JavaScript · ★ 13,647 · 🍴 730 · 📈 3,006 stars today
A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings
中文介绍 Cloudflare 开源的编码智能体技能，把安全审计拆成多个阶段执行，输出机器可读、可独立复核的发现结果，降低 LLM 审计的误报与漏报。适合集成进 Claude Code 等 agent，对代码库做持续、可复现的安全扫描。
TypeScript · ★ 146,288 · 🍴 23,788 · 📈 444 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 官方推出的终端内编程智能体：能读取并理解整个代码库，执行常规开发任务、解释复杂代码、处理 git 工作流，让开发者用自然语言驱动编码。适合希望在命令行中直接获得 AI 结对编程能力的工程师。
Go · ★ 36,648 · 🍴 2,612 · 📈 2,704 stars today
Secure, fast, efficient, battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in multi-language ruleset (NPE, thread-safety, XSS, SQL injection), OpenAI & Anthropic compatible.
中文介绍 阿里巴巴开源、经内部大规模验证的代码审查工具，采用确定性流水线加 LLM Agent 的混合架构：先由规则引擎静态检查，再由大模型生成精确到行的评审意见，内置多语言规则集（含 NPE 空指针检测）。适合在 CI 中落地自动化 code review。
JavaScript · ★ 262,057 · 🍴 39,212 · 📈 958 stars today
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
中文介绍 面向前沿编码智能体的 harness 性能优化系统，围绕 skills、instincts、memory、security 与 research-first 开发流程，为 Claude Code、Codex、Opencode、Cursor 等工具提供统一增强层，提升 agent 在真实工程任务中的稳定性与效率。
TypeScript · ★ 5,285 · 🍴 366 · 📈 1,306 stars today
Let AI agents use your real, logged-in browser without interrupting your work. CLI + extension for browser automation across any shell-capable AI agent.
中文介绍 腾讯开源的浏览器自动化方案，由 CLI 加浏览器扩展组成，让 AI agent 直接操作用户已登录的真实浏览器，不打断日常工作，也无需重新登录或处理验证码。适用于需要复用登录态抓取数据、操作后台的自动化场景。
JavaScript · ★ 96,396 · 🍴 10,190 · 📈 675 stars today
Production-grade engineering skills for AI coding agents.
中文介绍 为 AI 编码智能体准备的生产级工程技能集合，把代码规范、测试、重构、发布等成熟工程实践封装成可复用的 skill，供 agent 在真实项目中调用，减少生成代码的随意性。适合在团队内规范 AI 辅助开发的流程与产出质量。
Python · ★ 3,950 · 🍴 406 · 📈 569 stars today
A smarter, self-hosted AI assistant — multi-user, multi-agent.
中文介绍 腾讯云开源的自主托管 AI 助手，支持多用户与多 agent 协作，数据和模型部署在自有环境内，兼顾隐私与可控性。适合企业或团队搭建内部智能助手平台，用于知识问答、任务编排等日常协作场景。
TypeScript · ★ 69,341 · 🍴 4,750 · 📈 296 stars today
Spec-driven development (SDD) for AI coding assistants.
中文介绍 面向 AI 编码助手的规格驱动开发（SDD）框架，主张先写结构化规格说明再让模型生成代码，把需求、设计与实现对齐，减少大模型自由发挥带来的偏差。适合需要可追溯、可评审的 AI 辅助开发流程的团队。
Rust · ★ 31,208 · 🍴 3,237 · 📈 174 stars today
Anki is a smart spaced repetition flashcard program
中文介绍 老牌开源间隔重复记忆卡片软件，基于遗忘曲线安排复习计划，支持图片、音频、数学公式与插件扩展，并可跨平台同步。广泛用于语言学习、医学与各类考试备考等需要长期记忆大量知识的场景。
Python · ★ 24,874 · 🍴 2,970 · 📈 299 stars today
Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork
中文介绍 Anthropic 开源的插件仓库，主要面向知识工作者，在 Claude Cowork 中可直接使用，把文档处理、信息整理等常见办公流程封装为可复用能力，降低非工程用户上手 AI 协作工具的门槛。
TypeScript · ★ 30,268 · 🍴 2,640 · 📈 140 stars today
Memory and context engine + app that is extremely fast, scalable, and can be run fully locally. The Memory API for the AI era.
中文介绍 面向 AI 时代的内存与上下文引擎，提供极速、可扩展的 Memory API，支持完全本地运行，帮助应用在多轮对话和长任务中持久保存并检索用户记忆与上下文。适合构建个性化助手、为 Agent 加装记忆层的开发者。
JavaScript · ★ 6,463 · 🍴 2,717 · 📈 79 stars today
AI-assisted TradingView chart analysis — connect Claude Code to your TradingView Desktop for personal workflow automation
中文介绍 把 Claude Code 接入本地 TradingView Desktop 的 MCP 工具，让 AI 读取并分析图表行情，用自然语言完成技术面分析、标注与个人工作流自动化。适合希望用 AI 辅助盯盘和复盘的个人交易者。
Rust · ★ 33,159 · 🍴 1,482 · 📈 267 stars today
RustFS is an open-source, S3-compatible high-performance object storage system supporting migration and coexistence with other S3-compatible platforms such as MinIO and Ceph.
中文介绍 用 Rust 编写的高性能开源对象存储，兼容 S3 协议，支持与 MinIO、Ceph 等其他 S3 平台的数据迁移与共存。相较传统方案更强调性能与资源占用，适合自建私有存储、AI 训练数据湖等需要弹性扩展的场景。
TypeScript · ★ 110,142 · 🍴 14,301 · 📈 120 stars today
The Postgres development platform. Supabase gives you a dedicated Postgres database to build your web, mobile, and AI applications.
中文介绍 基于 Postgres 的开发平台，为 Web、移动端与 AI 应用提供专属数据库，并集成认证、存储、实时订阅与边缘函数等能力。开发者无需自建后端基础设施，即可快速搭建可扩展的全栈应用。
Go · ★ 15,281 · 🍴 1,504 · 📈 478 stars today
Secure environments for developers and their agents
中文介绍 为开发者及其 AI agent 提供安全、隔离的远程开发环境，把工作区集中托管在服务器或云上，统一管理权限、镜像与凭据。适合需要标准化开发环境、支持远程协作或让 agent 安全执行代码的工程团队。
TypeScript · ★ 16,463 · 🍴 1,255 · 📈 152 stars today
Free, simple, fast interactive diagrams for any GitHub repository
中文介绍 免费的 GitHub 仓库可视化工具，输入任意仓库地址即可生成可交互的架构与依赖关系图，帮助开发者快速理解陌生项目的模块划分与代码结构。适合接手新项目、做技术调研或撰写文档时使用。
Go · ★ 4,959 · 🍴 204 · 📈 889 stars today
Your own search engine
中文介绍 作者自研的个人搜索引擎，让你拥有并运行自己的检索服务，索引与查询都在自己掌控的环境中完成，减少对商业搜索引擎的依赖。适合注重隐私、想自建搜索或做垂直领域检索的开发者。
👍 73
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人类知识，但工具链碎片化、领域约定隐式、正确性判据特殊，难以转化为可靠学习经验。ScienceIDE 将全球科学代码库改造为 agent 可学习的交互环境，用于科学智能体的训练与评测。
👍 57
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 长程 agent 使模型负载日益偏重输入，prefill 计算依旧昂贵，庞大 KV cache 持续挤压 HBM 与 SSD 的容量和带宽。DeepSeek-V4.1-Flash 面向这一瓶颈推进 KV cache 压缩极限，在维持长上下文能力的同时降低显存与存储开销。
👍 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 可靠置信度估计是 LLM 可信部署的核心，决定何时上线、升级或重试。现有方法共享同一设计前提，难以覆盖从推理到 agent 的真实经历。论文提出「经验置信度」估计框架，让模型从自身推理与交互经验中校准输出正确的概率。
👍 46
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 编码 agent 从监督式代码补全转向无人值守的持续探索，工作形态变为推理、工具调用与反馈组成的长轨迹，token 效率成为递归自我改进的关键瓶颈。SoL-Pi 通过递归扩展 auto-research 循环，为高效 agent harness 提供可扩展方案。
👍 46
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 现有编码 agent 评测多以 issue 或指令描述目标行为，而真实 web 开发中 agent 需从可运行软件反推行为并在不完整应用中实现。ProgramDistill 提出从交互式 web 应用自动构造可验证、参考引导的 SWE 任务与基准，用于评测该能力。
👍 41
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 等自主研究循环表明单个编码 agent 可无人值守改进训练配置，但多个 agent 并行时各自从零开始，导致重复搜索多于新发现。Agora 以 Git 作为共享内存，让研究 agent 复用彼此成果与经验，减少重复探索。
👍 37
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 编码 harness 决定自主 agent 如何把模型能力转化为长程软件工程表现，但已有工作多把 harness 当作整体评测，各组件作用不明。该实证研究在组件级别拆解并对比 harness 设计要素，厘清哪些设计真正提升 agent 的长程任务性能。
👍 31
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 能融合行情、新闻与可执行分析，但其行为由部署前写死的工具使用策略控制，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，持续改进决策流程。
👍 30
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）中的长度膨胀问题：学生回复可能过长乃至耗尽生成预算。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，为抑制长度膨胀提供依据。
👍 25
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 通过综合证据、评估提案与提出新解释参与假设发现，近期系统将科学 agent 与演化搜索结合，以批判、比较与修订迭代。HypoEvolve 用遗传算法驱动多 agent LLM，考察不同协作形式对科学假设发现效果的影响。
👍 23
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测后果、指导干预并从交互中学习，但现有预测模型多局限于单一领域。JEPA-Anything 提出领域无关的预测学习框架，尝试用统一的学习原则在差异极大的系统上训练世界模型，实现跨「世界」的预测建模。
👍 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 直接偏好对齐因计算与显存高效而被广泛用于 LLM 人类偏好对齐，但似然位移问题促使人探索从似然间隔较小的偏好对中提取信息的替代方案。该工作提出零阶（zeroth-order）优化范式做偏好对齐，规避反向传播的代价与局限。
👍 21
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 MiniMax-H3 将多模态上下文理解与共享隐空间中的音视频联合生成结合，代表 Omni-Modal 生成模型的统一化趋势。该工作评测这类全模态生成模型是否真正具备物理世界推理能力，检验统一建模是否带来对物理规律的理解。
👍 17
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 平台滥用常以 emoji、谐音、拆字与冗余符号隐藏跳转指令，再经伪装链接把用户导向色情、诈骗、赌博或非法交易服务。RiskChainBench 提出面向混淆平台信息还原与证据支撑式 web 调查的基准，评测模型在此类任务上的能力缺口。
👍 16
Enabling robots to adapt to unfamiliar environments as readily as humans remains a moonshot goal of embodied AI. No finite collection of demonstrations can cover every task and situation a robot will encounter, making the ability to learn from context at deployment essential for generalization. Such
中文介绍 让机器人像人一样快速适应陌生环境是具身智能的长期目标，有限示范无法覆盖所有任务与情境，部署时的上下文学习能力因而至关重要。该工作用 VLM agent 实现上下文机器人学习，使机器人在部署阶段直接依据上下文适应新任务。
👍 15
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 借助具备特权任务技能的自我教师提供 token 级稠密监督，供无技能学生内化。RetireOPD 提出自我退场机制，在监督不再必要时关闭蒸馏，避免其干扰后续 RL。
👍 13
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 助手广泛用于日常社交建议，但其在该类咨询场景中的社交推理难以评测：助手需从主观用户叙述中理解社交情境，且「他人感受」等社交属性难以客观验证。该工作提出可验证的社交推理评测方法，检验助手的社交建议能力。
👍 11
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程任务，弹窗、延迟加载与控件移位常使执行前固定的计划失效，而现有 skill 设计固化、难以适配。该工作提出「反思—修订—复用」的训练无关 skill 演化方法，让 GUI agent 在运行中持续更新可复用过程知识。
👍 10
Mixture-of-experts (MoE) inference on consumer hardware is bounded by weight memory: a 35B-class model is 19.5GB at 4-bit, and sparsity shrinks the compute per token, not the bytes that must be held. Naive offloading to SSD does not help on its own, because layer N+1's experts must be chosen before
中文介绍 消费级硬件上 MoE 推理受权重内存限制：35B 模型 4-bit 下需 19.5GB，稀疏性只减少每 token 计算量而非驻留字节；朴素 SSD 卸载无效，因第 N+1 层专家必须在读取前确定。该工作用可训练的路由预测从 SSD 服务 35B MoE。
👍 10
Training a Mixture-of-Experts (MoE) model at long context or large batch size fails as soon as any one component's peak allocation exceeds device memory, so the target is every peak at once, not the average footprint. Four are left unbounded by the parallelism plans in common use, and each grows dif
中文介绍 长上下文或大 batch 训练 MoE 时，只要任一组件峰值分配超过设备显存就会失败，因此目标是压平所有峰值而非降低平均占用。该工作针对常用并行方案中四个无上界峰值逐一处理，实现长上下文 MoE 训练显存峰值的全面压平。
@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 作者复盘让 AI agent 真正经营一家公司的实验：7 月起给 Devin 配了 Ramp 企业卡和电话号码，让它自主运营并赚到第一笔 75 美元，分享这个过程中踩到的坑与经验，属于 AI 自动化创业的实操记录。
@mustafasuleyman · 1.1M 粉丝 · 204.3K 阅 · 502 赞 · 86 转
AIs are not conscious. They do not feel, experience, or suffer. However, there’s a growing movement in support of model welfare; the idea that we might soon owe them a duty of care. I think this
中文介绍 Mustafa Suleyman 撰文谈「模型福利」：他先明确 AI 并不具备意识、不会感受或痛苦，但指出主张给予模型某种照护义务的声音正在增多，认为这一议题值得提前展开讨论，属于偏哲学与治理层面的观点输出。
@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 面向新手的教程帖，目标人群是囤了大量照片视频却没时间发社交媒体的商家。作者用 GPT-6 Astra 搭配 DaVinci MCP，尝试把写文案、剪视频这类内容工作自动化，解决「素材多但不会写不会剪」的痛点。
@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 用大白话解释 Jev 是什么：它不做聊天式写作，而是像「AI 选择题」，输出软件可直接执行的决策，比如是否垃圾信息、agent 该调哪个工具、是否需要人工介入。作者还整理了人们已经用它做出来的 9 类实际应用。
@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 一篇讲怎么用 Jev 搭 agent harness 的技术贴。作者先拆解 agent 的循环结构：LLM 决定动作、工具执行、模型评估结果，如此往复直到任务完成，再说明如何借助 Jev 把这一循环落地成可用的框架。
@JacquelineSYC19 · 1.0K 粉丝 · 72.4K 阅 · 501 赞 · 47 转
At the time of writing, Artie is a team of 17. Every one of us works alongside Hermes, an open-source AI agent harness from @NousResearch. It runs on a single physical machine in Germany, has a
中文介绍 Artie 团队分享内部部署：17 人全员日常使用 @NousResearch 开源的 AI agent harness Hermes，整套系统跑在德国一台物理机上。属于小团队自建 agent 基础设施的实战案例，涉及开源方案选型与部署细节。
@ShaneLegg · 82.3K 粉丝 · 40.6K 阅 · 585 赞 · 86 转
We are on the cusp of a profound transformation. Today’s AI systems have impressive capabilities and the rapid pace of innovation suggests we’re now approaching artificial general intelligence (AGI),
中文介绍 DeepMind 联合创始人 Shane Legg 宣布成立 DeepMind Institute。他表示当前 AI 能力已相当可观，创新速度意味着正逼近 AGI，新机构大概率为应对这一转折而设，属于机构发布加方向性表态。
@Didotxyz_ · 64.6K 粉丝 · 5.7K 阅 · 502 赞 · 325 转
The important shift in AI agents is not the disappearance of chat. It is that chat is no longer the whole product. An assistant explains a task. An agent system has to carry it forward: call a tool,
中文介绍 观点帖：AI agent 的真正变化不是聊天消失，而是聊天不再是产品的全部。助手只负责解释任务，agent 系统则要接着往下推进——调工具、执行、闭环。讨论的是从对话界面转向可执行系统的产品形态迁移。
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 9 月 18 日，MIT 科技评论为订阅用户举办线上圆桌活动，探讨「AI 是否真的会杀死我们所有人」，因 30 分钟问答时间不足，编辑汇总读者提问并邀请资深记者作答。
a quiet day
中文介绍 Latent Space 的 AI 新闻日报表示，当日行业消息平淡，没有重大事件发生。
中文介绍 TLDR AI 9 月 18 日简报要点：Claude Projects 发布 v2 版本，Google 推出面向家庭的 AI 智能体。
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 Cooley 律所基于 ChatGPT 打造 GO Public 工具，将 AI 引入 IPO 流程，帮助律师更早发现问题，并把判断力集中在最关键环节。
A dash of cold water keeps the foomers away.
中文介绍 Latent Space 称 AI 新闻出现「现实检验」：Yegge 关闭 Gas Town 项目，Databricks 的 Astra 成本上升 60%。
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，提供前沿模型能力、律所定制工作流、接入法律数据源，并针对机密客户工作提供法律级管控。
中文介绍 TLDR AI 9 月 17 日简报要点：Claude 与 Cowork 合并，ChatGPT 推出赞助式智能体，以及 harness tax 相关讨论。
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，讨论该公司 A 轮融资，主题为「为超级智能承保」——为可被起诉的 AI 智能体提供保障。
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 公布模型错位报告框架，用于追踪、调查和披露模型错位问题，并同时发布六份关于模型意外或令人担忧行为的报告。
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，将在美国 10 座城市为 1000 名老年人提供免费 ChatGPT 实操工作坊，帮助其安全掌握实用 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，包括赞助式智能体与面向营销人员的工具，并宣布与 HubSpot、Shopify 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 AI 热潮正演变为材料挑战：随着算力扩张，半导体与数据中心在性能等方面逼近物理极限，支撑基础设施的材料与算法同等重要。
GPT-6 Astra helps Hex’s data agents turn answers into interactive visualizations that employees are proud to share.
中文介绍 Hex 借助 GPT-6 Astra，让数据智能体把复杂分析答案转化为交互式可视化报告，便于员工分享。
Learn how ChatGPT Work and Codex analytics help teams understand AI usage and spend, identify training needs, and connect adoption to business outcomes.
中文介绍 OpenAI 介绍如何用 ChatGPT Work 与 Codex 的分析功能了解 AI 使用量与支出、识别培训需求，并把 AI 采用与业务成果关联起来。
congrats to TypeSafe!
中文介绍 Jev 是一种「系统一模型」，只负责决策、分类、路由和打分，据称速度比小型前沿大模型快 100 倍以上，成本低 200 倍以上。
美股方面，SPY 现价 761.69，略低于 SMA20 764.35 但高于 SMA50 759.53 与 SMA200 715.87，RSI14 50 居中，MACD -0.6328 低于信号 0.8372，动能读数偏弱；QQQ 单日 +2.37%、5 日 +1.8% 相对领先，RSI14 53，呈多头排列。个股分化：AAPL RSI14 65.4 且距 52 周高 -2.45%，NVDA 单日 +3.91%，而 META RSI14 74 已入超买。加密方面，恐慌贪婪指数 71 处于「贪婪」区间，总市值 2.78T 美元（24h +2.53%），BTC 主导率 58.3%、ETH 11.4%；BTC 单日 +6% 报 80990.1，RSI14 63.8，MACD 1121.35 低于信号 1522.13，SOL 短端弹性最大，ETH、BTC、SOL 均为多头排列，但三者距 52 周高分别 -35.82%、-45.03%、-54.54%。中概整体最弱：BABA、PDD、0700.HK 全部空头排列，RSI14 分别 37.2、34.3、36.9，JD RSI14 30.4 逼近超卖。商品外汇：黄金 4415.9 报 RSI14 49.6，位于 SMA200 4553.31 下方，形态中性；WTI 原油单日 -6.32% 报 95.47，但仍守 SMA20/50/200 多头排列；美元人民币 RSI14 21.7 超卖。宏观端 VIX 14.81 空头排列，10Y 收益率 5%、RSI14 68.8 且距 52 周高 -0.36%，美元指数 100.22 多头排列。
现价 112.96，单日 +11.18%、5 日 +13.82%，为样本中短期涨幅最大者；RSI14 66.6 尚未进入超买，MACD 3.0549 略低于信号 3.3645，柱状仍偏负；价格站稳 SMA20 102.22、SMA50 90.73、SMA200 83.54，多头排列完整，但距 52 周高仍有 -54.54%。
读数报 5，单日 +1.03%，距 52 周高仅 -0.36%；RSI14 68.8 逼近超买区，MACD 0.0858 高于信号 0.0705，红柱延续；价格位于 SMA20 4.83、SMA50 4.72、SMA200 4.4 之上，多头排列完整，趋势字段 bullish，动量与技术形态同步走强。
现价 364.27，单日 +1.73%、5 日 +0.2%，RSI14 54.8 居中；MACD 3.7869 刚上穿信号 3.5925 触发金叉，但价格仍低于 SMA200 397.65，仅高于 SMA20 358.64 与 SMA50 351.09，trend 字段为 neutral，距 52 周高 -26.98%，多空信号交织。
现价 665.75，RSI14 74 为样本最高并进入超买，单日 -1.12% 逆势回落；价格较 SMA20 608.61 高出约 9.4%，短期偏离度偏大；MACD 23.979 高于信号 14.2832，红柱仍厚，但 trend 字段为 neutral，距 52 周高 -15.81%，高位动量与回落压力并存。
现价 6.7，单日与 5 日均 -0.16%，距 52 周低仅 0.18%；RSI14 21.7 处于超卖，MACD -0.0101 略高于信号 -0.0104，柱状接近零轴；价格低于 SMA20 6.71、SMA50 6.74、SMA200 6.84，空头排列，趋势字段 bearish，弱势形态明确。
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出的技术指标读数作客观复述，不构成任何投资建议或买卖依据。过去走势不代表未来表现，指标信号可能随时反转。仅供技术指标解读参考，请读者独立判断并自行承担风险。
Trump says the US will gain ‘permanent control’ over security in Greenland under a deal with Denmark.
中文摘要 特朗普表示，根据与丹麦达成的协议，美国将在格陵兰安全事务上获得「永久控制权」。
Witnesses say men were hooded and beaten in Equatorial Guinea. The autocratic government there is a partner in President Trump’s deportation program.
中文摘要 证人称，被美国驱逐至赤道几内亚的男性遭蒙头殴打。该国威权政府是特朗普驱逐计划的合作方。
The US president has had a tense and fiery relationship with the US media throughout his political career.
中文摘要 特朗普宣布禁止 CNN 和 Politico 进入白宫。他在政治生涯中与美国媒体关系一直紧张激烈。
Mary Timby flew to Ireland on a whim last year and met her boyfriend at a matchmaking festival.
中文摘要 玛丽·廷比去年一时兴起飞往爱尔兰，在一场相亲节上结识了男友，她希望更多人也能如此收获爱情。
In Super Macklemore 64, players are chased by a giant Ed Sheeran, dodge tiny billionaire Robert Krafts.
中文摘要 一款名为《Super Macklemore 64》的游戏以希兰演唱会争议为题材，玩家被巨型艾德·希兰追赶，并躲避缩小版的亿万富翁罗伯特·克拉夫特。
The war in Yemen intensifies with mounting clashes and reciprocal attacks between Saudi forces and the Houthis.
中文摘要 胡塞武装指控沙特阿拉伯在24小时内发动26次空袭。也门战事升级，沙特部队与胡塞武装相互攻击不断。
Group trapped at Hotel Bamy after US removal as rights groups sound alarm over poor conditions and violence Two men that the Trump administration expelled to Equatorial Guinea were bound, fitted with bags over their heads, beaten and pushed down a flight of stairs, all in plain view of other US depo
中文摘要 律师称，两名被特朗普政府驱逐至赤道几内亚的男子在巴米酒店遭捆绑、蒙头殴打并被推下楼梯。人权组织对当地恶劣条件与暴力表示担忧。
The Tata boardroom battle has intensified over a re-appointment of its chairman and a diktat to go public.
中文摘要 围绕董事长连任及要求上市的指令，塔塔集团董事会内斗加剧，这家印度企业巨头正面临动荡。
Russians head to the polls to choose lawmakers for their parliament, the Duma. It's Russia's first nationwide parliamentary race since launching its full scale invasion of Ukraine in 2022.
中文摘要 俄罗斯民众前往投票站选举国家杜马议员。这是俄罗斯自2022年全面入侵乌克兰以来的首次全国性议会选举。
Dineo Motapane is one of nine recent victims whose deaths have left a nation traumatised over the level of violence against women.
中文摘要 南非近期发现九名被害女性，迪内奥·莫塔帕内是其中之一。这些案件令全国对针对女性的暴力程度感到震惊与恐惧。
Increasingly frequent and ferocious attacks on the Ukrainian capital have left many on edge.
中文摘要 俄罗斯对基辅的袭击日益频繁猛烈，当地居民长期处于紧张不安之中，四分钟的预警时间被认为远远不够。
The agreement would allow the US to expand its military presence, while Denmark and Greenland say it preserves their sovereignty and right to self-determination Donald Trump announced on Friday that the US had struck an Arctic security pact with the leaders of Denmark and Greenland, after threatenin
中文摘要 美国与丹麦及格陵兰达成北极安全协议，将允许美方扩大军事存在；丹麦与格陵兰表示协议维护其主权与自决权。
Trump said the agreement will give the US "permanent control over security, and all other needs, in Greenland", though specifics haven't been confirmed by Danish officials.
中文摘要 特朗普称与丹麦的协议将让美国获得格陵兰安全及所有其他需求上的「永久控制权」，但丹麦官员尚未确认具体内容。
Edgars Rinkēvičs says drone warfare in Ukraine is moving at ‘lightning speed’ and efforts to prepare for an aerial war could be obsolete by next year Latvia’s president has warned that Europe is not keeping up with developments of drone and aerial warfare in Ukraine and that most countries in Nato c
中文摘要 拉脱维亚总统林克维奇斯警告，欧洲未能跟上乌克兰无人机与空战的发展速度，相关备战努力到明年可能已经过时。
Denmark and Greenland plan to sign a US defence pact next week, pending approval from both parliaments.
中文摘要 特朗普称与丹麦的协议让美国获得对格陵兰的「永久控制权」。丹麦与格陵兰计划下周签署防务协议，尚待双方议会批准。
CEO Dario Amodei says labs should more deeply embed third-party testers as concerns grow around risks associated with the technology
中文摘要 Anthropic 引入埃森哲参与人工智能安全测试。首席执行官达里奥·阿莫代伊表示，随着外界对相关技术风险的担忧上升，AI 实验室应更深入地引入第三方测试机构。
This week, a special edition of Wall Street Week from the fastest growing large economy in the world: India. The country's technology sector has helped power its growth, but its success has led to uneven progress that could turn its strengths into weaknesses. Bangalore became India’s tech capital by
中文摘要 彭博《华尔街一周》推出印度特辑。印度是全球增长最快的主要经济体，科技行业推动其增长，但发展并不均衡，班加罗尔的基础设施问题可能令其优势转为劣势。
AI start-up has projected deeply negative cash flows as it invests in infrastructure and faces price pressures
中文摘要 OpenAI 预计到2030年将烧掉2800亿美元。这家AI初创公司因投资基础设施并面临价格压力，预计现金流将深度为负。
President bars CNN, MS Now and Politico as he escalates assault on media
中文摘要 特朗普禁止CNN、MS Now和Politico等美国主要新闻机构进入白宫，此举被视为其对媒体攻击的进一步升级。
Bloomberg's Lisa Mateo joins Scarlet Fu and Tom Keene on "Bloomberg Money." They discuss New York City's Plaza Hotel as it is set to welcome events and guests tied to the United Nations General Assembly and the Qatar Economic Forum. The menu includes the $370 "Home Alone Sundae". (Source: Bloomberg)
中文摘要 纽约广场酒店将接待与联合国大会和卡塔尔经济论坛相关的活动及宾客，其菜单中包括售价370美元的「小鬼当家圣代」。
Caitlin Donovan, the Global Head of Sales at JOOPITER, discusses the rapid growth of sports memorabilia as an alternative asset class, saying the category is one of the fastest-growing parts of the collectibles market. JOOPITER, the digital auction house founded by Pharrell Williams in 2022, has put
中文摘要 JOOPITER全球销售主管凯特琳·多诺万表示，体育纪念品作为另类资产类别快速扩张，是收藏品市场中增长最快的领域之一。该数字拍卖行由法瑞尔·威廉姆斯创办。
Bloomberg's Chris Rovzar joins Scarlet Fu and Tom Keene on "Bloomberg Money" to discuss the must-read fashion books. (Source: Bloomberg)
中文摘要 在「Bloomberg Money」节目中，彭博的克里斯·罗夫扎尔讨论值得一读的时尚类书籍，包括《Dressing The Man》与《Preppy Handbook》。
Start-up has won a huge contract to supply computing power to Anthropic
中文摘要 英国数据中心集团Nscale申请在美国上市，规模350亿美元。这家初创公司此前赢得向Anthropic供应算力的巨额合同。
Bloomberg's Chris Rovzar joins Scarlet Fu and Tom Keene on "Bloomberg Money." They discuss the resurgence in suit sales for both men and women. (Source: Bloomberg)
中文摘要 在「Bloomberg Money」节目中，彭博的克里斯·罗夫扎尔与主持人讨论称，消费者重新找回打扮的乐趣，男女西装销量随之明显回升。
A nonprofit group with ties to US Health and Human Services Secretary Robert F. Kennedy Jr. solicited payment from companies in exchange for access to top agency officials at an upcoming event. John Tozzi explains. (Source: Bloomberg)
中文摘要 一家与美国卫生与公众服务部长小罗伯特·肯尼迪有关联的非营利组织，向企业收取费用，以换取在即将举行的活动中接触该机构高层官员的机会。
Alicia Levine, chief investment officer at BNY Wealth, joins Scarlet Fu and Tom Keene on "Bloomberg Money." They discuss retirement planning, building wealth and asset allocation. (Source: Bloomberg)
中文摘要 BNY财富首席投资官艾丽西亚·莱文在「Bloomberg Money」节目中表示，积累财富时不要像对冲基金那样操作，并讨论了退休规划与资产配置。
Alicia Levine, chief investment officer at BNY Wealth, joins Scarlet Fu and Tom Keene on "Bloomberg Money." They discuss retirement planning, building wealth and asset allocation. (Source: Bloomberg)
中文摘要 BNY财富首席投资官艾丽西亚·莱文建议分散资本配置并让其长期运行，她在节目中与主持人讨论了退休规划、财富积累与资产配置。
4 回复 · 程序员 节点
5 回复 · 程序员 节点
17 回复 · 程序员 节点
5 回复 · 程序员 节点
17 回复 · 程序员 节点
9 回复 · Apple 节点
13 回复 · Apple 节点
13 回复 · Apple 节点
14 回复 · Apple 节点
9 回复 · Apple 节点
该源今日无内容。