Years Old – 131 line of Glory
- 212I requested a copy of my data from McDonald’s loyalty program
- 215HTML over WebSockets: real-time SPAs with barely any JavaScript
- 201Magnitude 7.7 Earthquake – 68 km NNW of Ende, Indonesia
- 201RustDesk now supports true unattended remote access on Wayland
- 201Dear people who work at the airport
- 202DeepSeek peak/off-peak pricing update
- 202Bluesky Protocol Services
- 201How Compaction Works in Pi
- 2,499The top 10% of enterprises use plugins twice as often and skills six times as often as typical firms.
- 202Choosing an AI model: one prompt, 11 models, different results
- 2,491My lawyer is obligated to in all but the most extreme circumstances; he will even defend me if he knows I’m guilty.
- 2,285Qwen3.8-27B can now be run locally! ✨
- 2,167코덱스·챗GPT, '컴퓨터 히스토리'로 작업 이어간다
- 2,238The weights for the X algorithm have been publicly released.
- 2,150Less than 2 hours to say hi👋. It's almost time!
- 1,815오픈AI, GPT-5.6 Sol '초고속 모드' 예고…최대 14배 빠르게
- 55DreamX-Phi 1.0: Action-Conditioned Video World Model for Robotic Manipulation
- 1,715Travis Kalanick says entrepreneurs need pissed-off energy:
- 1,619We’re excited to welcome the Firetiger team to Cursor!
- 1,661Ultrafast mode for GPT-5.6 Sol is now in limited preview, powered by Cerebras.
- 1,314Auto mode is rolling out today as the default permission mode in Claude Code for Pro, Max, and Team.
- 1,412Travis Kalanick says new media is why he's back after eight years of silence:
- 1,249앤스로픽, 두 번째 리스크 리포트 공개
- 1,153Cloud agents now start 3x faster so you can hand them ambitious, long-running tasks to execute from start to finish.
- 1,022Hermes has a new slash command: /loop.
- 1,011Yes, we are back👑, with 206 tok/s on a single RTX 5090!
- 1,046제미나이 3.7 플래시, 지능 4점 올리고 시간은 줄였다
- 1,024Garry Tan says we'll see a wave of exceptional older AI-native founders:
- 1,023Hermes profiles are now portable and shareable:
- 855"Anysphere—which most people just called 'Cursor'—became the fastest software company in history to reach $100 million in annual recurring revenue."
- 853Previewing Ultrafast mode for @OpenAI's GPT 5.6 Sol, powered by Cerebras.
- 866Qwen3.8 can now be run locally! 🔥
- 847You know whats crazy? Grok 4.6 is not only cheaper than Opus 5 and 5.6 Sol. Its even cheaper than Sonnet 5. While at the same time, at least on par with the sota models.
- 726https://t.co/fbjLdFGKez
- 773Qwen-3.8 27b released! Significant jump!
- 758Promises were made. https://t.co/Ny6I7eqQNe
- 789Garry Tan says AI enables a single markdown file to act like an employee:
- 794Are you hiring PMs, engineers, or designers?
- 759구글 공동창업자 브린, 딥마인드에 '자기개선 AI' 자원 배분 압박
- 734As yet unverified DeepSeek V4 Pro benchmarks from WeChat. Seismic if accurate. https://t.co/CG5s8Q1PJJ https://t.co/fEOONbaFEO
- 642From now on, if you use Claude just to translate text into another language, it will be considered AI-generated.
- 648We’ve written an FAQ to answer some of the questions we've received about watermarking.
- 613https://t.co/wR6cII8uW6
- 594딥시크 V4 프로 GA 벤치마크 유출, 오픈소스 최상위권 근접
- 557Travis Kalanick: "A lot of folks think I'm back. I've been working my ass off the whole time. I just haven't been talking about it."
- 538Qwen3.8 27B brings a new state-of-the-art dense model for local AI development.
- 561Introducing Prime Flash MoE - a set of Blackwell-optimized CUDA kernels for MoE inference.
- 554Gemini 3.7 Flash is here.
- 552Garry Tan on YC, First Principles Thinking, and Progress in the Age of AI
- 492Grok 4.6 is now available in Perplexity and Perplexity Computer.
- 509LongCat-2.0 from @Meituan_LongCat is free in Nous Portal for one week.
- 502We’re launching Optima. Now anyone can create a custom benchmark for their use case, leveraging Artificial Analysis’ leading research and platform
- 459AI 영상으로 만든 '더 오피스' 패러디, 로켓리그 밈 화제
- 482Cursor once self-described as "compute-starved."
- 453One GPU, 1M context, Day-0 ready. Big props to the vLLM team for the seamless integration!👍
- 452챗GPT, 컴퓨터 사용 기록 기억하는 '컴퓨터 히스토리' 추가
- 428As has already been expected, Anthropic internally uses a model that is significantly better than Mythos 5, but they have no plans to release it. https://t.co/fB5JcB8cz3
- 410The deal was completed today. https://t.co/r9xzTqqFar
- 428A few days changed the entire game:
- 404Robots can already fold laundry, make espresso, clean kitchens, and assemble things. The harder problem is getting them to do those tasks reliably, for long periods of time, without a human babysitting them.
- 394Qwen 3.8 27b was the big leap I had hoped for and is a complete success.
- 371NVIDIA just raised prices again, it's $15k for the cards :( https://t.co/BYrOpOTLAR
- 368„A new timeline view gives you the ability to look back on your work and build skills from your frequent tasks.“
- 392Gemini 3.7 Flash official released. Surprisingly solid jump over 3.6!
- 388We're thrilled to invest in Vals.
- 387DeepSeek has open-sourced DeepSeek Harness, their framework for building and running AI agents. Digging into it now. https://t.co/HPPa2zH8X1 https://t.co/Fa06UVakkV
- 383Grok 4.6 made large gains on AA-Briefcase, our agentic knowledge work benchmark and cost substantially less than other leading models
- 348"HERMES-4000"
- 354WHAT: Zai just launched GLM-5.3, and its biggest leap may be in cybersecurity.
- 338Qwen Live now, an appetizer 👀 https://t.co/zML048IhM5
- 360머스크 "Grok 4.7, 현존 모든 모델 능가할 것" 예고
- 320"We are in a market that's had an iPod moment, and it's gonna have an iPhone moment, and another iPhone moment."
- 32627B on 17GB RAM. Are you ready to create something incredible? 😎
- 324Qwen 3.8-Max is live on Modal, with the full 1M context window and a custom DFlash speculator under the hood.
- 308OpenAI started sending waitlist invitations for Premium $100/month seats on ChatGPT Business.
- 309Congrats to @AndersForslund1 and @heartaerospace (W19) on the first flight of X1!
- 315Shots fired at Anthropic https://t.co/M8xRM5MARA https://t.co/J66I2no3Vj
- 321힉스필드, 20분 AI드라마 '오네이릭' 프롬프트 전면 공개
- 301Another reset is here; Codex has surpassed the 15 million user mark. Congrats to OpenAI and congrats to us. https://t.co/gY8AnEfTHP
- 318업스테이지 Solar Pro 4, 지능지수 14에서 42로
- 269ANTHROPIC 🔥: Claude desktop will get a dedicated Browser inside that app that will always be accessible in Cowork sessions! A new Super App is rising.
- 296Qwen3.8-27B is available on @ollama ！Any harness you run, the model always delivers. Let's code together, and show us what you build.😎 https://t.co/zqZahbEVCI
- 275Sharing confidential information with AIs is a norm these days! We can do better. Our latest encoder models find and strip 40 types of PII across 16 languages, all in one simultaneous pass for a secure information sharing with AIs.
- 277Many late night developments. They say in the blog it's the same base model as GLM-5.2, and all the gains came from post-training. https://t.co/F7EP292ps8 https://t.co/ZsBSOFZRx8
- 273In a suprise move Apple has trained their own custom model specifically for use in China, the training was done in partnership with Alibaba. This would be first foreign company approved by Beijing to offer a proprietary AI model in China. https://t.co/XzC95WkqYY
- 271The team at @mixedbreadai used Prime Intellect to train Toast 1.
- 278Matic just got over-the-air updated to be able to listen to your commands 🙌 https://t.co/Ii1FFNYYYm https://t.co/fBUcDwpvu2
- 272You can now fine-tune Meta Muse Glimmer 30B for free! 🔥
- 20AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design
- 2962-bit NVIDIA Nemotron 3.5 Lightning ran tool calls nonstop for 10 minutes on just 22GB of VRAM. 🤯
- 253Qwen3.8-27B is now part of your everyday life — from smartphones to vehicles. ⚡️🚀Appreciate your work! @MediaTek https://t.co/tbs0Yrc5mX
- 259Share your results from Qwen3.8-27B! 👇
- 256To those who doubted me, this is from this morning's report released by Anthropic. Section 1.4 'Notes on coverage of unreleased models' with frontier capabilities which have not been publicly released. https://t.co/ZjeDFHz1ht https://t.co/Sv8gfrKNdA
- 259🎁A gift for developers: Qwen3.8-27B, running locally on AMD from day zero.
- 258Anthropic now has an "auto-complete checkbox" for when a task isn't finished but the allocated resources have been used up.
- 267Studio 2.0 is live.
- 261백악관, AI 감독 대상에 오픈모델도 포함 검토
- 267Elon on Grok 4.7, and a new timeframe. https://t.co/ZvSvD3ubGg https://t.co/tCWymj48BH
- 232OpenAI is on track to clear $40 billion in annualized revenue, roughly double its run rate at the end of 2025, Bloomberg reports, citing people familiar with the figures.
- 225Visiting the dentist is such an interesting way to track AI adoption over time. 6 months ago, zero AI. Today, AI is analyzing my x-rays and transcribing the conversation. Wonder what 6 months from now will look like. https://t.co/PxEJoYObkR
- 232Gemini 3.7 Flash is now available in Devin.