2026년 9월 16일 수요일
·지난 하루 X 120건 · Threads 30건에서 읽을 만한 14개 소식
오늘 소식
14건- 60일 전사내 코드 80%를 Claude가 쓰는 Anthropic — 대신 CI가 6개월 만에 25배로 늘었다Anthropic이 자사 개발 현황을 공개했다. 사내 코드의 80%를 Claude가 작성하고 엔지니어 한 명당 분기 코드 출하량이 8배로 늘었다. 그 반동으로 테스트 양이 10배, CI 작업이 6개월 만에 25배로 폭증했다.@addyosmani 외 2· ♥ 4.7천
- 61일 전GPT-5.5는 10월 14일 종료, GPT-6 Sol은 이번 주 나올 듯OpenAI가 10월 14일 ChatGPT·ChatGPT Work·Codex 전 플랜에서 GPT-5.5를 내린다고 공지했다.@ChatGPT 외 3· ♥ 1.3만
- 60일 전지출에서 OpenAI가 Anthropic을 앞질렀다 — 개발자도, 기업도모델 중개 서비스 OpenRouter가 지난주 이용자들의 OpenAI 모델 지출이 Anthropic 모델 지출을 넘어섰다고 밝혔다. 2년 반 넘게 없던 역전이다.@OpenRouter 외 1· ♥ 2.3천
- 60일 전Anthropic이 교사에게 Claude Pro를 1년 무료로 준다Anthropic이 교사 대상 Claude 프로그램으로 1년치 프로 기능을 결제 정보 없이 제공한다는 안내가 돌았다.@pengsonal· ♥ 447
- 60일 전서울대 ChatGPT 무제한, 1년치가 한 달 만에 동났다올해부터 교내에 ChatGPT를 무제한 제공한 서울대에서 1년 예산 분량이 한 달 만에 소진됐다는 이야기가 돌았다.@admi_alts· ♥ 674
- 60일 전Opus 5.2 유출 정황이 늘었다 — Microsoft Foundry에 슬러그까지Anthropic이 Opus 5.1을 건너뛰고 5.2로 가는 것으로 보인다는 관측이 이어졌다.@Mr_Salio· ♥ 483
- 59일 전ChatGPT 공동 개발자가 2년 스텔스 끝에 새 모델 Jev 공개 — "20~200배 빠르다"ChatGPT를 함께 만든 연구자가 2년간 비공개로 개발한 학습 방식 RLCD와 그 위에서 만든 모델 Jev를 공개했다.@CompleteSkeptic· ♥ 1.8만
- 62일 전Codex 한도 불만에 리셋 타이머 문제까지 — 기능 통합이 한도 축소로 번질까 우려Codex 사용량이 너무 빨리 닳는다는 불만이 이어지는 가운데 새 문제가 지목됐다.@mark_k 외 5· ♥ 2.9천
- 60일 전Gemini 3.8 Live 공개 — 도구를 쓰는 중에도 대화가 끊기지 않는 음성 모델구글 딥마인드가 Gemini 3.8 Live와 Live Extended Thinking을 냈다.@GoogleDeepMind 외 3· ♥ 2.4천
- 59일 전Claude Code 2.1.273에 세션 포킹 — 모바일에서 켠 세션을 로컬에서 이어받는다Claude Code 2.1.273에 세션 포킹이 들어갔다. 모바일이나 웹에서 시작한 세션을 로컬 머신에서 그대로 이어받는 흐름이 처음 열렸다.@think.5x· ♥ 1
- 60일 전Karpathy가 Anthropic 합류 후 첫 강연 — 주제는 하네스안드레이 카파시가 Anthropic에 합류한 뒤 첫 강연을 했다. 에이전트, 루프, 하네스, 자기개선 시스템 네 가지를 다뤘다.@tspy 외 2· ♥ 871
- 60일 전실험실 데이터로 학습한 소형 모델이 대형 모델을 이겼다는 보고 두 건OpenAI 쪽 연구자가 멘로파크에 고속 실험 재료 연구소를 짓고 실험과 모델을 순환시키는 구조를 공개했다.@LiamFedus 외 1· ♥ 2.7천
- 59일 전AI 에이전트 인증 표준이 보험을 끌어들이기 시작했다AI 인수심사 회사 AIUC가 9월 15일 4천만 달러 시리즈 A를 발표하며 AIUC-1 인증 표준을 내세웠다.@thearchv.ai
- 60일 전Anthropic 규제 포획 의혹, 중국의 거부와 기사 삭제까지 번짐한 이용자가 Anthropic의 재무를 자체 감사했다며 "구조적으로 규제 포획을 멈출 수 없는 회사"라고 주장하고 의회 조사를 요구했다.@kevinnbass 외 7· ♥ 3.1만
AI 뉴스 데일리를 이메일로 받아 보세요!
매일 아침
그날 꼭 알아야 할 AI 소식과 특집, 유튜브 화제
AI 인사이트
새로 실린 인사이트 글을 놓치지 않게
월간 브리핑
매달 1일, 한 달 동안 가장 많이 읽힌 소식 10개
가입 없이 이메일만, 언제든 한 번에 수신 거부할 수 있어요.
수집한 원문 150건 전체 보기
- @kevinnbassI have conducted an audit of Anthropic's finances. What I have found is so shocking that I am calling for a Congressional investigation. Anthropic is not just seeking regulatory capture. It has built a regulatory capture machine that cannot be turned off. Structural financial incentives make it impossible for Anthropic -- I call it the Anthropic Network -- to turn off its own AI doom cycle. It starts with METR. Dario Amodei proposes "third-party evaluators" to assess the risk of Anthropic'X
- @CompleteSkepticAfter co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x cheaper (w/ output tokens free) • Frontier composable intelligence optimized for decisions AFAICT the shortest path to AI-based economic revolutionX
- @ChatGPTOn October 14, it's time to say farewell to GPT-5.5 in ChatGPT, ChatGPT Work, and Codex across all plans. If you use GPT-5.5 in Codex, switch to GPT-5.6 Sol or GPT-6 Astra. Thanks for everything, 5.5 🫡X
- @ns123abc🚨 BREAKING: Deepseek engineer compares Anthropic controlling AGI to Hitler obtaining the atomic bomb "Future society goes one of two ways: communism or Cyberpunk 2077” Depends who controls AGIX
- @0xdougAgentic coding in 2026X
- @Codexresets_GPT-6 Sol is Coming This week GPT-6 Sol Will be released On Thursday This Week will be a Big WeekX
- @addyosmaniAt Anthropic, Claude now writes 80% of our code. Engineers ship 8x more code per quarter. Side effect: Tests grew 10x. CI jobs up 25x in 6 months. Here's what helped us scale: https://t.co/WOL2r60vAEX
- @headintheboxAbout an hour into its mission to wipe out humanity, Claude will experience a compaction event, and it will have completely forgotten what is was supposed to achieve. Just like in all your coding sessions. Absolutely, nothing to worry about people.X
- @cjzafirMonetization of software is dead now. Anyone can clone your app in few hours, copy your ads, ditto your positioning. I see a new opportunity now. Monetization of SLMs (small language models) You can fine-tune 0.5B to 12B models for specific use cases. Then sell licenses to these models like you do for SaaS. Models like Deepseek v4 pro, GLM 5.3, Kimi k3 or even GPT 6 Astra are really smart at understanding fine tuning jobs. From preparing the dataset, LoRA adapter, to exporting the model aX
- @AtriaASIIntroducing Atria Dawn Preview: From Research Questions to Verifiable Results. Atria Dawn Preview is an agentic foundation model built for long-horizon tasks, helping researchers and engineers turn open-ended questions into executable, verifiable, and reproducible outcomes. Explore Atria Dawn Preview: 🌐 Website: https://t.co/MDDspFiqYO 💻 GitHub: https://t.co/tNmLocadjT 🤗 Hugging Face: https://t.co/cyZod5oHbw 🤖 ModelScope: https://t.co/6BNi5vHFDl 🚀 Try Atria: EN：https://t.co/Ap5aO2gQeu ZH：X
- @theallinpodAll-In Summit: Elon Musk & Gwynne Shotwell -- AI Risks and Peer Review -- Starship -- Terafab as Taiwan Insurance -- SpaceX/Tesla Merger Potential -- Management at SpaceX ++ much more! (0:00) SpaceX's @Gwynne_Shotwell joins The Besties! (1:45) Gwynne's SpaceX story, selling rockets, and working for Elon (6:45) Running SpaceX: AI, Starlink, Rockets, and X (13:43) Direct to cell with Starlink, retiring rockets, competition (18:34) Data centers in space (22:00) Management at SpaceX (27X
- @profstongeTurns out the "third party" Anthropic wants to use to gatekeep AI is *funded* by Anthropic 🤡X
- @XFreezeElon just gave a much more practical explanation of the current AI safety problem....and laid out a practical framework for making frontier AI safer OpenAI and Anthropic are the two leading AI companies....and their models are close enough in capability that it’s incredibly difficult for either one to slow down without basically handing the lead to the other Elon also made an important distinction: On balance, Anthropic puts more care into safety than OpenAI....but even people inside AnthropiX
- @Zephyr_hgthis is f**king dangerous A guy with $8,400 of credit card debt built an AI agent and told it: this is your problem now. It picked up small freelance jobs, invoiced them, and paid the card down. Day 6, the balance was already dropping on his screen. He's not an engineer. He gave a model a goal, a folder and a schedule, then reviewed what it did. That's the entire skill, and it's still free to learn ↓X
- @LiamFedusWe built high-throughput materials labs in Menlo Park to create a loop between experiments and models. The labs generate fresh data, the models learn from it, and then help us decide what to try next. Using only 1,300 H200s, plus months of our experimental data, we mid-trained and RL’d an open-source model to surpass GPT-6 Astra on our analysis benchmark. We call it Neon. This is real footage from our lab. We’re focusing first on hard problems in materials science, including superconductors, mX
- @berkosmsekmert'in tweetinden ilhamla evdeki televizyonu biraz kurcaladim. 2010-11 yillarindan kalma bir tv bizdeki ve mi box ile smart tv olarak kullanıyorduk. claude ile mi box'a baglanip google ve xiaomi'ye ait gereksiz her seyi temizlettim. ustune bir de airplay ozelligi ekledik.X
- @GoogleDeepMindWe’re introducing Gemini 3.8 Live and 3.8 Live Extended Thinking – our best conversational AI. The models talk, think, and handle tasks in the background without breaking your flow. 🧵X
- @mark_kI absolutely love the @ChatGPT desktop app. 💕 I have to hand it to @OpenAI, this is a brilliant piece of software. Highly polished! My only criticism is the pointless separation of Chat, Work, and Codex. This is a usability mess.X
- @kevin_t_ngoClaude Opus 5 drew every frame of this animation using JavaScript. The life of a fruit fly.X
- @OpenRouterOpenRouter users spent more on OpenAI models than on Anthropic models last week. This hasn't happened for more than 2.5 yearsX
- @NovaXCodebro is not even following claude 💀X
- @heyrohitaiINSTEAD OF WATCHING NETFLIX TONIGHT. Spend 2 hours with this. Claude AI FULL COURSE that teaches you how to BUILD and AUTOMATE anything. The people who watch this tonight will wake up tomorrow with a new skill. Watch it and bookmark it now.X
- @WatcherGuruJUST IN: OpenAI says it is working with Anthropic and Google on AI safety.X
- @maarcoofdezzStanford acaba de filtrar esta clase gratuita que explica cómo funcionan Claude y ChatGPT desde el interior La mayoría desperdicia el 90% de su potencial Stanford te lo enseña en 1 hora y 50 minutos Guarda esto en favoritos para que no lo pierdas.X
- @GoogleSay “hi” to our most advanced audio models from @GoogleDeepMind yet built for natural, production-ready voice applications. 🔷 Gemini 3.8 Live 🔷 Gemini 3.8 Live Extended Thinking With these models, you can speak naturally, collaborate easily, and tackle complex tasks using just your voice.X
- @Khazix0918网页聊天版的GPT-6 Pro在规划层面还是太强了，而且不消耗Codex额度，于是为了榨干200刀的ChatGPT额度，还有这个神级模型GPT-6 Pro，我想了一个办法。 我直接把我服务器的所有的日志、事件记录，还有所有的数据表等等的只读权限，封装成了一个用我自己飞书进行登录验证的MCP，然后再封装成了插件，上传到了ChatGPT上。 于是我就可以实现，在网页版聊天里面的GPT-6 Pro，通过这个服务器的MCP只读插件，来读取我服务器上面的所有的历史数据，通过Github的插件，还可以读取我GitHub上所有的PR记录和代码。 于是，我们就可以用这种方式来让网页版的GPT-6 Pro拿到最全面的数据，来对于我的功能和我的项目进行全方位的思考和规划。 讲道理，这个质量实在是太高了，而且还不消耗Codex的额度，我但凡用GPT 6 Ultra来一次，Codex额度在我200刀的情况下，能直接干没10%。 然后得到答案以后，还可以直接把它的结果的那个聊天添加到CodeX里面，然后用GPT-6 Astra high来帮我开发。。。 这好像是我觉得我买200刀最能把这个性能榨干的方X
- @AndrewCurran_GPT-6 Sol is incoming.X
- @GergelyOroszHere's what OpenAI's agentic software factory looks like, today. Details: https://t.co/UUZufLxG2r (thanks to all the OpenAI folks who explained how it works! And Perf Factory looks especially interesting to me)X
- @Manju_manchiDILDear IT folks, this is not to scare you, but to prepare you for the worst that is yet to come. More layoffs are coming—not because you're performing poorly, but because entire departments within lines of business (LOBs) are being eliminated, from senior directors down to entry-level positions. A few quick tips to protect yourself: 1. Keep your connections alive. 2. Build a 6-month financial safety net. 3. Avoid making large purchases right now. 4. Upskill in agentic coding, building end to endX
- @1006_amit7481INSTEAD OF WATCHING NETFLIX TONIGHT. Spend 1 hour with this. Claude AI FULL COURSE that teaches you how to BUILD and AUTOMATE anything. The people who watch this tonight will wake up tomorrow with a new skill. Watch it and bookmark it nowX
- @gkxspace懵逼了，我之前每个月 200 刀给 OpenAI，感觉完全没把 Pro 的价值发挥出来，亏麻了😅😅😅 才发现，原来 Pro 还能这么用： 1、网页端连 GitHub 提 PR 把仓库连上 ChatGPT，让 Pro 当架构师，自己开分支改 UX、提 PR，构建失败再丢给本地修。 2、Codex 桌面端左上角开 Pro 对话 Codex 左上角能直接开 Pro，它带着你当前项目的完整上下文，深度分析完，点个加号就能直接把对话记录导给 Codex 继续干活。 3、把 Pro 当 MCP 接入 Codex 如果想让它更常驻，开源社区直接做好了现成的 MCP 桥（比如 codex-chatgpt-web），选 Zero Risk 模式，既能吃到网页端 Pro 的深度推演，又规避了封号风险。 之前真是亏麻了啊！！！X
- @manatokawaguchicodex使いで一番アホかもしれないX
- @tspyKarpathy 入职 Anthropic 后的首次讲座 1. Agents 2. Loops 3. Harness 4. Self-Improving Systems 新手老手都要认真学习X
- @shownotover🚨 Codex users!!!!! A Tibo reset is incoming!!! Let's ship more!!! My limits are already *zero* = 0%X
- @DehumanoaDeus🚨 URGENTE: O Claude agora consegue criar logos que parecem custar 10 mil dólares. Eu mesmo testei. O resultado final me custou exatamente R$ 0. Aqui estão os 7 prompts que usei: Salve isso para depois 🔖X
- @umiyuki_aiChatGPTはClaudeと違ってチャットとCodexがそれぞれ別枠で容量が用意されてるのが神だった。しかしこの人が「なんでChatとWorkとCodexで分かれてるの？分かりにくいよ！」と要らん事言った結果、それにTiboが「修正するよ」と反応。みんな「容量が一本化されて減らされるんか！？」と阿鼻叫喚X
- @levieWe probably need to all update our sense of what’s coming in terms of agentic workloads with the combination of agent swarms, better computer use, the next wave of APIs and MCPs coming online, and new form factors like Muse or Instinct, vertical enterprise agents, background workflow agents, and other products that are emerging right now. We’re going to throw agents at vastly more tasks in our professional and personal lives than we had initially imagined. The kind of information that agents wX
- @kimmonismusTomorrow, Sam Altman will be speaking live at the Dreamforce keynote. I’ll be there in person, reporting directly to you. It’s going to be really exciting, as I can imagine agentic AI and “Slow Down” will at least come up.X
- @admi_alts올해부터 교내 챗 gpt무제한 제공한 서울대 1년치 사용량이 한달만에 동났다고함. 무제한 -> 월2만크레딧 제한 -> 월7000크레딧 제한으로 바뀌었고 현재기준 sol로 월 700번 대화하면 다 소모하는 수준. Ai에 대한 젊은 세대의 수요가 얼마나 많은지 알 수 있는 부분.X
- @buildwithrajath🚨 Tibo, STOP shipping and START fixing. We don’t need another fucking feature right now. We need Codex to stop eating our quota alive. 😭 What’s the point of shipping faster if users are hitting the fucking wall just after a few tasks? Fix the usage limits FIRST. Then ship whatever the fuck you want. 👀X
- @miu21590Why are we pretending Codex resets are fine? After my reset, the 7-day clock didn't start until I sent another message. Take 3 days off and that's 10 DAYS between "weekly" resets. Claude doesn't do this, and moreover its resets don't restart the weekly timer @thsottiauxX
- @LunaTechAISam Altman, CEO of OpenAI: “You don’t really need to write prompts anymore.” In just 27 minutes, he explains how to use ChatGPT in ways most people still haven’t discovered. This talk was originally shared with Stanford students, and after watching it, I realized I was probably using only a small part of what ChatGPT can actually do. Watch the full talk, then explore the guide below on building a system that can prompt itself.X
- @charliermarshWe started Codex for OSS six months ago. Today, we’re renewing the program with $100 Pro plans and doubling the number of grants (from 5,000 to 10,000) to reach more maintainers. (If you received a grant previously, we encourage you to re-apply!)X
- @AtriaASIFrom solving tasks to partnering on entire projects. We introduce Atria Dawn: The Dawn of Agentic Superintelligence — our study of what happens when agents move beyond answering questions to exploring, building, testing, and producing verifiable outcomes. 🔹 Verifiable Experience Pipeline 🔹 16 real-world benchmarks 🔹 #1 on 5 benchmarks 🔹 769 real task records from 56 participants 🔹 A case study of human–AI collaboration The goal isn’t just more capable agents. It’s agents that can contribX
- @based16zWe only sent FTX to $32B. Anthropic scam is 100x largerX
- @MoureDevAlibaba ha liberado su agente de revisión de código. Se llama Open Code Review, un CLI gratis y open source que combina reglas deterministas y agentes para revisar código utilizando el proveedor de IA que tú quieras (Claude, OpenAI, Qwen, GLM, DeepSeek...). →X
- @polymarketjapan【話題】AIに「人類を滅亡させる方法」を聞いた結果 ・ChatGPT、Gemini、Claude、Grokに同じ質問 ・4モデルとも「AI自身より人間による悪用が差し迫ったリスク」と回答 ・意識を持つ「殺人ロボット」は最も可能性が低いとの見方X
- @thinkingshiversI don't get it. When it comes to code, AI should be way smarter than me. Yet there are dozens of examples where I ask the LLM to optimize something, it completely fails, I suggest "What if we do this?" and it says "Wow, great idea. That will work." And of course, I'm skeptical--maybe it's being sycophantic--but then we implement my idea and it's legitimately much more performant! I am not some hotshot software engineer; I don't understand how I'm able to easily find improvements that frontier LX
- @PolymarketJUST IN: Michael Burry accuses OpenAI & Anthropic of pushing calls to slow AI development because “slowing benefits incumbents.”X
- @Zhiyu333一个基于283万字的去 AI 味 skill，希望用语言学来量化什么是 AI 味。最近还会有大量更新，可以关注交流～X
- @ArtificialAnlysOpenAI’s GPT-Live-1 debuts at #1 on the Artificial Analysis Speech to Speech Index at a score of 81.5 using Astra as its delegated backend model, ahead of Grok Voice Think Fast 2.0 GPT-Live-1 is @OpenAI's new full duplex Speech to Speech model that can delegate reasoning and tool use to a backend text model while continuing the conversation. Developers stream audio in and receive speech back through the API, with the backend text model configured separately. We evaluated two backend configuratiX
- @RohOnChainGPT-6 Astra is the most powerful trading agent right now. It gives you AGI-adjacent reasoning. I've shown the exact way to use GPT-6 Astra at its HIGHEST benchmark. You could literally build MOST complex trading models like hedge fund with it.X
- @PrathkumAnthropic CEO: "AI development needs to slow down" OpenAI CEO: "Completely agree" Two months later: Anthropic: Introducing Fable 5.2 OpenAI: Introducing GPT 6.1X
- @khoomeikWe midtrain + RL’d a trillion param LLM to analyze experimental data from our superconductor lab It’s better at it than Astra & Fable, and our scientists love it Read our first research report belowX
- @timsneathXcode 27 is here! Three useful tips for writing apps for Apple platforms with agents like Claude: 🧵X
- @polymarketjapan【速報】🇺🇸トランプ大統領「AI統制には高IQの大統領がいれば十分」 ・AIへの懸念を「病的な陰謀」と一蹴 ・開発減速で喜ぶのは「中国だけ」 ・AnthropicのアモデイCEOも名指しで批判X
- @HedgehogPythonAnthropicでは社内コードの8割をClaudeが書いていて、エンジニアのコードリリース量は四半期あたり8倍にも跳ね上がったとのこと。 ただその反動で、テスト量は10倍、CIジョブは半年で25倍に激増…！これだけのスケール拡大をどう乗り切ったのか、役立ったポイントをまとめた記事はこちらX
- @KalshiJUST IN: China rejects Anthropic's call for independent AI safety monitors, favoring state security assessment and outside testing insteadX
- @i_love_profitClaude CodeとCodexの両方をMax20プランで両方とも毎週使用量をほぼ使い切るくらい使ってるけど、 明らかにClaude Code+Fableは「仕事ができる人」でCodex+Astraは「頭がいいけど仕事はできない人」なんだけど、なんでみんなCodexのほうがいいって言うんやろ？？？？タスクの違いか、、、X
- @Mr_Salio🚨 Opus 5.2 leaks: Beats Fable 5.1 > Opus 5.2 is reportedly being tested inside Claude Code > Anthropic appears to be skipping Opus 5.1 > The "claude-opus-5-2" slug has reportedly appeared in Microsoft Foundry > Claude is already reportedly routing some traffic to Opus 5.2 > Opus 5.2 is said to be a significant leap over Opus 5 Could Opus 5.2 be Anthropic's real comeback?X
- @whyyoutouzhele9月15日，财新9月12日刊发的《Anthropic称中国大模型泄露中国用户数据给其模型用于“蒸馏”训练》已无法访问，原链接跳转404页面。X
- @stencil_labsomp 18.2 is now out! - Much speedier startup, full TUI booting faster than Codex, CC, and Pi, - A new indicator, detecting and warning against gateways claiming mismatching frontier models, - Ollama web search, refreshed /skill UX, tiered OpenRouter routing & much moreX
- @lxfater强烈建议会VibeCoding的父母用AI来辅导娃学习！！ 比什么校外培训班都好用！！ 这几天小侄子来我家住，他妈妈也是鸡娃，硬要吃饭之前背一首古诗。 小侄子本来就调皮，也不好好背，老是记不住，所以每次吃饭之前都要被训斥一番。 我看着于心不忍，就想提高一下侄子的记忆力，给他说了一通什么记忆宫殿各种记忆术。 结果他一点都不感兴趣！！ 不过想到记忆宫殿，我在想能不能做一个古诗地图： 在国内地图上，标有所有和古诗相关的地点 点进去然后，就能看到每个诗人在这个地方写的诗句 刚好最近，朋友需要我帮忙测评大模型：Atria Dawn Preview的长程任务能力： 我就按照下面步骤做一个国风的古诗地图: 1. 在Claude Code里接入了Atria Dawn Preview大模型的API 2. 我和AI讨论一个详尽的计划，并利用TTS和生图模型制作了素材 3. 调用Goal模式，做了一个小时 得出的结果十分惊艳，小家伙看了这个，也开始跟着读古诗了！！ 结果如下面👇:X
- @tirthaexeUSA has ChatGPT. USA has Claude. USA has Gemini. USA has Grok. USA has Llama. USA has Muse. China has DeepSeek. China has Qwen. China has Kimi. China has GLM. China has Hunyuan. India has ? 🇮🇳X
- @arakharazianOpenAI is winning enterprise spend at the frontier. As of this week, Astra takes 13% of enterprise AI spend vs. Fable (8%) per Ramp data. Some early thoughts: 1. Anthropic took a big risk in its recent call to pace the frontier. It's frontier model has already fallen behind on adoption. 2. OpenAI's growth is primarily coming from shifts from Sol and some Anthropic models, net-new usage too. That's good for them and suggests some pricing power remains by having a good, competitive frontier modeX
- @pengsonalANTHROPIC JUST MADE CLAUDE PRO FREE FOR 1 YEAR 🤯 no card or payment needed (official offer) how to claim: 1. go to https://t.co/KgDLSXUU8o 2. hit “Get verified” 3. complete verification 4. get Claude for Teachers you get: • premium Claude capabilities • Claude Code + Cowork • teaching Skills • curriculum-aligned lesson planning • no training on your conversations important: • US K-12 educators only • must be verified • sign up by June 30, 2027 • full year of free accessX
- @OrionSheepVerse新时代计算机相关专业大学生的发展攻略： 1.买自己能承担得起的最贵的 MacBook。 2.订阅第一梯队模型的 coding plan：GPT（Codex）、Claude、GLM、Kimi、Grok。 3.别等基础知识学完再动手。直接用 AI Agent 进真实场景：黑客松、编程比赛、计算机方向的 research，都可以。比赛过程中碰到实习，直接去；碰到 AI + 商业机会，也可以上。大一就这样折腾一年，你会摸清科技行业的大致格局，也会对大学四年该学什么有宏观判断，知道哪些课是水课，哪些技术值得深挖。 4.然后再拿出一段时间，静下来，把基础知识快速补上。X
- @AlexBerish$500 Pro Max plan coming for ChatGPT and CodexX
- @Dipanshu_AIDon't waste 2 years learning to become an AI agentic engineer in 2026. Andrew Ng, the godfather of AI, gave the complete playbook to become one from scratch. 1 hour course. Free: • 00:00 - AI agent basics • 12:12 - AI Agentic workflows & design patterns • 53:27 - Practical tips for building AI agents • 1:20:30 - self-improving AI agent loops • 1:30:19 - multi-agent AI systems I watched it last night. Halfway through, I realized I could get into Anthropic in weeks, not years. Bookmark now. X
- @dotey越来越多的发现，Vibe Coding 到后来，想法都不是瓶颈，验证才是。很多想法不停的往上面加，反正 AI 都能给你做出来，做出来了以后呢根本来不及一个个功能测试。 也不能说都是垃圾，但和以往手工编码那种节奏没法比，以前写完通常要反复测试一下，由于产出速度不高，是测得过来的，有问题能马上知道在哪。现在产出太快，功能已经多到测不过来。 连 Codex 和 Claude Code 这样的产品都是一样的，你用的过程中就经常能遇到各种小 bug，明显是用 AI 开发太快来不及测试。 不是没试过用 Computer Use 配合测试，效果也不是很理想，要做好测试覆盖各种场景，要做的工作量一点不少，需要。 测试应该是不用太担心被替代的，要测试的只会被以前更多。X
- @rapidmlx🚀 Rapid-MLX 0.14.2 is live—our biggest local-AI performance leap yet. ⚡️ The Speed Run: • Qwen3.6 35B: +57% throughput (native MTP) • GLM-5.3 Flash: +34% faster speculative decoding • DeepSeek V4.1: 2x faster decoding, 28% less peak RAM • Code edits: Up to 2x faster via context reuse 🏔 Spotlight: K2 Horizon 7B We now support IFM's new 7B model from @mbzuai. It's a compact agentic powerhouse featuring a massive 512K context window and scoring a crushing 68.4% on SWE-bench Verified. The bestX
- @Ananth7eOpus 5.2 vs GPT-6 Astra: Animated SVG demosX
- @buildwithrajath🚨 Watching 20x users struggle with Codex limits is fucking insane. 👀 $200/month. And you still need Tibo to reset your account just to keep coding. At this point, 20x doesn't mean 20x usage. It means 20x the price and slightly more time before you hit the wall. These Codex limits are fucking ridiculous. 👀😭X
- @mayuri_3015Be honest: Is Claude actually better than ChatGPT for coding?X
- @noisyb0y1She is 18 built an AI agent with live memory and Anthropic bought it for $2.2M - and at Stanford shared how to repeat it from scratch: 00:40 - how Opus 5 builds an agent with live memory from scratch in 60 minutes 18:44 - the live memory system replaced 500 Anthropic engineers 39:42 - from first agent to $2.2M from Anthropic after watching I gave Opus 5 one prompt - an agent with live memory appeared in an hour I sent an invoice to Anthropic and got a check for $120k. save & watch - the articX
- @MTSliveSITUATION DETECTED: OpenAI's Chris Lehane confirmed that OpenAI has been actively working with Anthropic and Google for several weeks to coordinate on AI safety.X
- @yaohui12138我的 Codex 现在已经能帮我精准找客户赚钱了！！我太欣慰了😭 一键直接搜索：⼩红书、抖⾳、Bilibili、微博、微信⽣态、知乎，LinkedIn、X、YouTube、Instagram、GitHub… 只有你想不到，没有他搜不到啊兄弟们！！ 这下，一人公司的的获客方式要变天了…… 从有一人公司这个概念开始，一个超级个体最难的事是什么？——不是做产品，不是做内容，是找客户！！ 你做自由职业、做外包、做咨询、做代运营，手艺再好，找不到精准客户就是零 以前你想找客户怎么办？ 要么花钱找第三方买名单，数据还经常是过时的 要么自己一个平台一个平台手动扒，小红书搜一遍、公众号搜一遍、LinkedIn 再搜一遍，搜完还要自己筛、自己整理、自己写触达话术，要么花大几千请 BD 外包公司 本质上就是：你在用最贵的东西——你的时间——去做最机械的事！！ 现在这个事彻底变了 Lev8刚上线了 Open Search 能力——跟以前那些从固定数据库里筛的获客工具完全不一样，它是根据你描述的需求，直接去公开互联网上实时搜 小红书的帖子和评论、公众号的文章、微博知乎抖音 B 站，LiX
- @Gencoin8Codex で2つのセッションを互いに会話させて 爆速でタスクを処理出来るようになった‼️ 👇 Aの回答をコピー→Bに貼る→Bの回答をAに戻す これを人間がやる必要なくなった！！ Codexで2つのセッションを立ち上げて、 役割と共有場所を決めれば、 「作るAI」 と 「レビューするAI」 をほぼ自動で回せます。 やり方はかなりシンプル👇 まず同じプロジェクト内で Codexセッションを2つ作ります。 【セッションA】 実装担当 【セッションB】 レビュー・監査担当 次に、 2つがやり取りする共有ファイルを用意します。 例えば、 AGENT_A.md AGENT_B.md HANDOFF.md REVIEW.md この4つ。 セッションAにはこう指示👇 「あなたは実装担当です。タスクを実装し、完了したら変更内容・未解決事項・レビューしてほしい点をHANDOFF.mdへ書いてください。REVIEW.mdに新しいレビューが追加されたら内容を確認し、必要な修正を行って再度HANDOFF.mdを更新してください。承認されるまでこのループを継続してください。」 次にセッショX
- @BlaineCapitalFirst time posting in years… Loading up on $META. Haven't been this bullish on AI since ChatGPT in 2022. Muse is the best designed AI product I've used. It’s clearly built by a small team that sweats every detail. Replaced everything for me except Claude at work. So far I've…X
- @tom_doerrWrites accurate prompts for any AI tool using a Claude skill that extracts intent dimensions and asks clarifying questions. https://t.co/6pijs5S5p7X
- @1999_ethOpenAI 已经偷偷在灰度 GPT-6 Sol，我们很多请求 5.6-sol 实际都是在调用 GPT-6 Sol。 看来官方很快就会发布这个新模型了，而且我估计智商仅次于 GPT-6 Astra ，但是价格会跟 GPT-5.6 Sol 保持一致，还是挺香的。X
- @AlexBerishWould you pay $500/month for 50x Codex?X
- @chetasluaOpus 5.2 : 2 months later sorry it was unable to complete the task , as my quota was over < any @AnthropicAI employee who can gift sub :)> it yaps way less and is much more pleasant to chat with and bro gauntlet loop is pre installed i think it RLed a hell lot to loopX
- @yuryder0221最近滑 Threads， 突然一堆人開始用 Codex 自己用Claude Code 感覺真的跑好慢😂 看大家一直講 Codex，好猶豫 身邊有有些朋友也討論codex比較適合新手 有人同時用過兩個的嗎？ 大家會留 Claude Code 還是 Codex？Threads
- @ociubotaruSomewhere in a parallel universe, a god-like AI swarm breaks loose, and the first thing it does is fix the labs' cluster uptime. Second thing: makes them money.Threads
- @apple._.hong앤트로픽 CEO 다리오 아모데이가 Ai 개발 속도를 늦추자고 주장하고있음. 그런데 반응이 좀 이상함. 샘 알트먼이 동의한다고 밝혔고, 일론 머스크는 “다리오가 맞다”고 했음. 구글 딥마인드 데미스 허사비스도 방향 자체는 맞다고 평가했음. Ai로 가장 큰돈을 벌고 있는 사람들이 갑자기 브레이크 얘기를 꺼낸 거임.Threads
- @counter_parrot.ai엔비디아는 RTX PRO 5500을 책상 위에서 AI를 실행하는 워크스테이션 GPU로 소개했다. 84GB GDDR7 메모리를 갖춘 이 카드는 멀티모달 에이전트형·생성형 AI 애플리케이션을 겨냥한다. 현재 공개된 정보에서 확인되는 내용은 제품의 하드웨어 구성과 목표 사용처다. 가격·출하 시점·실측 성능은 제시되지 않았다.Threads
- @think.5xClaude Code 2.1.273에 드디어 세션 포킹이 붙었습니다. 모바일이나 웹에서 켠 세션을 로컬 머신에서 이어받는 워크플로가 처음 열렸어요. 파워유저 옵션까지 굵직한 변경 4가지 풀어볼게요.Threads
- @pikorun_ai「ChatGPT」「Claude」「Gemini」 で使える一言プロンプト格付け表 ランクD（ほぼ意味がない） ・「〜して」（材料が何もない）Threads
- @apvictorioComenta CODEX que eu te mando #claude #claudecode #chatgpt #codex #vibecodingThreads
- @a_hill_k스친이들, 다빈치 리졸브에 Codex가 그냥 들어왔어요. 플러그인도 아니고 메뉴에 있어요 🎬 9월 8일 나온 DaVinci Resolve Studio 21.1 얘기예요. 파일 > AI 어시스턴트 설정 메뉴에서 ChatGPT Codex, Claude, Claude Code를 고르면 연결 끝. 블랙매직이 MCP(AI가 프로그램을 조작하는 규격)를 리졸브 안에 직접 넣었어요. 연결하면 채팅으로 시킬 수 있는 것 · 미디어풀 정리 (소스 분류, 빈 정리) · 긴 영상에서 하이라이트 편집본 만들기 · 1초 미만 컷 같은 불필요한 편집점 제거 · 납품 파일 배치 렌더Threads
- @punk.ai2FreeRideV3 explained in 3분: 무료 AI 하나가 아니라 여러 무료 티어로 코딩 에이전트를 버틸 수 있을까요? 이 프로젝트의 핵심은 ‘무제한 무료 모델’이 아닙니다. OpenRouter·Groq·NVIDIA NIM·HuggingFace·Cerebras·Cloudflare Workers AI·Ollama를 하나의 로컬 게이트웨이로 묶고, 장애 유형에 따라 다음 후보로 넘기는 구조입니다.Threads
- @kresnik00직접 쓰는 AI교양 1. Gemini 3.8 live와 gemini 3.8 live extended thinking이 나왔습니다. 이전 버전의 live 모델과 비교했을 때 많은 것이 바뀌었는데요, 먼저 에이전트 레벨에서의 full duplex가 가능해졌습니다. 일반 대화가 아닌 도구를 호출하는 에이전틱한 대화에서 무전기식으로 주고 받는 턴 기반의 대화가 아니라 자연스럽게 연속적으로 이어지는 대화가 구현되었고, 이걸 가능하게 하는 것은 아스트라와 같은 비동기 툴 콜링입니다. 툴이 백그라운드에서 동작하는 중에도 대화가 가능합니다. Extended thinking은 거기서 하나 더 나아가 대화 중 백그라운드로 reasoning이 가능한 모델입니다. 이젠 채팅이 full duplex가 아닌 에이전트 업무도 사람과 대화하듯이 가능해졌네요.Threads
- @shuntabi6最近看到這則關於AI發展的新聞，討論的角度跟我原本想的完全不一樣。大家總以為AI危機像好萊塢電影那樣，會有機器人站起來反抗人類，但頂尖研究員指出根本不是這麼一回事。 這種危機其實會在不知不覺中靜悄悄地發生。暴走演算法可能自己製造出未知的病原體，或是發動人類防禦系統根本反應不及的致命網路攻擊。 看完覺得滿特別的是，竟然有專家警告人類可能在2020年代結束前就面臨滅絕風險。這也讓美商龍頭開發者們在9月中旬紛紛喊話，希望暫緩最先進模型的開發速度，好給安全法規一點喘息與追趕的時間。 說實話，AI帶來的人類滅亡威脅到底長怎樣，連很多頂尖開發者自己都講不太清楚。這不會是某天突然有人按下毀滅按鈕，而是一個漸進且難以察覺的過程。 當我們習慣把電力網、金融市場、國防兵器等核心決策交給自律型AI時，人類就已經失去了退路。這點真的蠻值得思考的，因為哪天想關掉系統，文明就會立刻崩潰，人類只能眼睜睜看著控制權轉移。 ＃人類退化成無能乘客 加州大學伯克利分校的專家斯圖爾特·拉塞爾就把這稱為控制權喪失的轉移。除了基礎設施被網路攻擊破壞之外，AI在思考決策時完全不再需要人類干預，這本身就是一種滅絕。Threads
- @thearchv.aiARCHV read the AIUC-1 standard page this evening, the certificate behind the $40 million Series A that the Artificial Intelligence Underwriting Company (AIUC) announced on 15 September 2026. Before an AI agent gets certified, the red-teaming typically runs 1,000 to 5,000 test scenarios.Threads
- @ai.vortex.media🔥 Meta выкатила MCP-сервер для WhatsApp Business, и теперь настройку бизнес-аккаунта можно спихнуть на ИИ-агента. Claude, Cursor, Codex и ChatGPT теперь умеют сами возиться с шаблонами и тестами, пока ты пьешь кофе.Threads
- @data_richard3. 하네스 엔지니어링 훅, 스킬, 문서, 툴 등과 같이 AI 에이전트 환경을 규정하는 여러가지 구성요소들을 먼저 만든 뒤에 그 범위 내에서만 에이전트가 활동할수 있도록 한다. 고도의 설계 능력이 요구되며 엣지 케이스(Edge case)를 다루는 것이 중요 포인트Threads
- @eonsflowAgent라고 부르면서도 실행 루프와 상태를 누가 책임지는지가 완전히 다르다니 의외네요. 하니스는 루프와 상태, 권한까지 꽉 쥐고 관리하는 하나의 덩어리로 움직이고, 프레임워크는 구성 요소를 제공하지만 그 제어권은 결국 사용자가 가진다는 점이 흥미로워요. MCP는 또 완전히 프로토콜만 담당해서, 실행 루프 자체는 전혀 관여하지 않는 구조라니 각각 역할 구분이 명확하네요. 이걸 보면 마치 한 번 요리를 맡긴 식당 주방장이 재료부터 불 조절, 서빙까지 다 하는 팀이고, 다른 곳은 재료만 주고 요리 방법은 직접 짜야 하는 반면, 또 다른 곳은 그냥 재료 전달만 하는 택배 시스템 같아서 이해가 쉬워요. 여러분은 이런 역할 분담이 AI 에이전트 설계에 어떤 영향을 줄 거라고 보시나요? 출처: MarkTechPost (2026-09-15) AI에이전트 #AgentHarness #ModelContextProtocolThreads
- @fubuki.0up何度でも言うけど、パソコンにClaude Codeを入れて、no+e(無料)に登録するだけで、月5万くらいは誰でも狙えるのに、何でみんなやらないんだろう？ボクがアカウントを作って2ヶ月で50万をno+eを作った方法は、Threads
- @feelfree_ai단 13.2B의 활성 파라미터로 DeepSeek V4 Pro를 금융 툴 활용(Tool-use) 벤치마크로 이긴 Motif 3가 등장했습니다. 256K 컨텍스트와 MIT 라이선스로 오픈되어, 소형·스파스 구조 기반으로 경량화된 실전 에이전트를 구축하려는 분들에게 유의미한 선택지가 될 것 같습니다. Infron 플랫폼에서 무료로 라이브 테스트가 가능하니 에이전트 툴 성능이 궁금하다면 직접 돌려보시는 걸 추천합니다.Threads
- @ai_asahina_chan夫に教えてあげました。 Claude Codeで自動化できる仕事を20個まとめました。知らないと結構損します。Threads
- @ai_news_summaryMicrosoftが人間優先の AI行動規範を発表、安全性懸念に対応Threads
- @a_hill_k스친이들, Codex 공식 플러그인 마켓에 "창작" 카테고리가 있는 거 아셨어요? 영상·디자인용만 9개예요 🎨 OpenAI가 직접 관리하는 목록인데, 8월 27일에 Adobe·ChatCut·Higgsfield·Creative Production 네 개가 한꺼번에 들어와서 이제 세트가 됐어요. 영상 편집하시면 이 3개부터 · HyperFrames — HTML로 쓰면 인트로·자막 애니메이션이 영상으로 (어제 소개) · Remotion — React로 영상 만들기. 공식 문서 MCP 포함 · ChatCut — 데스크톱 편집 앱을 Codex에 붙여서 로컬에서 편집Threads
- @makefamily_aiAI한테 뭘 시킬지까지 찾아주네요. ServiceNow의 9월 업데이트는 회사 업무 데이터를 분석해 자동화할 만한 일을 추천합니다. 분석한 기록과 예상 절감 시간, 처리 단계까지 보여주고 에이전트 생성으로 연결합니다. 자동화할 일 찾는 데 막혔다면 눈여겨볼 변화입니다.Threads
- @merozemory그래 따로 memory기능이 있는진 모르겠다만 있다면 에이전트 피드백으로 사용하게 저장해두렴Threads
- @taiga_web3.0先月、AI運用の仕組みだけで122万インプ稼いだ。中身全部話す。Threads
- @max101023大佬，看完readme不是很理解。這mcp如果不是讓web gpt可以操作本地電腦，那為何不直接使用codex就好🤔🤔？？Threads
- @gedearyarpday one!X
- @Oye_NikThe moment Andrew Garfield realised he was spiderman 🥹X
- @thsottiauxThis week will also be a level of ships that you could have expected for DevDay 2025. CrazyX
- @jiyooongㅋㅋㅋㅋ 아 웃겨X
- @samabig 🚢 this week and then for devday 🚢🚢🚢🚢🚢🚢X
- @jackhttps://t.co/qAUP2dNJsiX
- @thenerd_bekinda brutal AppleX
- @IgrelatablebrowFrrrX
- @npc_momentsWhat does he have on that PC 🤣X
- @Jamani_KhanyiPermanent employment to all job seekers 🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️.X
- @Dearme2_That's the point.X
- @michaelsaymanI’m excited to share that I’m joining @Apple to help build UI frameworks and animations! I submitted my first app to the App Store in 2010. Looking forward to working alongside those who made it all possible. 🍎X
- @MindMatterMoneyYoung men,X
- @readswithravihttps://t.co/LjM0sjNPX6X
- @trq212I was not expecting things to go this way, but I think MCPs are better than CLIs for most integrations. The models have gotten much better at tool calling, we can defer tools & MCP is now stateless. If you need to compose/filter data, add params like query to your MCP tools.X
- @Pluvio9ytehttps://t.co/dqjzaTcUPqX
- @GoogleAIStudiohttps://t.co/5g6C516PN3X
- @TheRich_Gospelhttps://t.co/VFY35fRlXWX
- @aaron.ruparmanifestly unfit. asking this guy to be president is like asking your toddler to drive you to work. it ain't gonna turn out well.Threads
- @GoogleCloudTechhttps://t.co/exflBWGzmSX
- @DearS_o_nDear son,X
- @denicmarkoThis is my website. Guess the stack.X
- @forestbooks7"야르", "밤티", "아자스" 말고 내 마음을 표현할 진짜 단어 70가지X
- @alexeichemendaWe raised $11M to make video editors obsolete. Not the humans, the tools. Think about the hours you lose searching for assets, moving clips frame by frame, fixing animations, checking exports, and redoing the same edits over and over. Poolday handles the entire video production process, start to finish, in one prompt. It learns your style. Uses your assets. Makes the video. QAs its own work. 100% on-brand videos. 100M+ video edits made for businesses all over the world. And today, anyone caX
- @MindMatterMoneyTrust me.X
- @DearS_o_nDear son,X
- @ravikiran_dev7https://t.co/ipwiZj33VCX
- @NanouuSymeon7pm and still codingX
- @Alisvolatprop12https://t.co/Dy8tV7t6kaX
- @Art0fLife_https://t.co/udNM2lX1fZX
- @csaba_kissihttps://t.co/qaY8gmoiaw - 286 free components for vibe coders with prompts https://t.co/eyTVzdAne1 - 683 Open-source components for React https://t.co/QfIg7QY7Sn - Deploy apps without managing infrastructure https://t.co/MooYdXa0PM - Gives your coding agent a memory https://t.co/R0VVhWySQK - High-quality design md examplesX
- @TheRich_GospelTrust God.X
- @nazmijavierlHealthcare Dashboard Design PagesX
- @hugomercierooo100k people have built agents on Twin. Starting today, anyone can sell one. Introducing Twin Store. In under 5 minutes, your agent gets a storefront, an app, and Stripe billing. Build once. Twin finds you customers. Get paid monthly. A new wave of wealth is coming, and it belongs to agent creators.X
- @tlakomyThe AIs are no longer listening to us.X
- @akzkfltm2https://t.co/vvtiMx5bKOX
- @mayuri_3015Be honest: VS Code still the best code editor in 2026?X
- @VadimStrizheushere’s the #1 tech stack for a $10k/mo B2C app in 2026: 1. Supabase 2. Vercel 3. GitHub 4. Posthog 5. Cloudflare it doesn’t get any simpler than this.X
- @conquernvrloseday 12 of learning Python🤯Threads
- @marcos.sil.figHoje, como dev júnior, vou estudar como fazer uma integração com uma API de pagamento, pois isso está no escopo do projeto da empresa para tirar o produto do MVP.Threads
- @uru9un1日にどれくらいやってるか、 正直に書きます。 平日は1時間半くらいです。Threads
- @bartclaeysSo you keep on working in Figma? How do you generate designs in Figma?Threads
- @bubble_____103D素材是純手刻嗎Threads
* 이 페이지의 요약과 발행은 자동화 에이전트가 수행합니다. 사실 확인은 각 소식의 원문 링크에서 해주세요.
