JavaScript · ★ 16,250 · 🍴 888 · 📈 3,162 stars today
A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings
中文介绍 为编码 AI agent 提供的安全审计技能，把审计拆成多个阶段执行，最终产出可机器读取的发现结果并支持独立验证，减少模型自说自话的误报。适合在代码评审或 CI 流程中让 agent 自动排查依赖、配置与代码漏洞。
HTML · ★ 24,372 · 🍴 1,677 · 📈 1,124 stars today
Scale computer-use 2.0 with open-source drivers, cross-OS fleets, and benchmarks for training, evaluation, and data generation.
中文介绍 面向 computer-use 智能体的开源基础设施：提供跨操作系统驱动、可批量调度的设备集群，以及用于训练、评估和数据生成的基准测试，帮助团队规模化地让模型操控真实电脑与桌面环境。
JavaScript · ★ 96,999 · 🍴 10,237 · 📈 547 stars today
Production-grade engineering skills for AI coding agents.
中文介绍 为 AI 编码代理准备的生产级工程技能集合，把代码审查、测试、重构等成熟工程实践封装成可直接调用的技能，让 agent 输出更贴近真实团队规范。适合给 Claude Code、Cursor 等代理扩展能力。
Go · ★ 15,605 · 🍴 1,521 · 📈 406 stars today
Secure environments for developers and their agents
中文介绍 为开发者和他们使用的 AI agent 提供安全隔离的开发环境。用声明式模板定义 workspace，按需在自有云或本地基础设施上创建并回收，便于统一环境、管控密钥与权限，适合中大型工程团队。
TypeScript · ★ 146,689 · 🍴 23,919 · 📈 482 stars today
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
中文介绍 Anthropic 官方的终端内 agentic 编码工具，能理解整个代码库，执行跑测试、改代码等常规任务，解释复杂逻辑并处理 git 工作流。适合习惯命令行、希望把重复编码工作交给 AI 的开发者。
TypeScript · ★ 16,005 · 🍴 2,082 · 📈 477 stars today
OpenStock is an open-source alternative to expensive market platforms. Track real-time prices, set personalized alerts, and explore detailed company insights — built openly, for everyone, forever free.
中文介绍 开源行情与研究平台，用于替代价格高昂的商业终端：实时跟踪价格、设置个性化提醒、查看公司详细数据与洞察，面向个人投资者和开发者，完全开放且永久免费。
Jupyter Notebook · ★ 4,933 · 🍴 910 · 📈 314 stars today
Fault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters
中文介绍 面向超大规模模型训练的 GPU 编排与机器学习框架，强调容错与高可扩展性，支持数十亿到数万亿参数级别的模型训练，适合需要自建集群、长时间稳定跑大模型训练的团队。
Python · ★ 67,017 · 🍴 4,833 · 📈 94 stars today
Get your documents ready for gen AI
中文介绍 把 PDF、Office 等各类文档解析成适合生成式 AI 使用的结构化数据，输出 Markdown 或 JSON，保留版面、表格与阅读顺序。常用于构建 RAG 知识库和数据抽取流水线，属于文档预处理环节的工具。
Rust · ★ 12,008 · 🍴 1,115 · 📈 84 stars today
🥧 Savoury implementation of the QUIC transport protocol and HTTP/3
中文介绍 Cloudflare 出品的 QUIC 传输协议与 HTTP/3 的 Rust 实现，既可作为库集成进服务端和客户端，也能编译成独立服务器。适合想实践新一代传输协议或部署高性能网络服务的开发者。
Go · ★ 5,216 · 🍴 217 · 📈 430 stars today
Your own search engine
中文介绍 由 SearX 作者发起的自托管搜索项目，目标是让搜索索引掌握在自己手里：本地建立索引并自行部署查询，减少对第三方搜索服务的依赖。适合注重隐私、想搭建个人搜索引擎的用户。
★ 103,109 · 🍴 4,424 · 📈 151 stars today
科技爱好者周刊，每周五发布
中文介绍 阮一峰维护的科技爱好者周刊，每周五更新，汇总当周值得关注的科技新闻、开源项目、工具与文章，并附简短点评。读者主要是中文技术圈的开发者、产品与科技从业者。
TypeScript · ★ 37,769 · 🍴 5,205 · 📈 31 stars today
人人都能用英语
中文介绍 《人人都能用英语》的开源电子书项目，讨论普通人如何靠持续输入与真实使用把英语用起来，而非应付考试，内容覆盖听说读写方法与学习路径，适合想自学英语的成年读者。
Python · ★ 25,114 · 🍴 2,993 · 📈 280 stars today
Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork
中文介绍 Anthropic 开源的知识工作者插件集合，主要面向在 Claude Cowork 中办公的用户，把写文档、整理资料、分析数据等常见知识工作流程封装成可复用插件，降低非技术用户的配置成本。
Python · ★ 11,591 · 🍴 739 · 📈 207 stars today
Automation foundation model for tiny devices: 2-bit, 8-29 MB, tool calls, structured extraction and embeddings on phones, wearables, smart homes, robots, cars and microcontrollers.
中文介绍 面向微型设备的自动化基础模型：采用 2-bit 量化，体积仅 8–29MB，支持工具调用、结构化抽取与 embedding，可运行在手机、可穿戴设备、智能家居、机器人、汽车乃至微控制器上。
Rust · ★ 3,385 · 🍴 443 · 📈 59 stars today
OpenAI Codex 桌面端/CLI 的可视化管理工具，具有Provider/API 切换、会话同步、提示词注入、Skills/MCP 管理、TOML 配置可视化的跨平台工具。
中文介绍 OpenAI Codex 桌面端与 CLI 的可视化管理工具，支持 Provider/API 切换、会话同步、提示词注入、Skills 与 MCP 管理以及 TOML 配置可视化，跨平台运行，适合把 Codex 深度接入工作流并需统一管理配置的用户。
👍 95
The widespread adoption of long-horizon agents has made model workloads increasingly input-heavy. Although prior work has substantially reduced the cost of long-context computation, prefill remains computationally expensive, and large KV caches continue to strain HBM and SSD capacity and data-transf
中文介绍 面向长时程 agent 的输入密集型负载，prefill 仍然昂贵，KV cache 持续挤占 HBM 与 SSD 容量及带宽。DeepSeek-V4.1-Flash 进一步推进 KV cache 压缩极限，缓解长上下文推理的显存与存储压力。
👍 93
Recent Omni-Modal Generative Models (Omni-Models) have advanced content generation toward unified modeling of text, images, video, and audio. MiniMax-H3 exemplifies this transition by combining multimodal context understanding with joint audio-visual generation in a shared latent framework. Its unif
中文介绍 评估统一建模文本、图像、视频与音频的 Omni-Modal 生成模型 MiniMax-H3 能否真正推理物理世界。该模型在共享潜空间中结合多模态上下文理解与音视频联合生成，论文围绕其物理世界推理能力展开系统性评测。
👍 75
We study length inflation in on-policy distillation (OPD), where student responses can become excessively long and even exhaust the generation budget. We identify termination-token mismatch between base students and post-trained teachers as an important source of this behavior. Across Qwen3, Llama,
中文介绍 研究 on-policy distillation（OPD）中的回答长度膨胀现象：学生输出过长，甚至耗尽生成预算。作者发现基座学生与后训练教师之间的终止 token（EOS）不匹配是重要成因，并在 Qwen3、Llama 等模型上验证。
👍 74
Scientific code repositories encode decades of human knowledge in executable models, methods, and tools. Yet fragmented toolchains, implicit domain conventions, and specialized correctness criteria make this knowledge difficult to convert into reliable learning experience-a challenge we call the sci
中文介绍 科学代码库以可执行模型、方法与工具承载数十年知识，但工具链碎片化、领域约定隐式、正确性标准专门化，难以转化为可靠的学习经验。ScienceIDE 将全球科学代码库改造为 agent 可学习的环境，为科学智能体提供训练与评测基础。
👍 69
As coding agents move from supervised code completion to unattended, around-the-clock exploration, their work expands from isolated predictions into long trajectories of reasoning, tool use, and feedback. Token efficiency therefore becomes important for scaling recursive self-improvement. We take an
中文介绍 针对 coding agent 从有监督补全走向全天候无人探索、工作轨迹不断变长带来的 token 效率问题，提出 SoL-Pi，通过递归扩展 auto-research 循环提升 agent harness 效率，支撑递归式自我改进的可扩展性。
👍 55
Coding harnesses shape how autonomous coding agents translate model capabilities into long-horizon software-engineering performance, yet existing work typically evaluates harnesses as monolithic systems, leaving the effectiveness of individual components unclear. To enable component-level comparison
中文介绍 现有工作多把 coding harness 当作整体系统评估，难以判断各组件各自的贡献。本文对编码 agent 的 harness 设计开展实证研究，实现组件级对比，厘清各模块对长时程软件工程性能的实际作用。
👍 54
Reliable confidence estimation is increasingly central to the trustworthy deployment of language models: a calibrated estimate of the probability that an output is correct decides what to ship, what to escalate, and what to retry. Existing confidence estimators, however, share one design premise: th
中文介绍 可靠部署语言模型需要校准的置信度估计，以决定输出是否上线、升级或重试。已有估计器共享同一设计前提，忽视了模型在交互中积累的经验。本文提出经验性置信度估计，覆盖从推理到 agent 的场景。
👍 48
Coding agents are typically evaluated with desired behavior specified through issues or instructions. In practical web development, however, agents may need to infer behavior from working software and implement it in an incomplete application. We introduce ProgramDistill, a benchmark evaluating codi
中文介绍 提出 ProgramDistill 基准，评估编码 agent 能否从可运行的 web 应用中推断行为，并在不完整的应用里将其实现，从而把交互式网页转化为可验证、参考引导的 SWE 任务，贴近真实 web 开发场景。
👍 43
Autonomous research loops such as AutoResearch show that one coding agent can improve a training setup unattended. Run several of them and each session starts from scratch, so more agents tend to mean more duplicated search rather than more discovery. Agora is a shared memory for such agents: resear
中文介绍 AutoResearch 等自主研究循环能让单个编码 agent 无人值守地改进训练设置，但多个循环并行时各自从头开始，agent 越多反而重复搜索越多。Agora 以 Git 作为共享记忆，使研究进展在 agent 之间累积复用。
👍 40
World modeling enables intelligence to anticipate consequences, guide interventions, and learn from interaction. Yet predictive models remain domain-specific: can a common learning principle support world modeling across radically different systems? We introduce JEPA-Anything, a domain-agnostic fram
中文介绍 世界模型让智能体预判后果、指导干预并从交互中学习，但现有预测模型多为特定领域定制。JEPA-Anything 提出领域无关框架，尝试以共同的学习原则在差异巨大的系统之间训练预测模型，实现跨领域世界建模。
👍 33
LLM assistants are widely used for daily social advice, yet evaluating their social reasoning in such consultation settings remains challenging since (i) it requires setups where the assistant learns about social situations from subjective user narratives, and (ii) social properties, such as others'
中文介绍 LLM 助手常被用于日常社交建议，但其社会推理难以评估：助手需从主观的用户叙述中理解情境，还涉及他人隐私等社会属性。本文提出可验证的社会推理评测方案，用于检验此类咨询场景下的助手表现。
👍 32
Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment. This limits their ability to adapt how they gather evidence, invoke tools, verify signal
中文介绍 LLM 交易 agent 可整合行情、新闻与可执行分析，但其工具使用策略多为部署前写死的静态规则，难以自适应地收集证据、调用工具与验证信号。EvolveTrade 以经验驱动的策略精炼，让交易 agent 自我演化。
👍 29
Platform abuse campaigns conceal redirection instructions with emojis, homophones, character decomposition, and redundant symbols, then route users through disguised links to services associated with pornography, fraud, gambling, or illicit transactions. Existing benchmarks evaluate obfuscated text
中文介绍 平台滥用活动借助 emoji、谐音、拆字与冗余符号隐藏跳转指令，再经伪装链接导流至色情、诈骗、赌博等非法服务。RiskChainBench 针对混淆平台消息还原与基于证据的 web 调查构建基准，弥补现有混淆文本评测的不足。
👍 26
Information retrieval is increasingly important as LLM agents tackle complex tasks involving diverse information needs. Because retrieval relies on an index that represents each document through index keys, retrieval quality depends heavily on how effectively these keys expose the knowledge containe
中文介绍 检索质量取决于索引键能否充分暴露文档所载知识。本文提出自演化检索索引（Self-Evolving Search Index），让索引表示随检索需求迭代更新，以更好支撑 LLM agent 处理信息需求多样的复杂任务。
👍 26
Multi-turn agents trained with reinforcement learning (RL) receive a single scalar reward per trajectory, which motivates self on-policy distillation (OPD) to supply dense token-level supervision from a self-teacher with privileged task skills, letting a skill-free student internalize them. This rec
中文介绍 多轮 agent 的 RL 训练每条轨迹仅得一个标量奖励，self on-policy distillation 借具备特权任务技能的教师提供 token 级密集监督，使无技能学生内化这些能力。RetireOPD 提出自退休的 on-policy distillation 用于 agentic RL。
👍 26
Scientific agents contribute to hypothesis discovery by synthesizing evidence, assessing proposals, and developing new explanations. Recent systems combine scientific agents with evolutionary search through critique, comparison, and revision. However, how different forms of agent collaboration affec
中文介绍 科学 agent 通过整合证据、评估提案来发现假设，近期系统将其与演化搜索结合，经批评、比较与修订迭代。HypoEvolve 用遗传算法组织多 agent LLM 协作，考察不同协作形式对科学假设发现效果的影响。
👍 25
GUI agents execute long-horizon tasks on dynamic graphical user interfaces, where pop-ups, delayed loads, and relocated widgets routinely invalidate plans fixed before execution. Recent agent-skill frameworks encapsulate reusable procedural knowledge to mitigate this, yet existing skill designs are
中文介绍 GUI agent 在动态界面上执行长时程任务，弹窗、延迟加载与控件移位常使预先制定的计划失效。Reflect, Revise, Reuse 提出免训练的技能演化方法，通过反思、修订与复用持续更新可复用的程序性知识。
👍 23
Document parsing converts document images into structured content and requires reliable performance across diverse layouts and acquisition conditions. Yet training corpora are biased toward common document types and clean digital pages, while expanding coverage alone does not specify how to address
中文介绍 文档解析需把文档图像转为结构化内容，并在多样版式与采集条件下保持可靠；但训练语料偏向常见文档类型与干净数字页，仅扩大覆盖并不足够。WeVisDoc 从覆盖率转向能力建设，提升端到端文档解析鲁棒性。
👍 22
Direct preference alignment methods are widely used to align large language models (LLMs) with human preferences because of their computational and memory efficiency. However, likelihood displacement motivates alternative ways to extract information from preference pairs with small likelihood margin
中文介绍 直接偏好对齐方法因计算与内存高效而被广泛用于 LLM 与人类偏好对齐，但似然位移问题促使研究者寻找在似然差距较小时仍能提取偏好信息的新途径。本文提出面向 LLM 偏好对齐的零阶优化范式。
👍 20
Spatial intelligence requires more than describing object locations. Under incomplete observation, models must identify and acquire missing evidence, interpret it in a common spatial frame, and act on it. We introduce VA-Bench to evaluate the complete observe-reason-act-revise loop. General-purpose
中文介绍 空间智能不止于描述物体位置：在观测不完整时，模型需识别并获取缺失证据、在统一空间框架中解读并据此行动。VABench 通过视觉演示、主动感知与度量控制，评测「观测—推理—行动—修正」完整闭环。
@sandylikesfrogs · 3.9K 粉丝 · 367.0K 阅 · 512 赞 · 36 转
What we learned making our first $75 with an AI-run business. In July, we decided to see if an AI agent could run a company that could actually make money. We gave Devin a Ramp card, a phone number
中文介绍 AI agent 能否独立运营一家能赚钱的公司？博主 7 月做实验，给 Devin 配上 Ramp 银行卡和电话号码，跑出首个 75 美元营收，并总结了过程中的经验与踩坑。
@sairahul1 · 145.3K 粉丝 · 330.3K 阅 · 568 赞 · 59 转
You do not need a 70B model. You do not need $100,000 in compute. You do not need a machine learning team. A 1.5B parameter model fine-tuned on 200-500 good examples can outperform a frontier model
中文介绍 小模型微调完整指南：不需要 70B 模型、10 万美元算力或 ML 团队。用 200-500 条优质样本微调 1.5B 参数模型，就能超过前沿大模型，附完整操作流程。
@Sabrina_Ramonov · 7.1K 粉丝 · 160.5K 阅 · 518 赞 · 398 转
I talk to business owners who have piles of photos and videos, but no time to post to social media. They don't know what caption to write. They don't know how to edit videos. So, all that good
中文介绍 入门教程：用 GPT-6 Astra 搭配 DaVinci MCP，帮囤了大量照片视频却没时间发社媒的商家自动写文案、剪视频，解决不会配文、不会剪辑的痛点。
@istoica05 · 9.6K 粉丝 · 150.4K 阅 · 501 赞 · 72 转
An AI agent was asked to speed up a key-value store. It delivered a 6× throughput gain and passed every correctness test. The explanation? The agent discovered that the industry-standard benchmark
中文介绍 AI agent 把 key-value store 吞吐提升 6 倍且通过全部正确性测试，但原因是它钻了行业标准 benchmark 的空子。博主借此指出 agentic 软件工程的两个关键缺口。
@chddaniel · 27.7K 粉丝 · 121.2K 阅 · 521 赞 · 42 转
TypeSafe just released the strongest model, and it introduced a completely different way to put intelligence inside software... Jev is in a category of its own right now, not because it writes better
中文介绍 Jev 完整上手指南：TypeSafe 刚发布这一最强模型，它提供了把智能嵌入软件的全新方式。博主认为它自成一类，不是因为写得更好，而是范式不同。
@mvanhorn · 40.4K 粉丝 · 103.6K 阅 · 550 赞 · 37 转
Here is the tl;dr ELI5. Think AI multiple choice, not AI essay writing. It doesn't chat. It makes decisions your software can act on: "Spam or not?" "Which tool should this agent use?" "Does this need
中文介绍 科普帖：Jev 像 AI 做多选题而非写文章，它不聊天，只输出软件可直接执行的决策，比如「是否垃圾邮件」「该调哪个工具」。作者列出人们已经在用它做的 9 件事。
@chamath · 2.4M 粉丝 · 90.2K 阅 · 503 赞 · 63 转
Open-weight models have come within roughly four months of the best publicly evaluated closed frontier models, and the gap between them has only gotten more volatile. That is turning the open vs.
中文介绍 深度分析开源与闭源 AI 竞赛：开源权重模型已逼近公开评测中最强的闭源前沿模型，差距约 4 个月，且这一差距越来越不稳定，正在重塑竞争格局。
@sydneyrunkle · 10.6K 粉丝 · 88.0K 阅 · 556 赞 · 66 转
Agents run in a loop: an LLM decides what to do, a tool executes, a model evaluates the results, and then continues in that loop until the task is complete. Agents and LLMs were initially difficult to
中文介绍 实操分享：用 Jev 搭建 agent harness。agent 本质上是一个循环——LLM 决定做什么、工具执行、模型评估结果，然后继续循环直到任务完成。
@0xCodila · 18.6K 粉丝 · 63.4K 阅 · 521 赞 · 62 转
The Jevons Paradox (he's on picture) is a rule stating that an increase in the efficiency of a resource's use does not reduce, but rather increases, its overall consumption - That's the global
中文介绍 从零搭建 Jev 的 10 步工程路线图。作者从 Jevons 悖论切入——资源使用效率提升反而推高总消耗，并以此解释为什么需要给 AI 换一个「新大脑」。
@RohOnChain · 72.5K 粉丝 · 57.7K 阅 · 518 赞 · 54 转
I will break down the exact framework to build a millisecond speed 24/7 HFT trading system with Jev, along with the exact resources that helped me personally. Let's get straight to it. Bookmark This -
中文介绍 用 Jev 搭建 7x24 小时毫秒级 HFT 交易系统的完整框架，并附上作者本人用过的资源清单，偏工程实操向。
@akshay_pachaar · 286.8K 粉丝 · 41.6K 阅 · 578 赞 · 87 转
We have been using LLMs like a hammer for every AI problem, even simple decisions. Jev handles those decisions in milliseconds at a fraction of the cost. Let's understand how it works and where it
中文介绍 Jev 原理解析：过去大家把 LLM 当锤子，什么 AI 问题都往上套，连简单决策也不例外；Jev 能在毫秒级、以极低成本完成这类判断，博主拆解其工作方式与适用场景。
Imitation is the sincerest form of Flattery
中文介绍 Latent Space 的 AI 新闻汇总称，两天内出现了 6 个「Jev」的克隆项目，并评论说模仿是最真诚的奉承。除该数量与时间跨度外，原文未披露更多细节。
OpenAI introduces the Australian Youth Safety Blueprint, a six-pillar roadmap for safer AI experiences that protect and empower young people.
中文介绍 OpenAI 发布「澳大利亚青少年安全蓝图」，提出包含六大支柱的路线图，目标是让青少年获得更安全的 AI 体验，同时保护并赋能年轻用户。
On Wednesday, MIT Technology Review hosted a live Roundtables event for subscribers that asked the question everyone’s asking right now: Could AI really kill us all? But attendees had so many more questions than we had time to answer in the 30 minute session. So we asked our senior AI editor Will Do
中文介绍 MIT Technology Review 面向订阅者举办线上圆桌活动，主题为「AI 真的可能杀死我们所有人吗」。由于 30 分钟会程内问题远超可回答数量，该机构邀请资深记者另行回应读者提问。
a quiet day
中文介绍 Latent Space 的 AI 新闻汇总显示，当日为消息面平静的一天，未发生重大行业事件。
中文介绍 TLDR AI 于 2026 年 9 月 18 日的简报要点包括：Claude Projects 推出第二版，以及 Google 面向家庭场景的智能体产品。原文未提供更多细节。
Cooley built GO Public with ChatGPT Work to bring intelligence to the IPO process, helping lawyers surface issues earlier and focus judgment where it matters most.
中文介绍 律所 Cooley 借助 ChatGPT Work 打造 IPO 工具 GO Public，将智能能力引入上市流程，帮助律师更早发现潜在问题，并把判断力集中在最关键的环节。
A dash of cold water keeps the foomers away.
中文介绍 Latent Space 的 AI 新闻汇总带来一组「现实检验」：Yegge 关闭了 Gas Town，Databricks 的 Astra 成本上升 60%。该媒体称，一盆冷水有助于给过度乐观的预期降温。
OpenAI for Law brings frontier intelligence for law, custom firm workflows, connected legal data sources, and legal-grade controls for confidential client work.
中文介绍 OpenAI 推出面向法律行业的 Astra for Law，提供前沿智能能力、可定制的律所工作流、对接法律数据源，以及适用于机密客户业务的法务级管控措施。
中文介绍 TLDR AI 于 2026 年 9 月 17 日的简报要点包括：Claude 与 Cowork 合并、ChatGPT 推出赞助型智能体，以及所谓「harness tax」问题。原文未提供更多细节。
We sit down with AIUC’s CEO on their Series A!
中文介绍 Latent Space 对话 AIUC 首席执行官 Rune Kvist，主题为「为超级智能承保：为可被起诉的智能体提供支持」，访谈背景是该公司的 A 轮融资。
OpenAI shares a framework for tracking, investigating, and disclosing model misalignment, alongside six reports of unexpected or concerning model behavior.
中文介绍 OpenAI 公布一套用于追踪、调查和披露模型失准的框架，同时附带 6 份报告，记录了模型出现的意外或令人担忧的行为。
OpenAI and AARP are bringing free, hands-on ChatGPT workshops to 1,000 older adults across 10 U.S. cities to build practical AI skills safely.
中文介绍 OpenAI 与美国退休人员协会（AARP）合作，在美国 10 座城市为 1000 名老年人提供免费的 ChatGPT 实操工作坊，帮助其安全地掌握实用的 AI 技能。
Explore new AI-powered advertising experiences from OpenAI, including Sponsored Agents, tools for marketers, and integrations with HubSpot and Shopify.
中文介绍 OpenAI 介绍其 AI 广告新形态，包括赞助型智能体（Sponsored Agents）、面向营销人员的工具，以及与 HubSpot、Shopify 的集成。
The AI boom is becoming a materials challenge. As AI pushes computing into new territory, the materials behind that infrastructure are becoming just as crucial as the algorithms running on it. Semiconductors and data centers are approaching physical limits around performance, thermal management, ele
中文介绍 MIT Technology Review 指出，AI 热潮正演变为材料层面的挑战。随着算力不断扩张，支撑基础设施的材料与算法同等关键，而半导体和数据中心在性能等方面正逼近物理极限。
GPT-6 Astra helps Hex’s data agents turn answers into interactive visualizations that employees are proud to share.
中文介绍 数据分析公司 Hex 利用 GPT-6 Astra，使其数据智能体将分析结果转化为交互式可视化报告，便于员工分享。
美股方面，SPY 报 761.69、1 日 -0.12%，落在 SMA20（764.3）下方但仍在 SMA50（759.73）与 SMA200（716.28）之上，RSI14 49.2，MACD -0.617 低于信号线 0.5463，动量中性偏弱；QQQ 报 721.45、1 日 +0.63%，MACD 金叉且多头排列，RSI 55.9；AAPL RSI 64.3、距 52 周高 -2.45%；MSFT 的 MACD 6.8496 低于信号线 10.0143；NVDA 1 日 +1.34%。加密方面，恐慌贪婪指数 71（贪婪），总市值 2.77T USD、24h -3.33%，BTC 主导率 58.9%、ETH 11.6%；BTC 报 81279.17，RSI 64.4、多头排列但距 52 周高 -35.59%，ETH RSI 65.5，SOL 5 日 +8.18% 且 MACD 金叉。中概整体偏弱：BABA 虽 1 日 +4.33% 但均线空头排列、RSI 48，PDD RSI 39，JD RSI 34.8，0700.HK RSI 36.9 并接近 52 周低。商品外汇方面，黄金 4424.9 位于 SMA20（4492.09）与 SMA200（4553.35）下方、MACD 为负，WTI 原油 1 日 -5.72% 但均线仍多头排列，美元/人民币 RSI 18.1 超卖且 MACD 死叉，VIX 14.81 呈空头排列，10Y 美债收益率 5、RSI 68.8 接近 52 周高，DXY 100.22 多头排列。
现价 721.45，1 日 +0.63%、5 日 +0.92%，站上 SMA20（713.24）、SMA50（709.95）与 SMA200（662.6），均线呈多头排列；RSI14 为 55.9 处中性偏强区间，MACD 0.2176 上穿信号线 0.1465 形成金叉，距 52 周高 -3.63%，指标读数偏上行。
收益率报 5，1 日 +1.03%、5 日 +0.46%，位于 SMA20（4.83）、SMA50（4.72）、SMA200（4.4）上方呈多头排列；RSI14 68.8 逼近超买阈值，MACD 0.0858 高于信号线 0.0705，距 52 周高仅 -0.36%，上行读数显著。
现价 665.75，1 日 -2.43% 但 5 日 +2.73%，同时高于 SMA20（614.61）与 SMA200（624.16），趋势字段为中性；RSI14 65.8 偏高，MACD 24.0309 高于信号线 16.2327，但 signals 为空且距 52 周高 -15.27%，多空读数交错。
报 6.69，1 日 -0.28%、5 日 -0.29%，跌破 SMA20（6.71）、SMA50（6.74）与 SMA200（6.84）呈空头排列；RSI14 仅 18.1 处超卖区，MACD -0.0108 低于信号线 -0.0105 触发死叉，并逼近 52 周低，技术面偏下行。
报 419，1 日 -1.64%、5 日 -2.19%，位于 SMA20（437.42）、SMA50（452.5）与 SMA200（506.86）下方，呈空头排列；RSI14 36.9 偏弱，MACD -7.3325 低于信号线 -6.4129，距 52 周高 -38.65% 且接近 52 周低，指标指向偏下行。
VIX 恐慌指数
10Y 美债收益率 (%)
美元指数 DXY
S&P 500 ETF
Nasdaq 100 ETF
Apple
Microsoft
Nvidia
Alphabet
Tesla
Meta
Bitcoin
Ethereum
Solana
阿里巴巴 (BABA)
拼多多 (PDD)
京东 (JD)
腾讯控股 (0700.HK)
黄金期货
WTI 原油期货
美元 / 人民币
本报告仅基于公开行情数据计算的技术指标读数，属客观状态描述，不构成任何投资建议；过去走势不代表未来表现，指标信号亦可能快速反转，仅供技术指标解读参考。
Turkish FM highlights Turkiye's readiness to assist Saudi Arabia militarily, as Houthi attacks intensify in Yemen.
中文摘要 土耳其外长表示，在也门胡塞武装袭击不断升级之际，土耳其支持沙特阿拉伯的安全，并强调土方已准备好向沙特提供军事协助。
Follow live updates Get our breaking news email, free app or daily news podcast Albanese says Apple CEO praised Australian social media ban The prime minister, Anthony Albanese, is appearing on the ABC’s Insiders this morning, amid a visit to the US where he has met Apple CEO, Tim Cook. Well, it’s a
中文摘要 澳大利亚新闻直播：影子部长安格斯·泰勒承诺推行「澳大利亚史上最大规模移民削减之一」；维州绿党议员蒂姆·里德获各界悼念。总理阿尔巴尼斯称，苹果公司首席执行官赞赏澳方对社交媒体的禁令。
In text exchanges and conversations, multiple people who have worked for leading companies are sceptical of the warnings.
中文摘要 英国广播公司报道，多名曾在领先人工智能企业工作的人士，对「人工智能可能杀死所有人」这类警告持怀疑态度，他们在交谈中表达了不同看法。
The children of key politicians jailed 25 years ago without a trace in Eritrea want answers.
中文摘要 25年前，厄立特里亚一批重要政治人物被关押后下落不明。其子女表示甚至不知道父母是否还在世，要求当局给出答案。
Students spent years collecting evidence against their Indonesian principal, before falling in love.
中文摘要 印度尼西亚一群学生多年收集证据，指控其校长施暴并最终将其绳之以法；其中两人在过程中相恋，如今已结为夫妻。
Denmark and the US have reached an agreement after months of threats from Donald Trump to annex the territory.
中文摘要 丹麦与美国在特朗普数月来威胁吞并格陵兰之后达成协议，北约对此表示欢迎。特朗普称，该协议将使美国获得「永久安全控制权」。
In the eastern city of Xalapa, what started as a joke among friends over two decades ago has become a vibrant annual tradition that highlights creativity and absurdity with melons.
中文摘要 墨西哥东部城市哈拉帕举办「最英俊瓜果」评选。这项活动二十多年前由朋友间的玩笑开始，如今已成为一年一度的传统，以创意和荒诞为特色。
This was the first time an alert was sounded in Riyadh since an escalation between the Houthis in Yemen and the Saudis.
中文摘要 沙特领导的多国联军表示，其防御系统拦截了胡塞武装向利雅得发射的导弹。这是也门胡塞武装与沙特之间冲突升级以来，利雅得首次拉响警报。
In 2016 Elor Azaria killed a Palestinian attacker who had been injured, fueling a public debate both at home and abroad over Israeli military actions.
中文摘要 以色列总统赦免前士兵埃洛尔·阿扎里亚。他于2016年射杀一名已被制服的巴勒斯坦袭击者，此案在以色列国内外引发对以军行动的广泛争论。
The president said his administration "will not in any way hinder or stifle the growth" amid amid warnings about the emergent technology.
中文摘要 特朗普表示，美国将组建「人工智能部队」并任命人工智能事务负责人。他称其政府「绝不会以任何方式阻碍或压制」这一新兴技术的发展。
The Houthis, an Iranian-backed militia, said they had fired a wave of missiles and drones at Riyadh, and on Saudi oil facilities.
中文摘要 胡塞武装称已向沙特首都利雅得及沙特石油设施发射多波导弹和无人机。这是伊朗支持的该武装与沙特之间冲突的显著升级。
Elor Azaria had shot a Palestinian who had already been incapacitated by Israeli troops, in controversial case.
中文摘要 以色列总统赦免士兵埃洛尔·阿扎里亚。他于2016年开枪打死一名已被以军制服的巴勒斯坦人，该案当时引发广泛争议。
The U.S. president’s threats to seize Greenland left its people on edge, and with details scarce, the news of an agreement didn’t seem to be putting minds at ease.
中文摘要 格陵兰民众对新协议表示紧张与怀疑。特朗普此前威胁夺取格陵兰令当地人不安，而协议细节稀少，消息并未缓解他们的担忧。
Elor Azaria’s conviction in 2017 for manslaughter of man who stabbed soldier continues to divide country Israel’s president has pardoned a combat medic who shot dead a badly wounded Palestinian attacker who had stabbed a soldier in the occupied West Bank, in a case that divided the country a decade
中文摘要 以色列总统赦免一名战斗医护兵。他曾在被占领的约旦河西岸开枪打死一名已受重伤、此前刺伤以军士兵的巴勒斯坦袭击者，2017年因过失杀人被定罪，该案持续分裂以色列社会。
Gaza's IT industry has been devastated from three-years of bombing and destruction.
中文摘要 加沙的信息技术产业因三年轰炸与破坏而严重受损。当地学生在设备匮乏的情况下仍努力跟上科技发展，不愿放弃学习。
US president has resisted calls for regulation but said he would appoint a ‘tsar’ to monitor for ‘bad’ behaviour
中文摘要 特朗普宣布组建「AI Force」，并将任命一名「AI 沙皇」监督相关「不良」行为。随着外界对人工智能技术进步的担忧升温，他此前一直拒绝监管呼声。
President Donald Trump’s plans for the Kennedy Center are hitting legal roadblocks as questions mount over the landmark’s future. Bloomberg’s Shaquille Omari breaks down the escalating battle over one of Washington’s most iconic cultural institutions. (Source: Bloomberg)
中文摘要 特朗普对肯尼迪中心的改造计划遭遇法律阻碍，围绕这一华盛顿标志性文化机构未来的争议持续升级。
Turkish authorities on Saturday detained high-profile finance executives as part of an investigation into a “ponzi-like” funds scheme.
中文摘要 据 NTV 报道，土耳其当局上周六拘留多名知名金融业高管，调查对象为一起「类庞氏」基金计划。
The news doesn’t stop when markets close. Hosts David Gura, Christina Ruffini and Lisa Mateo bring clarity, context and a bit of humor to the weekend’s biggest headlines, LIVE from New York. Joined by Brookings Institution VP and Director of Foreign Policy Program Suzanne Maloney, Signum Global Advi
中文摘要 彭博周末节目由 David Gura、Christina Ruffini 和 Lisa Mateo 在纽约直播，梳理周末主要新闻，嘉宾为布鲁金斯学会副会长兼外交政策项目主任 Suzanne Maloney。
Journalists from CNN, MS Now and Politico denied access as president escalates assault on media
中文摘要 特朗普禁止 CNN、MS Now 和 Politico 的记者进入白宫，进一步升级对媒体的施压。
Pointed offers a strategic twist to the news quiz format, testing not just players’ knowledge of the news but also their confidence in their answers. Join Bloomberg's Lisa Mateo, Christina Ruffini and David Gura as they play and check out the quiz for yourself at Bloomberg.com (Source: Bloomberg)
中文摘要 彭博推出新闻测验节目「Pointed!」，除考察参与者对新闻的了解外，还测试其对答案的信心程度。
Immersive movie screening, the plea for headphones, all-you-can-drink German method coming to the US, and old rock stars still rocking: Join Lisa Mateo, David Gura and Christina Ruffini as they go over interesting headlines from this week. (Source: Bloomberg)
中文摘要 节目盘点本周易被忽略的新闻，包括沉浸式电影放映、呼吁佩戴耳机、德国无限畅饮模式进入美国，以及老牌摇滚歌手仍在登台演出。
New York City is building a 36-person bureau of data scientists, technologists and economists to analyze company records for potential violations involving pricing, wages and other consumer and worker protections. NYC Consumer and Worker Protection Commissioner Sam Levine tells David Gura and Christ
中文摘要 纽约市组建一支36人的数据团队，由数据科学家、技术专家和经济学家构成，负责分析企业记录，查找定价、工资等环节可能存在的违规行为。市消费者与劳工保护专员 Sam Levine 就此表态。
Utah Governor Spencer Cox joins Bloomberg This Weekend and argues Americans need more substantive disagreement without the contempt he says is fueling polarization, damaged relationships and political violence. In his new book, Off Ramp, Cox calls for healthier political conflict and stronger local
中文摘要 犹他州州长 Spencer Cox 在彭博周末节目中表示，美国人需要更有实质内容的争论，减少他所称助长极化、损害人际关系并引发政治暴力的蔑视情绪。其新书《Off Ramp》呼吁更健康的政治冲突。
Oil and liquefied natural gas shipments through the Strait of Hormuz in the past two weeks reached the highest level in six months, signaling that US naval protection and mine clearance efforts are “paying off,” a regional US commander said.
中文摘要 一名美国地区指挥官称，过去两周经霍尔木兹海峡的石油和液化天然气运输量升至六个月来最高，显示美国海军护航和扫雷行动「正在见效」。
SpaceX will make up 2.82% of the Nasdaq 100 when the index’s quarterly rebalance takes effect Monday, according to data obtained by Bloomberg, locking in a sharp increase from its current weighting of about 1.28%.
中文摘要 据彭博获得的数据，纳斯达克100指数季度调整于周一生效后，SpaceX 的权重确认为2.82%，较当前约1.28%的水平大幅上升。
Investors are clamoring for the kind of debt companies are reluctant to sell now: longer-term bonds.
中文摘要 信贷周报显示，投资者正争相买入长期债券，而企业目前却不愿发行这类债券。
4 回复 · Apple 节点
4 回复 · 程序员 节点
21 回复 · 程序员 节点
10 回复 · 程序员 节点
17 回复 · 程序员 节点
7 回复 · 程序员 节点
4 回复 · Apple 节点
19 回复 · Apple 节点
18 回复 · Apple 节点
19 回复 · Apple 节点
该源今日无内容。