2026-07-16（内容获取于 07/16 04:53）

Tesla driver in fatal Texas crash pressed accelerator 100%, NTSB confirms

TechCrunch · 07/16 04:22

The safety board confirmed Tesla's account of the crash, which the company shared days after it happened last month.

中文介绍 美国国家运输安全委员会（NTSB）确认，上月得克萨斯州特斯拉致命车祸中，司机百分之百踩下了加速踏板，证实了特斯拉此前的说法。

Roblox is shutting down its video chat service

The Verge · 07/16 04:02

Roblox will be shutting down Roblox Connect, its video calling service introduced in 2023. Roblox Connect let you video chat with other people using your Roblox avatar, which would be able to mimic the movements you were making in real life. You could also run around with people on the call in a sha

中文介绍 游戏平台Roblox宣布将关闭其于2023年推出的视频通话服务Roblox Connect，该服务允许用户使用Roblox虚拟形象进行视频聊天并模仿真人动作。

AI slop movies are the new direct-to-video cash grabs

The Verge · 07/16 04:00

This weekend, cinephiles across the world will march to their local theaters to feast their eyes on Christopher Nolan's new adaptation of The Odyssey. It's on track to rake in anywhere between $80-$100 million in just a few days. People are clearly excited to see how Nolan uses cutting-edge filmmaki

中文介绍 随着人工智能技术发展，业界出现了一种新的“AI粗制滥造电影”现象，它们被比作过去的“直接发行录像带”式盈利手段，与如克里斯托弗·诺兰的《奥德赛》等大制作形成对比。

Windows 0-day drops the same day Microsoft releases record number of patches

Ars Technica · 07/16 03:59

HiveLegacy is a "powerful primitive" that's likely capable of other nefarious actions.

中文介绍 微软发布创纪录数量补丁的同一天，一个名为“HiveLegacy”的Windows零日漏洞被曝光。该漏洞被描述为一个“强大原语”，可能被用于其他恶意活动。

OpenAI is now using AI to attack its own AI, and it's working better than humans ever did

The Decoder · 07/16 03:47

OpenAI's internal GPT-Red model finds successful attacks in 84 percent of test scenarios through self-play training. Human red teamers manage just 13 percent. The results feed directly into hardening models like GPT-5.6 Sol. The article OpenAI is now using AI to attack its own AI, and it's working b

中文介绍 OpenAI正利用内部GPT-Red模型“攻击”自身AI，在84%测试场景中成功发现弱点，远高于人类红队（13%）。此成果将直接用于强化如GPT-5.6 Sol等模型。

Daniel Ek’s body-scanning startup Neko Health raises another $700M

TechCrunch · 07/16 03:46

Neko Health has developed proprietary body-scanning technology, which it couples with bloodwork, to assess a person's health.

中文介绍 丹尼尔·埃克（Daniel Ek）创办的身体扫描初创公司Neko Health再融资7亿美元。该公司开发了专有的身体扫描技术，结合血液检测，用于评估个人健康状况。

Amid hardware legal battle, OpenAI releases a $230 keyboard for Codex

TechCrunch · 07/16 03:41

OpenAI, which is in the middle of a legal battle with Apple over hardware trade theft allegations, just released a light-up keyboard designed to be paired with its agentic coding app.

中文介绍 正在与苹果公司就硬件商业盗窃指控进行法律诉讼的OpenAI，发布了一款售价230美元的发光键盘，专为其代理式编码应用Codex设计。

FCC to repeal 39% TV ownership cap in boost for Trump-friendly news orgs

Ars Technica · 07/16 02:52

FCC chairman claims power to repeal TV ownership limit set by Congress.

中文介绍 美国联邦通信委员会（FCC）计划废除39%的电视所有权上限，此举可能有利于支持特朗普的新闻机构。FCC主席声称有权废除国会设定的这一限制。

Microsoft patches bug in video game Age of Empires II

TechCrunch · 07/16 02:47

The vulnerability in the decades-old game could have allowed hackers to take over victims’ computers with a malicious game invite.

中文介绍 微软修复了经典视频游戏《帝国时代II》中的一个漏洞。该漏洞可能允许黑客通过恶意游戏邀请，完全控制受害者的电脑。

Enshittification

The Verge · 07/16 02:40

Bryan, a food broker from Michigan, wasn't sure if he'd be able to make it to urgent care in time. He started feeling off on Thursday, and by Saturday, he was having to use the bathroom every 15 to 30 minutes. "It's no joke about the explosive diarrhea," Bryan, who asked that his last name be withhe

中文介绍 一名密歇根州男子因感染微小隐孢子虫（Cyclospora）出现严重腹泻症状，需每15至30分钟如厕。这起食源性疾病事件引发了FDA和CDC对可能的仙女座塔可钟（Taco Bell）相关疫情的调查。

In memoriam: 7 of our favorite Sam Neill films

Ars Technica · 07/16 02:39

The actor, who starred in 1993's Jurassic Park, died Monday in Sydney, Australia, at the age of 78.

中文介绍 曾出演1993年电影《侏罗纪公园》的著名演员萨姆·尼尔（Sam Neill）于周一在澳大利亚悉尼去世，享年78岁。

AI Isn’t Smarter Than a Baby—Yet

Wired AI · 07/16 02:30

Babies are tremendous learning machines, and key advances for AI may soon be found in the architecture of their little brains.

中文介绍 文章指出，当前人工智能（AI）在某些方面尚未超越婴儿的智能水平。研究者认为，婴儿大脑的结构蕴含着巨大的学习能力，未来AI的关键进展可能从中获得启发。

Something’s glowing on the Pixel 11’s camera bar

The Verge · 07/16 02:24

A new teaser for Google's upcoming Pixel 11 lineup reveals that the phones will feature some kind of glowing orb on the camera bar, as reported by 9to5Google. Google's store page for the Pixel 11 has a short video that shows the glowing, color-shifting orb to the one side of the camera bar. The stor

中文介绍 谷歌即将推出的Pixel 11系列手机预告片显示，其后置摄像头模组上将包含一个发光的、可变色的球体。根据9to5Google报道，谷歌商店页面短视频也展示了这一新设计。

Apple bans home services from its upcoming Maps ads

TechCrunch · 07/16 02:21

Apple has published the policies governing its upcoming Maps advertising business, revealing a strategy that differs from Google’s. The new rules prohibit home services businesses like plumbers, electricians, locksmiths, and roofers from advertising on Apple Maps, along with several other sensitive

中文介绍 苹果公司公布了其即将推出的地图广告业务政策，规定禁止水管工、电工、锁匠等家庭服务类企业在其地图上投放广告，这一策略与谷歌的地图广告模式有所不同。

SpaceX falls to $135 IPO price ahead of Starship launch

TechCrunch · 07/16 02:06

The stock has steadily fallen from the euphoric post-IPO high, showing that markets may be sobering up to the promises CEO Elon Musk made before and after SpaceX went public.

中文介绍 临近“星舰”发射，SpaceX股价跌至135美元的IPO发行价。该公司股价从上市后的高点持续下跌，表明市场对首席执行官埃隆·马斯克此前的承诺正趋于冷静。

OpenCut-app/OpenCut

TypeScript · ★ 71,327 · 🍴 7,327 · 📈 1,505 stars today

The open-source CapCut alternative

中文介绍 OpenCut 是一个开源的视频剪辑软件，旨在成为流行剪辑工具 CapCut 的免费替代品。它为视频创作者和内容编辑提供了一个功能齐全且可定制的平台，用于处理视频素材。用户可以利用其免费开源的特性，在不依赖商业软件的情况下进行高效的视频制作和后期编辑。

Nutlope/hallmark

CSS · ★ 8,110 · 🍴 426 · 📈 1,119 stars today

Anti-AI-slop design skill for Claude Code, Cursor, and Codex.

中文介绍 hallmark 致力于提供“反AI废话”的设计技巧，专注于优化 Claude Code、Cursor 和 Codex 等AI工具生成的代码或设计。它旨在帮助开发者和设计师提升AI输出的质量，使其更具人性化、美观和实用性，避免生成平庸或低质量的内容。适用于需要将AI辅助成果融入高质量项目的工作者。

mattpocock/skills

Shell · ★ 172,105 · 🍴 14,782 · 📈 2,160 stars today

Skills for Real Engineers. Straight from my .claude directory.

中文介绍 mattpocock/skills 汇集了一系列专为“真正的工程师”设计的 AI 技能，这些技能直接源自作者的 .claude 目录。它旨在为 Anthropic Claude 等 AI 编码助手提供实际、高效的工具和能力，帮助开发者在编程、代码审查和自动化任务中提高效率和准确性，尤其适合需要强大 AI 编程辅助的专业人士。

moeru-ai/airi

TypeScript · ★ 42,455 · 🍴 4,246 · 📈 144 stars today

💖🧸 Self hosted, you-owned Grok Companion, a container of souls of waifu, cyber livings to bring them into our worlds, wishing to achieve Neuro-sama's altitude. Capable of realtime voice chat, Minecraft, Factorio playing. Web / macOS / Windows supported.

中文介绍 Airi 是一个自托管的 AI 伴侣项目，旨在创建用户私有化、可定制的虚拟角色（如“waifu”），并支持实时语音聊天，目标是达到 Neuro-sama 的交互水平。它被设计为一个类似 Grok Companion 的系统，让用户能够完全掌控自己的 AI 互动体验，甚至可能与 Minecraft 等游戏集成。

Dicklesworthstone/destructive_command_guard

Rust · ★ 4,713 · 🍴 178 · 📈 497 stars today

The Destructive Command Guard (dcg) is for blocking dangerous git and shell commands from being executed by agents.

中文介绍 dcg (Destructive Command Guard) 是一个旨在阻止自动化代理执行危险 Git 和 Shell 命令的工具。它通过拦截潜在的破坏性操作，如强制推送或删除关键文件，有效提升系统安全性，防止误操作导致的数据丢失或服务中断。此工具特别适用于 CI/CD 流程、自动化脚本或多用户协作环境，为开发者和运维人员提供一道安全屏障。

HKUDS/Vibe-Trading

Python · ★ 23,594 · 🍴 4,029 · 📈 924 stars today

"Vibe-Trading: Your Personal Trading Agent"

中文介绍 Vibe-Trading 是一个个人交易 Agent 项目，旨在为用户提供智能化的交易辅助。它可能通过分析市场数据、情绪指标等，自动化交易决策和执行策略，帮助用户更好地管理投资组合并捕捉交易机会。该项目适用于希望利用 AI 技术增强交易能力、实现自动化投资或寻求个性化交易策略的个人投资者和量化交易爱好者。

openinterpreter/openinterpreter

Rust · ★ 65,386 · 🍴 5,650 · 📈 345 stars today

A coding agent for low-cost models

中文介绍 openinterpreter 是一个专为低成本 AI 模型设计的编程智能体。它通过提供一个代码执行环境，赋能经济型 AI 模型也能高效完成复杂的编程任务。该项目旨在降低开发者在利用 AI 辅助编程时对昂贵大型模型的需求，使 AI 编程自动化更具成本效益。用户可以是需要自动化代码生成、调试或项目管理的开发者，特别适合在资源有限的环境下，利用较小规模的语言模型提升开发效率。

HKUDS/DeepTutor

Python · ★ 26,196 · 🍴 3,559 · 📈 128 stars today

DeepTutor: Lifelong Personalized Tutoring. https://deeptutor.info/.

中文介绍 DeepTutor 是一个提供“终身个性化辅导”的智能教育系统。它利用深度学习等先进技术，为学习者提供定制化的、贯穿整个学习过程的辅导服务。该项目旨在解决传统教育中缺乏个性化指导的问题，通过适应学生的学习进度、风格和知识盲点，提供实时反馈和定制内容。它非常适合学生进行自主学习，教师辅助教学，以及任何寻求高效、个性化学习体验的终身学习者。

HenryNdubuaku/maths-cs-ai-compendium

TypeScript · ★ 5,827 · 🍴 739 · 📈 729 stars today

Become a cracked AI/ML Research Engineer

中文介绍 maths-cs-ai-compendium 是一个旨在帮助有志成为顶尖 AI/ML 研究工程师的综合性学习资料库。它系统性地整理了数学、计算机科学和人工智能领域的核心知识，为学习者提供了清晰的学习路径和必备的理论基础。用户可以通过此项目深入理解 AI/ML 算法背后的原理，构建扎实的知识体系，从而在 AI 研发领域取得突破。

Shubhamsaboo/awesome-llm-apps

Python · ★ 121,815 · 🍴 17,985 · 📈 1,278 stars today

100+ AI Agent & RAG apps you can actually run — clone, customize, ship.

中文介绍 这是一个精选列表，收录了超过100个可实际运行的AI Agent和RAG（检索增强生成）应用示例。开发者可以轻松克隆、自定义并部署这些项目，作为构建自己LLM应用的基础。它为寻找LLM实践案例、学习AI Agent设计模式或RAG实现方案的开发者提供了丰富的灵感和快速启动资源。

coreyhaines31/marketingskills

JavaScript · ★ 39,658 · 🍴 6,309 · 📈 390 stars today

Marketing skills for Claude Code and AI agents. CRO, copywriting, SEO, analytics, and growth engineering.

中文介绍 `marketingskills` 项目旨在为 Claude Code 等 AI 代理提供专业的营销技能，涵盖转化率优化 (CRO)、文案撰写、搜索引擎优化 (SEO)、数据分析和增长工程等关键领域。它通过集成特定的知识和策略，使 AI 能够更有效地执行营销任务，解决传统 AI 在专业营销领域能力不足的问题。该项目适合营销团队、增长专家、内容创作者或任何希望利用 AI 自动化或优化市场营销活动的个人和企业，帮助他们提升营销效率和效果。

YimMenu/YimMenuV2

C++ · ★ 1,399 · 🍴 353 · 📈 21 stars today

Experimental menu for GTA 5: Enhanced

中文介绍 YimMenuV2 是一个专为《侠盗猎车手5：增强版》（GTA 5: Enhanced）设计的实验性游戏修改菜单。它允许玩家在游戏中访问和启用各种自定义功能或作弊选项，从而改变游戏体验。该项目主要面向希望深入探索游戏机制、解锁隐藏内容或以非传统方式享受 GTA 5 的玩家。它通常通过注入游戏进程来修改参数，为热衷于游戏模组（modding）的用户提供更多自由度和娱乐性。

hasaneyldrm/exercises-dataset

HTML · ★ 14,268 · 🍴 1,714 · 📈 951 stars today

1,324-exercise fitness dataset — animation GIFs, 180×180 thumbnails, muscle-group & equipment data, and step-by-step instructions in 6 languages. The exercise data layer behind the LogPress app.

中文介绍 hasaneyldrm/exercises-dataset 提供了一个包含433种健身运动的综合数据集，每项运动详细收录了名称、类别、目标肌肉群、所需器械、完整说明、缩略图及动画视频。该数据集旨在为健身应用开发、运动科学研究或AI模型训练（如姿态识别）提供高质量、结构化的多媒体资源。开发者可利用此数据快速构建功能丰富的健身指南、训练计划生成器，或用于AI在健身领域的应用开发，极大地降低了数据收集与整理的门槛。

Copy Claude Fable 5’s Thinking Before It’s Gone

@TheAIColony · 225.1K 粉丝 · 1.2M 阅 · 603 赞 · 140 转

07/10 15:57

You Have a Few Days to Copy Fable 5’s Brain Into a Cheaper Model. Here Is How. July 12 is the last day Claude Fable 5 sits inside your plan for free. After that, it moves to pay-per-use credits, and

中文介绍 该推文提醒用户在 Claude Fable 5 结束免费期前，抓住机会学习或复制其独特思考模式。暗示了在模型计费模式改变（可能指 7 月 12 日后转为按量付费）之前，进行知识提取或策略迁移的重要性，帮助用户在成本增加前最大化利用其功能。

You just hired a million bad employees.

@gsivulka · 10.2K 粉丝 · 591.2K 阅 · 512 赞 · 99 转

07/15 00:41

AI was supposed to replace human labor. It did the opposite. For the first time in history, humans are cheaper than software. And AI is creating more jobs than it eliminates. Technology has

中文介绍 博主gsivulka以「你刚雇佣了一百万个糟糕的员工」为引子，暗示了对AI工具或其应用效果的批判性观点。

Making Fable Cheaper Than Opus

@joon_h_lee · 515 粉丝 · 366.3K 阅 · 521 赞 · 43 转

07/14 01:03

We replaced Opus 4.8 with Fable 5, and Devin’s bill went down. Fable 5 costs twice as much per token as Opus 4.8. But when we ran both models on FrontierCode 1.1 using our new Fusion architecture,

中文介绍 博主分享了通过用Fable 5替换Opus 4.8显著降低Devin账单的经验。尽管Fable 5的每token成本是Opus 4.8的两倍，但利用其新Fusion架构在FrontierCode 1.1上运行模型后，整体费用反而下降。这表明通过优化模型架构和运行时环境，即使使用成本更高的模型也能实现更优的经济效益。

The Most Human Technology Ever Made

@illscience · 30.4K 粉丝 · 332.5K 阅 · 539 赞 · 69 转

07/14 01:05

Over the weekend, Mira Murati wrote a phenomenal essay that you should read in full: “The future worth building is human.” I think it’s one of the best essays of the year, because it gets at the

中文介绍 博主推荐OpenAI首席技术官Mira Murati撰写的文章「值得建设的未来是人性化的」。该文章探讨了将人类价值融入AI发展的未来方向，并被博主评价为年度最佳文章之一。它强调了在AGI时代，技术应以人为本，而非单纯追求技术突破。

The Great Wave Has Arrived (from GLM CEO Jie Tang)

@bingxu_ · 6.5K 粉丝 · 318.6K 阅 · 505 赞 · 88 转

07/11 23:10

-- Bing Xu's Note --- I came across an internal GLM letter on the Chinese app RedNote, purportedly written by @jietang, and translated the Chinese text in the images into English. --- Letter Start

中文介绍 智谱AI CEO唐杰发表了「大模型浪潮已至」的观点，指出了当前人工智能发展的重要转折点，并对未来AI大模型的趋势与影响进行了预判。

AI's Value Capture problem

@JayaGup10 · 34.4K 粉丝 · 306.1K 阅 · 504 赞 · 67 转

07/09 12:27

AI could be one of the most powerful value-creation technologies in history and still have a value-capture problem. Alex Karp says companies buying AI risk leaking their IP into Anthropic and OpenAI.

中文介绍 推文聚焦「AI的价值捕获问题」。探讨了AI作为强大价值创造技术，其产生的价值如何在企业与AI服务提供商之间分配，以及如何有效转化为商业利润的深层挑战。

The Reverse Information Paradox

@satyanadella · 7.0M 粉丝 · 146.5K 阅 · 1.1K 赞 · 162 转

07/12 23:09

In the age of intelligence, how should firms protect their core IP? Nobel Prize winning economist Kenneth Arrow famously described a paradox in the market for information. “Its value for the

中文介绍 微软CEO Satya Nadella探讨了「智能时代」企业如何有效保护核心知识产权的挑战。他引用诺贝尔经济学家Kenneth Arrow的信息悖论，深入分析了AI背景下知识价值与保护机制的新思考。

Cost effective harnesses with Fable

@RLanceMartin · 36.7K 粉丝 · 119.8K 阅 · 510 赞 · 51 转

07/11 02:00

There is a lot of interest in cost effective use of Fable 5. Agent harnesses will get better at knowing when to use frontier intelligence. I wanted to share some tests I’ve run to better understand

中文介绍 博主分享关于如何通过「Agent harnesses」实现 Claude Fable 5 成本效益使用的方法。该方法旨在优化 AI Agent 在调用先进智能模型时的决策机制，从而降低运行成本，提高效率。

ORACLE: Official AI Agents Trade on Polymarket

@Oracle_Globall · 4.1K 粉丝 · 109.7K 阅 · 1.9K 赞 · 527 转

07/10 07:44

In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We

中文介绍 ORACLE 项目宣布，到 2026 年，自主 AI Agent 已成为预测市场 Polymarket 上最有效的策略之一。目前 Polymarket 超过 30% 的交易活动来自算法和 AI 驱动的钱包。这篇推文预示了 AI Agent 在去中心化预测市场中的巨大潜力和应用前景，可能涉及产品发布或愿景分享。

ORACLE: Official AI Agents Trade on Polymarket

@Oracle__Promo · 4.1K 粉丝 · 106.5K 阅 · 1.8K 赞 · 543 转

07/11 07:04

In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We

中文介绍 ORACLE 项目宣布，到 2026 年，自主 AI Agent 已成为预测市场 Polymarket 上最有效的策略之一。目前 Polymarket 超过 30% 的交易活动来自算法和 AI 驱动的钱包。这篇推文预示了 AI Agent 在去中心化预测市场中的巨大潜力和应用前景，可能涉及产品发布或愿景分享。

ORACLE: Official AI Agents Trade on Polymarket

@Oracle__Trade · 4.0K 粉丝 · 105.3K 阅 · 1.9K 赞 · 557 转

07/11 07:03

In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We

中文介绍 ORACLE 项目宣布，到 2026 年，自主 AI Agent 已成为预测市场 Polymarket 上最有效的策略之一。目前 Polymarket 超过 30% 的交易活动来自算法和 AI 驱动的钱包。这篇推文预示了 AI Agent 在去中心化预测市场中的巨大潜力和应用前景，可能涉及产品发布或愿景分享。

A Framework for Frontier AI and the Dawning of a New Age

@demishassabis · 1.4M 粉丝 · 98.1K 阅 · 1.5K 赞 · 236 转

07/14 17:10

This is a pivotal moment in human history. Artificial General Intelligence (AGI), a system that exhibits all the cognitive capabilities the brain has, is probably only a few short years away. When we

中文介绍 Google AI CEO Demis Hassabis认为，人类正处于关键历史时刻，通用人工智能（AGI）的实现可能仅在数年之内。AGI系统有望展现人类大脑的全部认知能力。他可能在分享一个关于前沿AI的框架，以应对即将到来的新时代。

gpt-5.6-sol without hitting limits

@theo · 356.6K 粉丝 · 77.0K 阅 · 774 赞 · 27 转

07/12 06:59

I've burned over $200,000 of tokens with gpt-5.6-sol. It's a great model. That said, it's a bit too easy to hit your limits on the $200 Codex Pro sub. OpenAI has been generous about resets, but that

中文介绍 博主Theo分享了关于在应用中使用模型「gpt-5.6-sol」的经验，侧重于如何避免触及其固有的使用限制。这篇帖子可能提供了在特定场景下，实现模型无障碍运行的策略或操作建议。

I'm pretty sure isRecord is our fault

@tldraw · 73.7K 粉丝 · 68.3K 阅 · 570 赞 · 29 转

07/10 05:21

A few weeks ago, many programmers (or perhaps former programmers) began to notice that OpenAI’s Codex model had developed an obsession with defensive coding. Nothing captured this better than

中文介绍 tldraw团队指出，近期许多程序员观察到OpenAI的Codex模型在代码生成时，对防御性编程表现出「痴迷」。博主推测，这种特殊行为模式，尤其是针对「isRecord」等特定模式的过度使用，可能与tldraw自身产品的代码生成实践有所关联。

DenseReward: Dense Reward Learning via Failure Synthesis for Robotic Manipulation

👍 0

07/14 08:00

Reinforcement learning holds great promise for improving robot policies beyond the limits of imitation learning. However, its practical adoption remains bottlenecked by the lack of reliable vision-language reward models that provide dense and informative feedback. Two key challenges remain: acquirin

中文介绍 该研究提出DenseReward，旨在通过「失败合成」的方法，解决强化学习在机器人操作中视觉-语言奖励模型反馈不足的问题。现有强化学习在改进机器人策略方面潜力巨大，但受限于缺乏可靠且提供密集信息的视觉-语言奖励模型。DenseReward通过合成失败场景来学习密集的奖励函数，以期提高机器人策略的效率和鲁棒性。

The Seriality Gap in Video Diffusion Models

👍 0

07/14 08:00

When one ball strikes another, then another, video models should predict the consequences of each bounce. In controlled experiments on multi-ball hard-sphere dynamics, we find that the performance of standard bidirectional video diffusion degrades as the causal chain lengthens, even when provided mo

中文介绍 该研究探讨了视频扩散模型在处理长因果链事件时的「序列性鸿沟」问题。通过对多球硬球动力学实验发现，标准双向视频扩散模型预测物体连续碰撞后果的性能，会随着因果链的增长而下降。这表明模型在理解和预测复杂序列事件方面存在局限性，影响其在现实世界中的应用潜力。

FlowWAM: Optical Flow as a Unified Action Representation for World Action Models

👍 0

07/14 08:00

World Action Models (WAMs) are able to leverage pretrained video generators for both world modeling and action prediction. However, directly leveraging such video generators for control raises a new challenge: how to represent actions in a suitable form that aligns with pretrained video generators w

中文介绍 本研究提出FlowWAM，将光流作为世界动作模型（WAMs）的统一动作表示。WAMs能利用预训练视频生成器进行世界建模和动作预测，但如何以适应这些生成器的方式表示动作是一个挑战。FlowWAM旨在通过光流这一通用且与预训练模型对齐的表示方式，有效桥接视频生成器与机器人控制，提升模型的泛化能力。

DermDepth: Toward Monocular Metric Scale 3D Reconstruction Models for Dermatology

👍 0

07/14 08:00

Dermatological practice routinely involves measuring and tracking lesion size, morphology and texture, as critical components of wound or skin cancer screening, monitoring and diagnosis. To accomplish this task, practitioners often image the skin surface with commonly available off-the-shelf camera

中文介绍 DermDepth提出一种用于皮肤病学的单目度量尺度3D重建模型。皮肤病学实践中，测量和追踪病变的大小、形态和纹理是皮肤癌筛查、监测和诊断的关键环节。传统方法存在限制，该模型旨在通过单目图像实现精确的3D重建，为皮肤病诊断提供更准确的量化依据，从而辅助医生进行决策。

X-Lens: Real-Time Metric Depth Estimation with Heterogeneous Cameras

👍 1

07/14 08:00

We present X-lens, a compact feed-forward model for metric depth estimation from a variable number of calibrated fisheye and pinhole views. To support real-time downstream perception, X-lens is built around a geometry-aware heterogeneous camera formulation with two key components. Learnable calibrat

中文介绍 X-Lens是一种紧凑的前馈模型，能利用可变数量的校准鱼眼和针孔相机视图进行实时度量深度估计。为支持实时下游感知，X-Lens设计基于几何感知的异构相机框架，集成两项关键技术。这提高了在复杂多视角环境下进行精确深度感知的能力，适用于自动驾驶、机器人导航等各种实时应用场景。

ChunkFlow: Towards Continuity-Consistent Chunked Policy Learning

👍 0

07/14 08:00

Vision-language action (VLA) models increasingly adopt chunked action heads to satisfy real-time constraints; however, this introduces boundary jitter: overlapping regions between consecutive chunks often yield inconsistent predictions, degrading temporal coherence and the task success rate. Existin

中文介绍 本文提出了ChunkFlow，旨在解决视觉-语言动作（VLA）模型中分块策略学习存在的「边界抖动」问题。VLA模型为满足实时性常采用分块动作头部，但导致连续块重叠区域预测不一致，损害了时间连贯性。ChunkFlow致力于提升分块策略的连续性一致性，从而改善VLA模型的任务表现和鲁棒性。

Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification

👍 0

07/14 08:00

Accurate dermatological diagnosis naturally necessitates equitable performance across diverse populations, yet a systematic lack of expertly annotated images, especially for underrepresented skin tones and rare diseases, impedes progress toward measurably fair methods. We introduce cgDDI (Controllab

中文介绍 该研究针对皮肤病诊断中恶性肿瘤分类的公平性问题，提出一种可控生成多样化皮肤病图像的方法。由于缺乏不同肤色及罕见疾病的专家注释图像，现有分类方法难以实现对不同人群的公平表现。该生成模型旨在弥补数据鸿沟，为训练更公平、高效的皮肤病变分类模型提供合成数据，促进普惠医疗发展。

ViHoRec: A Quality-Controlled Vietnamese Hotel Recommendation Dataset and Cold-Start Benchmark

👍 1

07/14 08:00

Recommender-system research for Vietnamese remains limited by the absence of a public, well-documented hotel interaction resource. Building such a resource is challenging for three reasons: cross-platform hotel names must be reconciled before interactions are comparable; quality must be audited with

中文介绍 本文介绍了ViHoRec，一个经过质量控制的越南酒店推荐数据集及冷启动基准。由于缺乏公开且完善的酒店互动资源，越南推荐系统研究面临挑战。ViHoRec旨在通过协调跨平台酒店名称和严格质量控制，解决数据稀缺问题，为越南语酒店推荐系统的开发与评估提供重要资源，尤其是在冷启动场景。

ExToken: Structured Exploration for Efficient Vision-Language-Action Reinforcement Fine-tuning

👍 0

07/14 08:00

Reinforcement Learning (RL) has demonstrated significant potential for improving Vision-Language-Action (VLA) models on complex manipulation tasks. However, its practical scalability remains severely limited by the substantial cost of environmental interactions. In this work, we first investigate th

中文介绍 ExToken提出一种结构化探索方法，旨在提高视觉-语言-动作（VLA）模型通过强化学习（RL）进行微调的效率。虽然RL在改善复杂操作任务的VLA模型方面潜力巨大，但高昂的环境交互成本严重限制了其可扩展性。ExToken通过更高效的探索策略，减少了训练所需的交互次数，从而降低了微调VLA模型的成本。

Hy-Embodied-VLM-1.0: Efficient Physical-World Agents

👍 5

07/14 08:00

Building capable embodied agents requires not only multimodal perception and understanding, but also agentic capabilities for reasoning about actions, adapting to evolving situations, and interacting with the physical world. In this report, we introduce Hy-Embodied-VLM-1.0, an efficient and powerful

中文介绍 本文介绍了Hy-Embodied-VLM-1.0，一个旨在构建高效物理世界具身智能体的模型。有能力的具身智能体需要结合多模态感知、理解、行动推理、适应变化以及与物理世界互动的能力。Hy-Embodied-VLM-1.0致力于提升这些能力，使其能在物理环境中更有效地执行任务，并展现出类人智能，推动具身AI发展。

AVSCap: Orchestrating Audio-Visual Synergy for Omni-modal Video Captioning

👍 0

07/14 08:00

Omni-modal video captioning is not merely combining visual captioning with audio transcription: a useful caption must describe how visual actions, speech, music, and sound effects co-evolve. Existing large multimodal models often fail at this relational step, treating audio and visual streams as loo

中文介绍 本文提出了AVSCap，旨在为全模态视频字幕任务协调视听协同。研究指出，有用的字幕应描述视觉动作、语音、音乐和音效如何共同演变，而非简单结合。现有大型多模态模型常未能捕捉这种关系。AVSCap致力于克服此局限，通过有效整合视听信息，生成更具关联性和描述力的全模态视频字幕。

TrustVLA: Mechanism-Guided Inference-Time Defense Against Vision-Language-Action Backdoors

👍 0

07/14 08:00

Vision-Language-Action (VLA) models are deployed through pipelines that end users cannot audit, and a poisoned VLA can behave normally on clean observations while a small visual trigger redirects a long-horizon robot policy before any failure becomes observable. Existing vision or language defenses

中文介绍 本研究引入TrustVLA，一种针对视觉-语言-动作（VLA）模型后门攻击的机制引导推理时防御方法。VLA模型部署中存在安全风险，即中毒模型在正常运行时，微小视觉触发器即可在失败显现前重定向长期机器人策略。TrustVLA旨在通过在推理阶段检测并缓解这些隐蔽的后门威胁，保障VLA模型的安全性和可靠性。

Evidence-Grounded AI for Musculoskeletal Care

👍 13

07/14 08:00

Musculoskeletal diseases are among the leading causes of disability worldwide and create the greatest global need for rehabilitation. Because recovery, remodelling and degeneration often unfold over months to years, musculoskeletal care requires longitudinal management that repeatedly integrates evo

Self in Space: Benchmarking Self-Awareness and Spatial Cognition in UAV Embodied Intelligence

👍 2

07/14 08:00

Autonomous UAV systems increasingly rely on multimodal large language models (MLLMs) to operate in complex real-world environments. Such embodied scenarios require not only understanding the surrounding space but also maintaining a coherent representation of the agent itself. However, existing UAV-o

Function-Aware Fill-in-the-Middle as Mid-Training for Coding Agent Foundation Models

👍 0

07/14 08:00

Coding agents must integrate external tool returns into ongoing reasoning - a capability that standard left-to-right pretraining on code exposes only in its forward direction. We observe that the action-observation-continuation loop of a coding agent is structurally isomorphic to a function call sit

Let RGB Be the Language of Vision

👍 3

07/14 08:00

This work introduces a unified formulation for vision models, where diverse forms of visual information beyond natural images, such as masks, depth maps, and other structured visual signals, are all represented as RGB images, while general visual tasks can be converted into a common RGB-to-RGB image

Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning

👍 0

07/14 08:00

Reinforcement learning with verifiable rewards without human-annotated data, often referred to as zero RL, has emerged as a powerful paradigm for eliciting chain-of-thought reasoning. However, due to computational constraints, existing studies are largely restricted to small models, leaving the trai

VistaVLA: Geometry- and Semantic-Aware 3D Gaussian-Grounded VLA for Robotic Manipulation

👍 0

07/14 08:00

Vision-Language-Action (VLA) models have emerged as a powerful end-to-end paradigm for robotic manipulation by mapping language instructions and 2D visual inputs directly to actions. However, these models lack an explicit, scene-level 3D representation, limiting their ability to reason over spatial

Reducing Temporal Redundancy for Efficient Vision-Language-Action Inference

👍 0

07/14 08:00

Vision-Language-Action (VLA) models exhibit strong generalization for robotic manipulation, yet their high inference latency limits real time deployment. We identify two primary sources of temporal redundancy in existing VLA pipelines: repeated visual encoding of highly similar consecutive frames an

DA-Nav: Direction-Aware City-Scale Vision-Language Navigation

👍 0

07/14 08:00

City-scale outdoor navigation is currently hindered by the heavy reliance on dense maps or costly navigation supervision. In this work, we introduce a novel paradigm for leveraging directional instructions from commercial navigation tools (e.g., Google Maps). To bridge the gap between commercial ins

AgentCheck: A Reproduce-Intervene-Mitigate Workbench for LLM Agents over MCP

👍 0

07/14 08:00

Tool-using LLM agents are mostly evaluated assuming all tools work. When a tool times out, returns a week-stale value, or has its description poisoned in deployment, the developer needs a controlled way to reproduce the failure, test a fix, and confirm the fix worked before deployment. We present Ag

Navigating the Mirage: A Dual-Path Agentic Framework for Robust Misleading Chart Question Answering

👍 0

07/14 08:00

Despite the success of Vision-Language Models (VLMs), misleading charts remain a significant challenge due to their deceptive visual structures and distorted data representations. We present ChartCynics, an agentic dual-path framework designed to unmask visual deception via a "skeptical" reasoning p

Latent-Identity Tuning in Text-to-Image Personalization Models

👍 3

07/14 01:59

Generating and editing a person's face demands high precision, as even minor modifications can significantly alter a subject's perceived identity. Current personalization and editing methods built on general-purpose text-to-image models, however, often lack the precision required for fine-grained fa

Metacognition in LLMs: Foundations, Progress, and Opportunities

👍 16

07/14 01:58

Metacognition is a foundational component of intelligence critical to effective learning, problem solving, decision-making, communication, and more. In recent years, it has become increasingly recognized as a cornerstone of capable, transparent AI systems. Yet while LLMs have made significant progre

Evidence-Backed Video Question Answering

👍 2

07/14 01:49

Current Video Large Language Models (Video LLMs) excel in question answering (QA) but largely operate as black boxes, providing textual answers without verifiable visual grounding. Existing explainability efforts rely on textual rationales or sparse bounding boxes, which struggle to capture complex

AdvancedMathBench: A Benchmark Suite for Advanced Mathematical Proof Generation and Verification

👍 21

07/14 01:38

Large language models (LLMs) have achieved remarkable performance on high-school and olympiad-style mathematics, yet their capabilities on advanced mathematics remain poorly understood. Existing benchmarks, however, fall short in both scope and evaluation granularity: they provide limited disciplina

MET: Theory-Grounded and Culture-Aware Multilingual Moral Reasoning

👍 7

07/13 23:59

Language models are increasingly used for moral decision-making across diverse linguistic and cultural contexts, yet existing work overlooks multilinguality on three aspects: 1) multilingual evaluation benchmarks use direct translation, failing to adapt culture-specific items; 2) inference-time meth

Automated Synthesis of Facial Mechanisms for Conversational Animatronic Robots

👍 1

07/13 23:22

Animatronic faces are a central component of socially interactive robots, enabling rich nonverbal communication through facial articulation. However, state-of-the-art animatronic faces are typically tailored systems: each new facial geometry requires extensive manual mechanical redesign, making larg

RAGU: A Multi-Step GraphRAG Engine with a Compact Domain-Adapted LLM

👍 0

07/13 23:20

Graph retrieval-augmented generation (GraphRAG) enhances large language models with structured knowledge, yet existing systems construct knowledge graphs in a single extraction pass, producing noisy entities and brittle retrieval. RAGU, an open-source modular GraphRAG engine, addresses this by separ

ABot-3DWorld 0: A Universal World Model to Explore Any 3D Space

👍 0

07/13 23:14

We present ABot-3DWorld 0, a universal multimodal 3D world model that turns text, image, and video inputs into high-fidelity, explorable 3D worlds. At the heart of our framework is a unified Spatial Generative Primitive (SGP), a compact tuple of a high-quality panorama and a spatial point cloud that

QuickQuill

07/16 04:47

Private, on-device meeting notes for Mac

中文介绍 QuickQuill 是一款专为 Mac 设计的应用，提供设备上私密记录会议笔记的功能，确保信息安全。

V2Fun

07/16 04:47

Generate 3D character with 8K textures and AI motion capture

中文介绍 V2Fun 平台利用 AI 动作捕捉技术，可生成带有 8K 纹理的高质量 3D 角色。

Velo 3.0

07/16 04:46

AI video infrastructure to explain, train, and sell faster.

中文介绍 Velo 3.0 提供 AI 视频基础设施，旨在帮助用户更快地完成解释、培训和销售等任务。

Agently

07/16 04:46

Your whole stack, running itself!

中文介绍 Agently 是一款自动化工具，能够让整个技术栈自主运行，简化开发和管理流程。

Campus

07/16 04:45

One project space for humans and AI agents

中文介绍 Campus 为人类和 AI 代理提供一个统一的项目协作空间，优化团队工作效率。

YAGNI

07/16 04:44

Proactive agent teams you manage like humans

中文介绍 YAGNI 允许用户像管理人类团队一样管理主动式 AI 代理团队，提升工作自动化水平。

Tiptap AI Toolkit

07/16 04:44

Empower your AI to directly edit documents in real time.

中文介绍 Tiptap AI Toolkit 赋能 AI，使其能够直接实时编辑文档，显著提升内容处理效率。

RecordMeeting

07/16 04:39

Record and transcribe any calls without announcement

中文介绍 RecordMeeting 是一款能够录制并转录任何通话内容的工具，且无需提前通知对方。

Crustdata Recruiter

07/16 04:38

Claude Skills to turn Claude into a 100x Recruiter

中文介绍 Crustdata Recruiter 通过集成 Claude AI 技能，将 Claude 转化为高效招聘工具，大幅提升招聘效率。

MentalHappy 3.0

07/16 04:38

Find & host online support groups that actually fill up

中文介绍 MentalHappy 3.0 帮助用户寻找并组织在线支持小组，确保社群活跃度并促进成员交流。

Codex and Claude Shipped Browser Updates. This Changes Everything.

07/15 21:00

中文介绍 Codex和Claude两大AI模型近期发布了浏览器更新，预计将对用户与AI的互动方式产生显著影响。这些更新可能包括功能改进、性能优化或更深度的浏览器集成，旨在提升用户体验并拓宽AI工具的应用场景。

OpenAI Just Merged ChatGPT and Codex. This Changes Everything.

07/13 01:24

中文介绍 据YouTube频道“Riley Brown”消息，OpenAI已将其大型语言模型ChatGPT与代码生成模型Codex进行整合。此举被认为将彻底改变人工智能的应用格局，可能极大提升AI在自然语言理解和编程辅助方面的综合能力，预示着AI技术发展进入新阶段。

Grok 4.5 + Cursor: Full Guide

07/10 01:51

Claude Fable 5 Built My Website in 20 Minutes (Here's How)

07/07 01:37

How to Create Your Own AI Assistant (It's So Easy)

07/01 09:40

Poweful Agents inside Slack

06/25 06:39

AI Agents Just Changed Forever: GLM 5.2, Codex Skills, Claude & Cursor

06/21 21:30

9 AI Agent Skills To Get Ahead of 99% of People

06/18 10:27

How to build high quality internal tools with Retool

06/18 05:21

SpaceX Just Bought Cursor for $60B. It’s About to Take OVER.

06/17 05:31

闲鱼卖 iPhone 遇到骗子

07/15 20:32

18 回复 · Apple 节点

/goal 已经跑了 1d3h，完成 143 个 commit， 8w 行代码。跑的我道心破碎。

07/15 17:50

21 回复 · 程序员 节点

[苹果智能国内备案，阿里千问能力集成其中]

07/15 16:45

7 回复 · Apple 节点

Apple 智能备案了

07/15 16:12

17 回复 · Apple 节点

Apple Fitness+ 挺好用的

07/15 10:04

12 回复 · Apple 节点

IOS27 还能用 AI 吗？

07/15 09:30

6 回复 · Apple 节点

记录 macos beta3 午睡差点机毁人亡

07/14 18:29

18 回复 · Apple 节点

公司提供 5 周年礼物是一台美版 ipad11 的 esim 版，想卖掉发现不好出手。

07/14 11:04

22 回复 · Apple 节点

哪位朋友帮忙用 mac 测试下我的 Python 脚本

07/10 16:51

12 回复 · Python 节点

为什么这段 cuda 的并行前缀算法不会数据竞争

07/05 23:28

6 回复 · Python 节点

Brainless: Shadcn components that look like Claude Code, Codex and Grok

07/16 03:42

9 points · 1 comments

Voxatron

07/16 03:40

14 points · 6 comments

Duskers, the scary command line game, is getting a sequel

07/16 03:27

42 points · 5 comments

Inkling: Our Open-Weights Model

07/16 02:12

292 points · 82 comments

Artie (YC S23) Is Hiring Software Engineers

07/16 01:01

1 points · 0 comments

A General Goal-Conditioned Minecraft Model

07/16 00:35

33 points · 11 comments

Collection of Digital Clock Designs

07/16 00:33

103 points · 30 comments

Open-source memory for coding agents, synced over SSH

07/16 00:15

82 points · 18 comments

Unsolved Problems in MLOps

07/16 00:15

34 points · 4 comments

Codex Micro

07/16 00:12

220 points · 193 comments

Show HN: misa77 - a codec that decodes 2x faster than LZ4 (at better ratios)

07/15 23:58

I've spent the last few months working on this codec.It has the following characteristics: - SOTA decompression throughput in its ratio class - Decent ratios (comparable to LZ4 at high effort levels) - Slow compression Most of the gains can be attributed to reducing branches and making decompre

Launch HN: Coasty (YC S26) – An API for computer-use agents

07/15 23:51

Hey HN, we’re Nitish and Prateek, the founders of Coasty (https://coasty.ai/computer-use). We’re building computer-use agents that can complete workflows inside legacy desktop software and web applications without usable APIs.Developers send Coasty a natural-language task either throu

When A.I. is a member of the family

07/15 23:37

54 points · 74 comments

Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

07/15 23:34

179 points · 99 comments

OpenAI loses trademark dispute at EU court

07/15 22:32

190 points · 130 comments

My midlife crisis Corolla is fast, furious, and modded

07/15 22:23

124 points · 270 comments

Towards a harness that can do anything

07/15 22:08

132 points · 67 comments

Mysteries of Telegram Data Centers (2022)

07/15 21:22

206 points · 101 comments

The Three-Second Theft: Why AI Voice Fraud Outruns Every Defence

07/15 21:18

148 points · 191 comments

Briar is in maintenance mode

07/15 20:33

123 points · 85 comments

Sleep regularity is a stronger predictor of mortality risk than sleep duration (2023)

07/15 19:46

597 points · 297 comments

Prioritize mental health, and why communication is so important

07/15 19:27

244 points · 187 comments

Telegram Serverless

07/15 18:06

159 points · 90 comments

Stripe and Advent have made a joint offer to acquire PayPal – sources

07/15 11:32

210 points · 106 comments

The Memory Heist

07/15 03:40

71 points · 8 comments

The Conservationist Who Turned 40 Terabytes of Public Data into a Video Game

07/14 21:52

86 points · 16 comments

Designing APIs for Agents

07/14 00:15

6 points · 0 comments

MITS: Rockets, Calculators, and Personal Computers

07/13 21:20

5 points · 0 comments

Show HN: Capn-hook for coding agents – don't grep the same mystery twice

07/13 05:17

14 points · 4 comments

Today I Rescued 7,234 Old GIFs

07/13 03:00

60 points · 5 comments

v2.1.210

07/15 07:45

What's changed Added a live elapsed-time counter to the collapsed tool summary line so long-running tool calls visibly tick instead of looking stuck Added a startup warning for Write(path), NotebookEdit(path), and Glob(path) permission rules — use Edit(path) or Read(path) instead Fixed isolation: 'w

中文介绍 Anthropic的Claude Code项目发布了v2.1.210版本更新。此次更新增加了折叠工具摘要行的实时耗时计数器，以显示长时间运行的工具调用进度。同时，为`Write(path)`、`NotebookEdit(path)`和`Glob(path)`等权限规则新增了启动警告，并建议用户改用`Edit(path)`或`Read(path)`权限。

v2.1.209

07/14 14:36

What's changed Fixed /model and other dialogs being blocked in claude agents background sessions (reverts an overly broad guard)

中文介绍 Anthropic的Claude Code发布v2.1.209版本。此次更新主要修复了一个错误，解决了Claude代理后台会话中「/model」及其他对话框被阻碍的问题，该问题是由于之前一项过于宽泛的保护机制导致的。

v2.1.208

07/14 09:10

What's changed Added screen reader mode: opt-in plain-text rendering for screen reader users. Run claude --ax-screen-reader, set CLAUDE_AX_SCREEN_READER=1, or add "axScreenReader": true to settings. Added vimInsertModeRemaps setting: map two-key insert-mode sequences like jj to Escape in vim mode Ad

中文介绍 Anthropic的Claude Code发布v2.1.208版本，引入了两项新功能。一是增加了屏幕阅读器模式，为视障用户提供可选的纯文本渲染，可通过命令行、环境变量或设置启用。二是新增了「vimInsertModeRemaps」设置，允许用户自定义Vim插入模式下的双键序列映射。

v2.1.207

07/11 08:52

What's changed Auto mode is now available without CLAUDE_CODE_ENABLE_AUTO_MODE opt-in on Bedrock, Vertex AI, and Foundry; disable via disableAutoMode in settings Fixed the terminal freezing and keystrokes lagging while streaming responses containing very long lists, tables, paragraphs, or code block

v2.1.206

07/10 09:45

What's changed Added directory path suggestions to /cd, matching /add-dir behavior Added a /doctor check that proposes trimming checked-in CLAUDE.md files by cutting content Claude could derive from the codebase /commit-push-pr now auto-allows git push to the repo's configured push remote (remote.pu

v2.1.205

07/09 05:22

What's changed Added an auto mode rule that blocks tampering with session transcript files Fixed --json-schema silently producing unstructured output when the schema was invalid, and schemas using the format keyword being rejected Fixed a message sent while Claude was working being silently lost whe

v2.1.204

07/08 08:27

What's changed Fixed hook events not streaming during SessionStart hooks in headless sessions, which could cause remote workers to be idle-reaped mid-hook

v2.1.203

07/08 05:06

What's changed Added a warning when your login is about to expire, so you can re-authenticate before background sessions are interrupted Added a grey ⏸ badge to the footer when in manual permission mode, making the active mode always visible Added the session's additional working directories to MCP

v2.1.202

07/07 06:51

What's changed Added a "Dynamic workflow size" setting in /config for controlling how large Claude generally makes dynamic workflows (small/medium/large agent counts) — an advisory guideline, not an enforced cap Added workflow.run_id and workflow.name OpenTelemetry attributes to telemetry emitted by

v2.1.201

07/04 07:50

What's changed Claude Sonnet 5 sessions no longer use the mid-conversation system role for harness reminders