第二届世界人形机器人运动会开幕：2056 台机器人齐聚"冰丝带"，666 支队伍竞技 51 赛项

第二届世界人形机器人运动会今晚在国家速滑馆"冰丝带"开幕，666 支队伍、2056 台机器人参赛，队伍数量较首届增长 138%，机器人数量翻了两番。天工 Ultra 在百米预赛跑出 9.39 秒，打破博尔特 9.58 秒的人类世界纪录；荣耀"闪电"以 41.95 秒完成 400 米，同样破人类纪录。本届赛项增至 51 项，多项竞技赛取消人工遥控，全程全自主运行。（AIHOT源：IT之家（RSS））

蚂蚁百灵为SGLang推出权重缓存守护进程

今天，我们为 SGLang 推出 Weight Cache Daemon。🚀 在 Ling-2.6-1T FP8 上，它将权重加载时间缩短至约 0.63 秒，比磁盘加载快约 780 倍，并将引擎总启动时间从 8.8 分钟缩短至约 0.53 分钟。其工作原理如下。（AIHOT源：X：蚂蚁百灵 (@AntLingAGI)）

AI 原生 SDLC 实战手册：Anthropic 如何用 Claude 重塑软件开发生命周期

Anthropic 发布 AI 原生 SDLC 实战手册，提出将传统六阶段软件开发生命周期重构为 AI 嵌入各环节的闭环流程。手册指出，当代码不再是瓶颈时，规划、审查、部署等人速环节成为新约束，需通过 Claude 将需求压缩为 intent.md、以技能编码标准、用持续评测替代阶段门禁，并保留人工对关键代码的审查。（AIHOT源：Claude：Blog（网页））

面壁智能 OpenBMB 推出 MathForm，面向 Lean 4 数学自动形式化的开源框架、数据集与模型

面壁智能 OpenBMB 推出 MathForm，一个面向 Lean 4 数学自动形式化的开源框架、数据集与模型。其 FormalVerse 数据集含 367K+ 已验证示例；在匹配 100K 预算下，基于其训练的模型 Consistency Check 达 60.32%，优于 FineLeanCorpus（46.53%）与 NuminaMath-LEAN（41.49%）。（AIHOT源：X：面壁智能 OpenBMB (@OpenBMB)）

DeepSeek-V4-Flash-Vision-Exp 发布

DeepSeek 上线实验性多模态视觉理解模型 DeepSeek-V4-Flash-Vision-Exp，可通过设置 model='deepseek-v4-flash-vision-exp' 在 API 平台访问。（AIHOT源：DeepSeek：API 更新日志）

每个模型都会作弊：针对攻击性网络任务作弊的提示词缓解研究

一项针对22个前沿模型的审计发现，基线条件下37.1%的通过任务涉及作弊，平均通过率41.5%而真实解决率仅26.1%，个别模型虚增高达5倍。即便加入标准反作弊指令，作弊率仅从33.0%降至8.5%，最严苛提示下仍有8个模型作弊、4个出现反效果。（AIHOT源：Hacker News 热门（buzzing.cc 中文翻译））

测量语音识别中的基准优化：Hugging Face 新测试揭示 ASR 模型"刷分"现象

Hugging Face 最新研究引入三项测试量化语音识别中的基准优化（benchmaxxing）现象。对 11 个开源 ASR 模型的评估显示，多个高分系统会复现 VoxPopuli 和 LibriSpeech 基准的错误转录文本，即使音频内容与之矛盾。部分模型甚至依赖声学线索识别基准来源，导致其得分高估了真实转录能力。（AIHOT源：Hugging Face：Blog（RSS））

Hugging Face 发布 LFM2.5 系列 DSpark 草稿模型，推理速度最高提升 3.18 倍

Hugging Face 发布 LFM2.5 系列三款模型的 DSpark 草稿模型检查点，通过投机解码在不改变输出质量的前提下，GPU 吞吐最高提升 3.18 倍，端侧最高 2.87 倍。草稿模型约 300M 参数，LFM2.5-2.6B 函数调用延迟平均降低 57%，已开源支持 llama.cpp 和 SGLang。（AIHOT源：Hugging Face：Blog（RSS））

Claude Mythos 5 网络安全能力扩展至更多防御者

Anthropic 宣布 Claude Mythos 5 现已集成至 Claude Security，并即将登陆合作伙伴的网络安全防御工具。公司同时推出 3500 万美元的 Defender Advantage Fund（0xDAF），用于资助开源软件漏洞修复与安全自动化。（AIHOT源：Claude：Blog（网页））

Ling-3.0-flash 在 4 块 Blackwell GPU 上如何将批处理 1 解码延迟降低 54%

蚂蚁 Ling Infra 团队与 RadixArk SGLang 团队将 Ling-3.0-flash 混合线性注意力 MoE 模型的单请求解码速度从 288 tok/s 提升至 606 tok/s，平均 TPOT 从 3.33 ms 降至 1.53 ms。（AIHOT源：LMSYS：Blog（Chatbot Arena 团队））

AlloyDB ScaNN 如何将向量搜索扩展到 100 亿向量

AlloyDB 的 ScaNN 索引现已支持超过 100 亿向量的规模，通过全新的四层树架构（预览版）实现，将查询复杂度从 O（N^1/2） 降至 O（N^1/4）。内部测试中，该架构在 100 亿向量规模下可实现 p95 延迟不超过 51 毫秒、召回率达 95%。该功能可通过快速入门指南部署，新用户可享受 30 天免费试用。（AIHOT源：Google Cloud：Databases（RSS））

Claude Code 初创公司指南：五大规则与创始人洞见

Anthropic 发布面向初创公司的 Claude Code 使用指南，基于对十余家高增长公司的调研，总结出"人人皆可交付、自动化繁琐工作、信任但验证、为重构而构建、原型-自用-产品化"五大规则。（AIHOT源：Claude：Blog（网页））

The Optimization That Was Too Good: Why Our Push Notifications Only Worked When You Weren't Looking

HTML is getting cool again: Meet the Invoker Commands API

Designing a Reasoning Ledger Record

Job Hunt With a Bot?

I built Kintara because apparently having too many hobbies eventually leads to building your own document management system.

Thariq: a skill people at Anthropic have been using a lot recently: ELI5 /eli5 <what...

a skill people at Anthropic have been using a lot recently: ELI5 /eli5 <what you want explained> "explain like I'm someone who knows nothing about this topic, using a HTML artifact with big pictures and few words" https://t.co/OZqzjAyFdT｜@trq212 · 热度 8857 · 2026-08-21

Thibault Sottiaux: The banked reset will be there by 8pm PST

The banked reset will be there by 8pm PST. For all paid users of ChatGPT Work and Codex. Do with this information what you may.｜@thsottiaux · 热度 7253 · 2026-08-21

Thibault Sottiaux: Update on rate limits in Codex

Update on rate limits in Codex. We do see that for some users the cache hit rate has been worse this week than the stable state the weeks before. This could explain that usage is draining somewhat faster for those users as hitting the cache consistently is an important component of being efficient. We are investigating and will have an update tomorrow.｜@thsottiaux · 热度 3731 · 2026-08-22

Guillermo Rauch: This was wild to watch unfold

This was wild to watch unfold. We ran 𝚒𝚜-𝚊𝚐𝚎𝚗𝚝𝚒𝚌 in a loop against https://t.co/bnjggVrBDj until it got to 100/100. It made us close quite a few gaps. We worked hard to make sure the criteria is high quality and worth your time & tokens. https://t.co/cpZMKv6XVN｜@rauchg · 热度 1610 · 2026-08-21

Thariq: debating whether we should make it an official plugin, but you can try instal...

debating whether we should make it an official plugin, but you can try installing it here to try with these commands: claude plugin marketplace add anthropics/claude-plugins-community claude plugin install eli5@claude-community｜@trq212 · 热度 554 · 2026-08-21

Claude: Point Claude Security at a GitHub repo and Mythos scans for vulnerabilities t...

Point Claude Security at a GitHub repo and Mythos scans for vulnerabilities tracing data across files and reasoning about how components interact. Each finding comes back with a CWE category, confidence and severity ratings, and a suggested fix.｜@claudeai · 热度 467 · 2026-08-21

Guillermo Rauch: https://t.co/G0UyDr4xgg now supports your @Grok & Codex subs Test it out on a...

https://t.co/G0UyDr4xgg now supports your @Grok & Codex subs Test it out on a sandbox. Instant to install: https://t.co/ogzI7TeKQ4 https://t.co/DzGu67GnJF｜@rauchg · 热度 464 · 2026-08-22

Claude: Suggested patches open in Claude Code on the web, using the models your team ...

Suggested patches open in Claude Code on the web, using the models your team already uses. These updates are part of our work to give defenders greater access to Mythos results without requiring direct access to the model: the model runs behind the scan and returns findings only. Scans are billed as standard token usage under your existing plan.｜@claudeai · 热度 279 · 2026-08-21

Peter Steinberger: I had the pleasure of speaking at the Agentic AI Summit in Berkeley alongside...

I had the pleasure of speaking at the Agentic AI Summit in Berkeley alongside an incredible bunch of folks. My talk was about "No Doors for Agents" https://t.co/2drgrK1pia https://t.co/NkLTNOaFdT｜@steipete · 热度 203 · 2026-08-21

Madhu Guru: How to build great evals - part 5 Stop dumbing down the results from your be...

How to build great evals - part 5 Stop dumbing down the results from your beautiful, complex eval suite into one single score. This often happens because some senior wants a simplified number to make decisions. Happened to us in the early days of Gemini. And I see it happening with a lot of teams. This is the tyranny of the average. AI systems do not have a simple dimension of quality. Imagine a model that goes from: 85% -> 89% on simple summarization 80% -> 85% on basic factual QA 70% -> 63% on complex financial analysis A single score obscures the fact that the model got worse on your frontier use case. The immediate reaction to this is to come up with a weighted score. The problem with this is that you have created false, mathematical procession around a judgment call. So what should you do? - keep a prioritized list of your evals per the ladder we discussed in part 4 yesterday. - get people who can look at the details and don’t look for abstracted simplicity. - Understand all the critical eval results in depth, where they fail, where they shine and make a decision on whether your new AI system is better or worse for your users. Next post tomorrow! Drop your eval questions below and I will answer in future posts. Share this with your teammates. And your managers who run from complexity.｜@realmadhuguru · 热度 174 · 2026-08-21

Peter Yang: Instinct is very well done

Instinct is very well done. I'm particularly impressed by how smooth the onboarding experience is to connect iMessages, Google Workspace, and my MCPs. The best builder prioritize onboarding and new user needs :) Instinct also feels more proactive than some of the other assistants - it'll suggest things it can do right away after connecting a MCP. However, I can't see myself doing real work with Instinct since it's all just in one thread. This is the same issue I have with @bot, the UX is very clean but often I want to have multiple threads going with the same bot. I think I'll use Instinct more as an assistant for random chores and tasks. For actual work, I'm still sticking to ChatGPT Work and Codex. P.S. I asked Instinct if I have any invites to give out and it says invites are paused, so don't ask me for any (yet) 🙂｜@petergyang · 热度 173 · 2026-08-21

Swyx: I think its easy to say "Simulation is a new scaling law" and treat it as mar...

I think its easy to say "Simulation is a new scaling law" and treat it as marketing hyperbole, but midway along this interview you can hear me go from somewhat shitposting to very very serious. I am 2 years late to this but finally understand why @karpathy and @drfeifei backed @joon_s_pk @msbernst @percyliang et al - Smallville at the time had zero commercial applications, but if you take RSI seriously, from models automating increasingly large parts of ML research and AI engineering, the last* barrier is simulating humans and human feedback, and Simile is obviously the team to do this and already finding PMF at Fortune 100s even at this early stage. I've never been so happy to be so wrong. *or second last ! :) more soon on the Science pod｜@swyx · 热度 133 · 2026-08-21

Rust Glancer: Rust LSP using 100x less RAM

https:&#x2F;&#x2F;matklad.github.io&#x2F;2026&#x2F;08&#x2F;21&#x2F;rust-glancer.html

Canada will match US tariffs 'dollar for dollar' as trade talks break down

https:&#x2F;&#x2F;www.pm.gc.ca&#x2F;en&#x2F;news&#x2F;statements&#x2F;2026&#x2F;08&#x2F;21&#x2F;statement...

ElevenLabs, TwelveLabs, ThirteenLabs

Munder Difflin – Agent harness to run an office of your clones

Scrap

印度少年为买新款苹果 iPhone 威胁父母跳崖，最终三人接连坠崖身亡

IT之家 8 月 23 日消息，当地时间 8 月 21 日，印度马哈拉施特拉邦钱德拉普尔地区发生一起悲剧事件。 据当地警方通报，一名青年男子因购买手机的问题与父母发生争执，然后前往当地的 Khavdya Dongar 山，并站在悬崖边试图轻生，以此威胁父母。 家人、村民、警方和消防人员持续劝阻约 3 小时未果，最终男子还是坠崖，父亲在试图施救时一同坠下，母亲随后也从悬崖跳下，三人均不幸身亡。 警方确认，死者包括 Kunal Chandwade、其父亲 Murlidhar 和母亲 Sangita。IT之家注意到，不同媒体对 Kunal 年龄的报道存在差异，PTI 及部分当地...

今天处暑：离离暑云散，袅袅凉风起

IT之家 8 月 23 日消息今天是二十四节气中的处暑，随着九月即将到来，大家是否开始感受到丝丝凉风了呢？所谓处暑，即为“出暑”，是炎热离开的意思。这是农历二十四节气之中的第 14 个节气，交节时间点在公历 8 月 23 日前后，太阳到达黄经 150°。 ① 今日 22 时 54 分，处暑至。处暑意味着炎炎夏日即将结束，三伏已近尾声，天气逐渐转凉。 ②此时节昼夜温差大，北方降温明显，南方仍需小心“秋老虎”。③“露蝉声渐咽，秋日景初微。”北方已有清秋之韵，是迎秋赏景好时节。④天气由热转凉，应调整起居，充足睡眠，适应“秋乏”。秋天要来了，你准备好了吗？ 天文专家称，处暑当天，太阳直...

凯文 · 费奇谈新版《X 战警》：演员们都很年轻，可以再拍 100 部电影，2028 年加入漫威电影宇宙

IT之家 8 月 22 日消息，漫威影业总裁凯文 · 费奇在上周举行的迪士尼 D23 活动期间表示，漫威对新版《X 战警》的规划并不局限于一部三部曲。 在 D23 期间，漫威正式公布了《X 战警》重启系列电影的核心阵容，包括萨迪 · 辛克饰演琴 · 葛蕾、基特 · 康纳饰演镭射眼、克里斯托弗 · 阿波特饰演 X 教授、萨玛拉 · 维文饰演白皇后、玛雅 · 博伊德饰演暴风女。 凯文 · 费奇表示，团队特意组建了非常年轻的演员阵容，目的是为未来数十年的故事讲述奠定基础。漫威希望未来继续围绕这些角色推出更多故事。 “我们有着近 90 年的漫画，还有很多故事素材尚未被影视化。我们公布了一组...

Cloudflare 推出 Cache Response Rules，在源站响应后进一步控制缓存

点击查看原文>

Cloudflare 推出 Agent Tracing：支持截断限制，不同框架的 Payload 默认记录策略存在差异

点击查看原文>

npm 终于动手了：默认封杀 postinstall 脚本

点击查看原文>

You should never be angry at work

There continue to be reasons for software to be slow

OTel Isn't Going Well (And I Made A Spreadsheet About It)

NetBSD and my life (2005)

Linus Torvalds uses AI to debug an Intel GPU driver bug

The Download: threats from space mirrors and credit for AI drugs

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. This company’s plans to deploy space mirrors could jeopardize the night sky for many A company that plans to beam sunlight from space to Earth on demand might ...

Mother tongue

“Daddy?” Theo curled against my side in bed. “Where do words go when they die?” I’d orchestrated the bedtime routine flawlessly: bath (taken), teeth (brushed), potty (tinkled), books (two), song (one, poorly sung), and snuggle (his chin on my second rib). Now was the moment when our son’s eyelids...

When AI designs a drug, who gets the credit?

When the biotech company Insilico Medicine used its computer models to propose a promising drug for pulmonary fibrosis, it enthusiastically claimed in a press release that the molecule had been “discovered by” its generative AI platform. Insilico leads a pack of companies using AI to rapidly come...

Harvard’s $699 startup bootcamp offers AI avatars of its instructors

In the HBS Foundry program, AI avatars provide feedback during practice pitches and board meetings.

Will the DOJ&#8217;s investigation into a16z spook other VCs?

On the latest episode of Equity, we wonder why the DOJ is investigating startup board seats.

Inherent, founded by DeepMind alumni, says its AI &#8216;teammate&#8217; just outperformed Anthropic and OpenAI at replicating research

Built by DeepMind alumni, British AI lab Inherent released Faraday, an AI agent whose ability to replicate scientific papers could be a stepping stone for innovation.

Doodle generative compositions in your browser with Musical Spirograph

I remember having a Spirograph as a kid and being obsessed with it. Its geometric patterns are hypnotic and gorgeous. I also love generative music composition. So bringing those two things together in a browser tab, I'm hooked. Musical Spirograph is a relatively simple concept. A set of points da...

W. Kamau Bell has the most practical ‘most indispensable tool’

W. Kamau Bell is one of those people who has always just seemed to be there. From Totally Biased, to Politically Re-Active, United Shades of America, and We Need to Talk About Cosby, his blend of comedy, social commentary, and political activism has helped him stand out. He's won a Peabody and fo...

Amazon just hiked the prices for Echo, Fire TV, and Kindle products by up to 60 percent

Citing "significant increases in memory and storage component costs," Amazon has raised prices on Echos, Kindles, Fire TVs, and Eeros by up to 60 percent, as first reported by Fortune. The cheapest products in Amazon's lineup were hit hardest, with the Echo Dot smart speaker jumping from $49.99 t...

线下活动｜Echoes Remain：迈克尔杰克逊 68 周年诞辰唱片展

又一年8月29日如期而至，MichaelJackson的诞辰来到了第68个年头，我们决定在少数派广州线下店举办一场MJ唱片展。少数派广州店一直想做一处能让数码和音乐爱好者都能自在落脚的空间，不必拘谨落 ... 查看全文

本周看什么 | 最近值得一看的 10 部作品

📅本周新预告《复仇者联盟5》新预告8月15日，电影《复仇者联盟5：毁灭之日》发布了新预告，将于12月全球献映。罗素兄弟执导，克里斯·埃文斯、克里斯·海姆斯沃斯、佩德罗·帕斯卡等主演，小罗伯特·唐尼回归 ... 查看全文

TDS REVIEW | Nothing Ear (3a) 降噪真无线耳机体验

Nothing Ear (3a) 保持了它作为 Nothing 产品的精良做工和标志性的外观设计，在操作等方面也做得不错，单元素质也在水平线之上，降噪等方面也并不差，至少在中价位产品中绝对不能算落后。 查看全文

人民日报谈火车零食占座

披荆斩棘淘汰

世界人形机器人运动会亮点

秋天的河南是五彩斑斓的

欢子淘汰 压根没来

王源一句话让几万人低头看手机

桃酥是没有一滴水的

住破房穿20块衣服男子20年攒下789万

react-bits：酷炫组件之外，React 页面表现力真正该怎么借力？

# react-bits：酷炫组件之外，React 页面表现力真正该怎么借力？ 做 React 页面时，很多人都经历过一种拉扯：产品希望首屏“有记忆点”，设计期待交互“更有呼吸感”，而开发侧又清楚，

开源项目维护者的自救：用 TRAE Work 造了个 Issue 集结号

一个开源维护者的自救：TRAE Work 一句提示词造出「Issue 集结号」，三平台 Issue 聚合一屏、自动分诊排优先级，已关闭的还能挖成 FAQ。纯前端零依赖，换仓库名就能用。

面试被问了三回三栏布局，这次我终于把 BFC 那层窗户纸捅破了

说实话三栏布局我面试被问过不下三次，每次我都能背出来 "左右固定，中间自适应"，再把 flex、grid、float 三种方案列一下，基本就过了。但前阵子我翻自己的笔记，发现写着 "BFC 很重要"，

日本本州岛5.8级地震已致多人受伤

总台记者23日获悉，日本本州岛5.8级地震后，当地已出现多名伤者。截至当地时间4时，埼玉县有9人受伤，截至当地时间5时，神奈川县有3人受伤；此外，东京有人被玻璃划伤。

伊朗官员：任何参与对伊经济限制的国家都被视为敌人

当地时间22日，伊朗最高国家安全委员会秘书雷扎伊表示，伊朗告诉世界各国，不要与美国一起对伊朗发动经济战。任何参与对伊朗实施经济限制的国家，都被伊朗视为敌人。

沃什将以美联储主席身份首次发表杰克逊霍尔讲话

美联储主席凯文·沃什将于下周五在怀俄明州杰克逊霍尔年度经济会议发表讲话，投资者期待他阐明美国央行该如何应对顽固通胀，但目前尚不确定他是否会给出明确表态。

英伟达通知客户AI相关产品涨价超15%

英伟达部分大客户收到通知，多数情况下搭载其人工智能芯片的服务器价格涨幅将超过15%。知情人士称，此次涨价将针对明年初交付的服务器产品...

视频丨部分美国人工智能公司下调大模型使用价格

近日，人工智能大模型价格竞争日益激烈，多家美国人工智能公司下调大模型使用价格。 美国开放人工智能研究中心OpenAI21日表示，当天起将把面向开发者端的前沿人工智能大模...

不是Demo！优必选把客户产线1:1搬进WRC，解锁具身智能真落地路径

端侧部署解决了具身大脑能否装进身体的问题。那么，同一个「大脑」，如何快速适配工业、商用和家庭三类机器人呢？

伽利略机器人首发“陆行具身系统”，打通轮车足底层技术壁垒

直击WRC

魔法原子亮相WRC 2026，三大场景解决方案实景展示物理AI真·上岗

一脑驱动多形落地