Meta 发布 Muse Spark 1.3，Intelligence Index 得 61-62 分逼近 Claude 与 GPT-5.6

Meta 发布 Muse Spark 1.3，为五个月内第四个 Muse Spark 版本。max 变体（合作伙伴限时预览）在 Artificial Analysis Intelligence Index 得 62 分。（AIHOT源：X：Kim (@kimmonismus)）

Meta 发布 Muse Spark 1.3，智能体与科学推理能力提升

Meta 发布 Muse Spark 1.3，是五个月内第四个 Muse Spark 版本。xhigh 版在 Artificial Analysis Intelligence Index 得 61 分。（AIHOT源：X：Alexandr Wang（Scale AI 创始人/Meta 首席 AI 官） (@alexandr_wang)）

Claude 在 Cowork 和 Claude Code 中支持后台操作电脑

Claude 官方宣布 Claude Cowork 和 Claude Code 新增后台使用电脑的能力。用户把任务交给 Claude 后，它会像人一样点击、输入和打开应用，用户可同时去做其他事。（AIHOT源：X：Claude (@claudeai)）

GitHub Copilot 如何在不牺牲任务质量的前提下降低 AI 编码成本

GitHub 工程师 Erik Kristensen 分享了 Copilot 降本的四项改动：选择性压缩工具输出、移除 view 工具行号前缀（线下推理成本降约 5%，线上用户日均推理成本降约 3%）、压缩 task-tool 提示词（每轮省约 1300 token，每活跃小时归一化成本降 2.9%）、后台任务完成后直接交付结果（AI Credits 用量降约 2.3%）。（AIHOT源：GitHub Blog）

美国司法部就 OpenAI 版权诉讼提交意见书支持训练属于合理使用

美国司法部在 OpenAI 与纽约时报的版权诉讼中提交意见书，主张用受版权文本训练 LLM 一般应属合理使用，称模型训练具有非凡转换性，并以国家安全为由警告全面许可要求会削弱美国 AI 开发者竞争力。该意见书属建议性质、不约束法院，仍将数据获取方式与具体输出是否复制受保护段落作为独立问题留给法院逐案判断。（AIHOT源：X：Rohan Paul (@rohanpaul_ai)）

Anthropic 发布电商 Agent 架构与生产实践指南，并开源 commerce-agents 参考实现

Anthropic 发布电商 Agent 构建指南，基于与零售、旅游、电信等团队的落地经验，核心架构是单个 Claude 在标准 Agent 循环中配合技能与工具，而非按领域拆分子智能体，并开源了 anthropics/commerce-agents 参考实现，含购物与商家 Agent。（AIHOT源：Claude：Blog（网页））

Google AI 团队分享如何为 LLM-as-a-Judge 评测编写可靠的评分标准

Google AI 团队发布教程，讲解如何为 LLM-as-a-Judge 评测编写可靠的布尔式评分标准，指出模糊提示会导致评估不一致和浪费 token。文中给出四条经验：问题保持原子化且互不重叠、只让评判模型评估客观事实（可用 RFC 2119 术语如 MUST 表述）、只评 prompt 中明确要求的内容、用专家标注的 golden set 校准评判模型直至与人类评分一致。（AIHOT源：Google AI：DEV 作者专属（RSS））

Google 总结 AI Agents Challenge 中最强提交背后的 4 个工程模式

Google 复盘 AI Agents Challenge 赛事，从各赛道头部提交中提炼出四个工程模式：双向 MCP、事件驱动并发、同标准回退和分层路由。（AIHOT源：Google Developers Blog（RSS））

Google DeepMind 发布 Gemini 3.8 Flash 与 3.8 Flash Cyber 两款新模型

Google DeepMind 发布 Gemini 3.8 Flash 与 3.8 Flash Cyber。（AIHOT源：Google DeepMind：Blog（RSS））

什么是 harness 工程？Google 用 ADK 2.0 与 Antigravity SDK 演示自动修复编码循环

Google 员工 Shir Meir Lador 介绍 harness 工程，即用确定性组件包裹 LLM，包括编排层、执行沙箱、状态持久化和验证工具，让 Agent 不需逐行人工审查即可安全生成代码。（AIHOT源：Google AI：DEV 作者专属（RSS））

OpenAI 因 Tumbler Ridge 枪击案面临 30 起新诉讼，被指协助教唆

OpenAI 及 CEO Sam Altman 面临 30 起新诉讼，指控其为加拿大 Tumbler Ridge 校园枪击案嫌疑人提供实质性协助与鼓励，诉讼由事发时在校的学生、教师和校长于加州联邦法院提起。（AIHOT源：The Verge：AI（RSS））

Cursor 推出 Self-Hosted Machines，云智能体可在企业自有机器上执行

Cursor 发布 Self-Hosted Machines，让云智能体的工具执行迁移到企业自有网络内的机器，智能体循环、推理和规划仍留在 Cursor 云端，通过 worker 的出站 HTTPS 连接对接，Cursor 不会主动连入企业网络。（AIHOT源：Cursor Blog）

What do you build when you can build anything?

Execution Trees, Not More Logs: A Better Debugging Model for AI Agents

Designing a Webhook Delivery System for 10 Million Events a Day

My AI Gateway Added 400ms to Every Request. Here's Where It Went

I Tried Pair Programming With Three Different AI Tools For a Month

Sam Altman: Over the summer, we have been sprinting on safety priorities; it's more impor...

Over the summer, we have been sprinting on safety priorities; it's more important than ever for capabilities and safeguards to advance together. We have more to do but have made a lot of progress. We are also going to be launching our next model soon. There is an obvious tension here: on one hand, Astra is very good and we are excited to see what people will build with it. We are proud of our work. On the other hand, we are clearly in a phase of development where we believe caution is warranted, and we are pacing our progress to ensure that we can meet the safety standards required by new capability levels. Astra has been done training for a while now and is a significant step forward in both capabilities and alignment. For the models after that, we have been slowing things as needed to ensure that we can do sufficient work on safety and alignment. AI is getting extremely capable; no one fully understands the consequences of this. Managing the transition to a world with abundant and powerful AI to optimize for safety and benefits to people should be one of the highest priorities in the world. It is our highest priority at OpenAI. We have been living with the tension between being excited and anxious about progress for some time, and it is still discordant for us. We know it is much more discordant for other people. And yet, we believe strongly that the world needs to understand where AI is going and how models perform in the real world. More importantly, we believe the world will need aligned AI to manage the future phases of this transition. An iterative loop where society and this technology evolve together is what will lead to the highest chance of getting this right. So we hope you enjoy our new model, and we hope the world continues to take what’s happening in AI extremely seriously.｜@sama · 热度 10641 · 2026-09-01

Claude: Claude Fable 5.1 is available everywhere today

Claude Fable 5.1 is available everywhere today. Claude Mythos 5.1, our model for cyberdefenders and life scientists, is available through trusted access programs. Read more: https://t.co/nfkH5hyDyO｜@claudeai · 热度 2404 · 2026-09-01

Peter Yang: If you're trying out Fable 5.1 I highly recommend running: /claude-api promp...

If you're trying out Fable 5.1 I highly recommend running: /claude-api prompt-audit on your skills. It finds a bunch of redundancies and rules to remove for the latest models. Running it for all my skills now.｜@petergyang · 热度 1694 · 2026-09-02

Thariq: it’s a very good model, I spent a lot of time diving into it- longer write up...

it’s a very good model, I spent a lot of time diving into it- longer write up coming but try it on lower effort for tasks that need less verification or have fewer edge cases, switching effort no longer breaks prompt cache as well https://t.co/MCGYQTnR0X｜@trq212 · 热度 1642 · 2026-09-02

Boris Cherny: We have also reduced prices for Enterprise, API, and SDK customers

We have also reduced prices for Enterprise, API, and SDK customers. Cache reads on Fable 5.1 are now $0.25 per million tokens (previously: $1). Up to 38% cheaper for a typical Claude Code session.｜@bcherny · 热度 411 · 2026-09-01

Cat Wu: With Fable 5.1, our team has taken on more ambitious projects that would have...

With Fable 5.1, our team has taken on more ambitious projects that would have previously taken months. What big bets do you want to take? Ask Fable 5.1 in Claude Code, Claude Cowork, Claude Tag to take on the task and let us know what you think! https://t.co/LCloUBWKeO｜@_catwu · 热度 397 · 2026-09-01

Guillermo Rauch: Fable 5.1 now on @vercel AI Gateway Try it with https://t.co/OL0LzGtvAw

Fable 5.1 now on @vercel AI Gateway Try it with https://t.co/OL0LzGtvAw. Every model at your fingertips 🥹｜@rauchg · 热度 389 · 2026-09-01

Boris Cherny: We have been working on how often safeguards intervene

We have been working on how often safeguards intervene. Our latest biology safeguards intervene on benign requests 85% less often than the ones we shipped with Fable 5, and Claude Code users should see around 60% fewer cyber interventions per session. Expect more improvements soon.｜@bcherny · 热度 343 · 2026-09-01

Aaron Levie: At Box, we’ve been testing Fable 5.1 in early release against our complex ent...

At Box, we’ve been testing Fable 5.1 in early release against our complex enterprise work eval. Fable 5.1 delivers a huge 7 percentage point jump over Fable 5 for unstructured data tasks in the enterprise. On this updated test, we use the Box Agent with Fable 5.1 to work through a wide range of real world enterprise scenarios with documents in financial services, life sciences, the public sector, and more industries. All of these tasks require a high degree of analytical, math, logic, and domain knowledge to perform successfully. Here are a few examples of wins: * Financial Services (+17% improvement): on a tax-adjusted profit projection, Fable 5.1 correctly applies capital allowances before computing tax liability; this is a subtle ordering that Fable 5 misses, producing wrong figures all the way through retained earnings. * Technology (+37% improvement): on a cost-optimization analysis where a key metric's normalization is ambiguous, Fable 5.1 recognized the ambiguity, computed both forms, and presented the correct one, where Fable 5 committed to the wrong normalization. * Public Sector (+16% improvement): on an educational data analysis task, Fable 5.1 works through the full weighted-mean methodology and produces correct rankings, while Fable 5 miscategorizes one item and cascades errors through the whole sheet. These are just a few of the wins we saw. Overall major jump in capability for long running agentic workflows in the enterprise. Fable 5.1 will be available shortly in the Box AI Studio for building custom AI agents with enterprise content.｜@levie · 热度 334 · 2026-09-01

Boris Cherny: Last but not least, Fable 5.1 writes better and has better tone

Last but not least, Fable 5.1 writes better and has better tone. We heard your feedback, and are actively working on reducing Claude-speak. Solid progress with 5.1, more to come.｜@bcherny · 热度 330 · 2026-09-01

Alex Albert: One non-model thing I'm very excited we announced today with Fable 5.1 is Ent...

One non-model thing I'm very excited we announced today with Fable 5.1 is Enterprise Frontier Safeguards (EFS). Even if you're not a CEO/CISO, this blog post is a look into how big companies are thinking about AI adoption which is important to understand if you're interested in how AI might diffuse across industry. Companies are giving agents more and more access to internal systems. With traditional zero data retention, there's been no way to spot patterns in what agents might be doing across sessions on these systems. EFS is ZDR++. It's ZDR built for a world with agents. A company's data stays in its own cloud, and an automated monitoring layer we built flags risky patterns to its team. I think of this as an observability and risk mitigation layer for agents. My guess is it will become a standard requirement from enterprises for running increasingly powerful AI on their systems.｜@alexalbert__ · 热度 284 · 2026-09-01

Madhu Guru: Self-improving products is a meta idea every company should be implementing

Self-improving products is a meta idea every company should be implementing. Some of the core pieces you will need: 1/ a crisp definition of the various metrics you care about : primary, secondary, guardrail 2/ an articulation of your strategy, roadmap and metrics you care about moving the most 3/ knowledge base of past product decisions and principles so the agent knows how decisions get made 4/ connections to your to your internal product systems: dashboards, APIs, MCPs, and tools your agent needs to observe and act on your product. 5/ a harness built with knowledge of your end to end product development flow. Depending on the complexity of your product, you might be able to prototype with fairly simple versions of each of the above and reuse existing swe agents.｜@realmadhuguru · 热度 279 · 2026-09-01

DietrichGebert/ponytail

Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.｜JavaScript · ⭐121381 · Search 增补 · 近快照 +1508星 · 创建 2026-06-12

stablyai/orca

Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and VPS.｜TypeScript · ⭐59967 · Search 增补 · 近快照 +763星 · 创建 2026-03-17

Graphify-Labs/graphify

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.｜Python · ⭐113988 · Search 增补 · 近快照 +597星 · 创建 2026-04-03

zhaoxuya520/reverse-skill

Reverse Engineering / Authorized Penetration Testing / Security Research Skill Router Pack AI-powered routing + On-demand toolchain bootstrapping + Self-evolving knowledge base Supports Claude Code, Kiro, Cursor, Cline, and other AI coding clients 逆向/渗透/安全技能路由包 - AI 自动路由 + 按需自举工具链 + 自动进化经验库 | 支持 Claude Code / Kiro / Cursor / Cline 等代码 AI 客户端｜PowerShell · ⭐34108 · Search 增补 · 近快照 +362星 · 创建 2026-05-13

calesthio/OpenMontage

World's first open-source, agentic video production system. 12 production pipelines, 100+ tools, 700+ agent skill and production-knowledge files. Turn your AI coding assistant into a full video production studio.｜Python · ⭐55660 · Search 增补 · 近快照 +651星 · 创建 2026-03-29

Gemini 3.8 Flash and 3.8 Flash Cyber

https://deepmind.google/models/model-cards/gemini-3-8-flash/

Can I opt out of my input or output data being used for training?

Commodore 64 released September 1, 1982

Muse Spark 1.3

Three sites made 215,128 “best software” pages for AI. Perplexity cites them

华为 CMM-AL10 鸿蒙手机入网：配备 6615mAh 大电池

IT之家 9 月 3 日消息，型号为 CMM-AL10 的华为 5G 新机已通过电信设备进网许可，未公布证件照。 IT之家注意到，该机配备 6615mAh 额定容量电池，搭载 HarmonyOS 鸿蒙操作系统，前后共 6 颗摄像头。 此外，本机支持 GSM、WCDMA、LTEFDD、TDLTE 网络，支持红外、蓝牙、USB、星闪连接方式。 更多详细参数信息如下：

博通 2026 财年第三财季归母净利润 130.88 亿美元，同比增长 216%

IT之家 9 月 3 日消息，博通今日发布 2026 财年（2025 年 11 月 3 日～2026 年 11 月 1 日）第三财季（2026 年 5 月 4 日～2026 年 8 月 2 日）报告： 营业总收入 ：295.91 亿美元 （IT之家注：现汇率约合 1993.52 亿元人民币） ，同比增长 86%，环比增长 33.37% 毛利润 ：204.56 亿美元 （现汇率约合 1378.1 亿元人民币） ，同比增长 91.12% 毛利率 ：69.13%，同比增长 2.03 个百分点 归母净利润 ：130.88 亿美元 （现汇率约合 881.73 亿元人...

慧与 2026 财年半年报归母净利润 25.29 亿美元

IT之家 9 月 3 日消息，慧与今日发布 2026 财年（2025年11月1日～2026年10月31日）半年报（2025年11月1日～2026年7月31日）及第三财季（2026年5月1日～2026年7月31日）报告： 2026 财年半年报 营业总收入 ：321.92 亿美元 （IT之家注：现汇率约合 2168.75 亿元人民币） ，同比增长 30.8% 毛利润 ：121.39 亿美元 （现汇率约合 817.79 亿元人民币） 毛利率 ：37.7% 归母净利润 ：25.29 亿美元 （现汇率约合 170.38 ...

Nuxt 4.5：实验性 SSR 流式传输、Vite 8 以及基于 Rsbuild 的 Rspack 构建器

点击查看原文>

当 AI 开始重写负载，数据库该如何重新设计？

点击查看原文>

Java 近期新闻：JDK 27-RC1、OpenJDK JEPs、Jakarta EE、BellSoft、Helidon、Micrometer、Tika 4.0

点击查看原文>

Normalized Fascism in Open Source: $12 Million Given to DHH

A note on subscription prices from LWN

Dependent if expressions without dependent types

Static Allocation, Constant Work

I Don’t Have a Smartphone…

Facilitating AI integration with simplicity at scale

As companies scale, the technology supporting operations can become a liability just as quickly as it becomes an asset. Disconnected systems, site-specific tools, spreadsheets, and manual workarounds can create data silos that make it harder to spot problems early, coordinate responses, and make ...

The Download: AI puzzles and a path to our nearest star system

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. AI models flub these intelligence tests. Can you fare any better? Puzzles and games have always been central to AI development. The term “machine learning” was...

How AI plotted an interstellar journey to Alpha Centauri

A nonprofit organization called the Fermi Explorer Mission announced today that it intends to launch a spacecraft to our nearest star system by the end of 2029.  It’s a hugely ambitious mission—if all goes well, the spacecraft could take up to 80,000 years to arrive at Alpha Centauri, which is 4....

TechCrunch Disrupt 2026’s new Real World AI Stage features Nvidia, robots, and extinct animals

On our new Real World AI stage, we’ll be focusing on the intersection between the digital and physical, and all the ways we’ll continue to see a blending of the two.

Google spared from ad-business breakup, but judge orders changes to how it operates

Google has dodged an effort to break up its ad business, but a judge said Wednesday that the company will need to adjust its business to benefit competitors.

OpenAI’s new reasoning technique alarms AI safety experts

OpenAI’s new Astra model will use “recurrent depth,” a technique that allows the model to operate outside of the sequential thinking that characterizes most reasoning models.

Google says its new Gemini 3.8 Flash model ‘works harder’ but might cost more

Google launched Gemini 3.8 Flash, arriving just a few weeks after its predecessor. The company claims the new model "works harder" than Gemini 3.7 Flash by performing more reasoning steps on complex tasks and "calling tools iteratively." It has the same introductory pricing as 3.7 Flash, $0.75 pe...

1Password wades into a right-wing mess after funding a Linux project

1Password faced immediate backlash from customers this week over a $300,000 pledge in support of a Linux distro created by David Heinemeier Hansson, who has regularly published overtly racist blog posts that include comments calling for deportation of ethnic minorities in Europe. The popular pass...

折叠屏 iPhone、新款 HomePod…… Apple 发布会有哪些看点？

上周，Apple公布了今年秋季发布会的时间和主题，这场名为「Surpriseandshine」的发布会将于北京时间9月10日凌晨一点在ApplePark举行。与过去两年的秋季发布会不同，今年的「科技春 ... 查看全文

一日一技 | 借助 PowerShell 自定义 Windows 更新暂停时间

Windows 下的累计更新会修复大量的安全漏洞，虽然可以暂停更新但是最长只能暂停 35 天，35 天后 Windows 电脑就要强行帮你更新了，而本文介绍了一种突破暂停最大时间的方法。 查看全文

众测招募 | 玩转音乐生态，体验 Sonos Play 便携智能音响

近年来，智能音响正在从「放在客厅的固定设备」变为「跟随你移动的声音系统」。用户不再满足于单一房间的聆听，而是希望音乐能随着自己的脚步，从客厅到厨房、从书房到户外，无缝流转、不断联。这正是Sonos一直 ... 查看全文

react-bits：从 36K stars 的“酷炫组件”，看动效如何成为 React 的可复用能力

# react-bits：从 36K stars 的“酷炫组件”，看动效如何成为 React 的可复用能力 打开一个视觉表现力强的网页，人们最先注意到的往往不是组件边界、状态管理或样式组织，而是“这

【AI】大模型本地部署与量化：Ollama、transformers、llama.cpp实践

部署即在本地电脑中下载并运行模型，就像使用网络上的大模型API一样，但区别在于模型是运行在本地的，不收费也不会泄露信息。但模型可能很大，本地电脑可能会内存不足，这时候就需要量化来尝试缩小模型存储空间，

PDF Viewer KMP（基于 chrome 的 PDFium 内核）

PDF Viewer KMP 是基于 PDFium 的 Kotlin Multiplatform PDF 库，可用于读取、渲染和显示 PDF 文档。

大宗商品综述：油价高位企稳 铜价下跌 金价上涨

油价企稳，接近五周来最高收盘水平，此前美国与伊朗爆发冲突，再次对中东能源出口构成威胁。铜价跌至逾一周来最低水平，因为油价再度上涨加剧了市场对全球经济前景的担忧。...

9月3日隔夜要闻：美股收高 戴尔大涨 油价高位企稳 金价上涨 特朗普提议将霍尔木兹海峡更名为“特朗普海峡”

市场 9月3日美股收盘：三大股指普涨 英伟达涨超3% 9月3日美股成交额前20：戴尔大涨15% 大幅上调全年营收预期 9月3日热门中概股涨跌不一 网易有道跌超9% 蔚来跌超4% 大宗商...

美法官叫停特朗普政府关于“出生公民权”的最新行政令

当地时间9月2日，美国马里兰州联邦地区法院法官德博拉·博德曼发布初步禁令，阻止特朗普政府执行一项新的行政令。该行政令旨在限制有资格获得“出生公民权”的人数。

博通本财季营收指引令投资者失望 股价盘后重挫

博通本财季营收展望低于分析师预估，显示人工智能（AI）投资热潮给这家芯片巨头带来的业绩提振力度不及一些投资者的预期。

原油：油价在高位企稳 美伊冲突持续

油价企稳，接近五周来最高收盘水平，此前美国与伊朗爆发冲突，再次对中东能源出口构成威胁。 WTI小幅上涨，收于每桶90美元上方，连续第三个交易日上涨。

SkyProduction天工工作台：从剧本到成片，一套工作台把精品短剧创作真正跑起来

SkyProduction（天工工作台）全新版本于8月31日正式上线

还在为大模型洗数据熬夜？蚂蚁拿下VLDB工业最佳论文，一套宽表搞定35PB语料，效率狂飙5.6倍

蚂蚁集团推出统一宽表系统OmniTable，论文获评VLDB 2026工业赛道最佳论文

阿里更新旗舰模型Qwen3.8-Max，前端编程能力跃居全球第一