每日简报

2026-11-06

← 历史归档

NVIDIA/auto-ontology

Python · ★ 7,548 · 🍴 599 · 📈 2,679 stars today

Auto Ontology is a Python framework for building semantic, graph-based reasoning layers over data. It enables deterministic query generation, better context retrieval, and more explainable AI for LLM and agent workflows.

中文介绍 NVIDIA 开源的 Python 框架,为数据构建基于语义图谱的推理层。通过确定性查询生成与上下文检索增强,提升 LLM 的可解释性,适用于企业级数据分析与智能问答。

mattpocock/skills

Shell · ★ 315,419 · 🍴 26,408 · 📈 2,071 stars today

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

中文介绍 面向真实工程师的技能集合,源自作者本人的 .agents 目录。提供可直接复用的智能体技能,帮助开发者快速搭建专业级 AI 工作流,适合需要工程化 AI 助手的团队。

garrytan/gstack

TypeScript · ★ 140,016 · 🍴 20,699 · 📈 612 stars today

Use Garry Tan's exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA

中文介绍 Y Combinator CEO Garry Tan 的 Claude Code 配置合集,含 23 个意见型工具,覆盖 CEO、设计、工程管理、发布、文档与 QA 角色。用于模拟多角色协作的 AI 工作流。

Usagi-org/ai-goofish-monitor

Python · ★ 17,709 · 🍴 2,964 · 📈 240 stars today

基于 Playwright 和AI实现的闲鱼多任务实时/定时监控与智能分析系统,配备了功能完善的后台管理UI。帮助用户从闲鱼海量商品中,找到心仪产品。

中文介绍 基于 Playwright 与 AI 的闲鱼多任务实时/定时监控系统,配后台管理 UI。自动分析海量商品,帮用户快速筛选心仪产品,适合二手交易场景的精准捡漏。

D4Vinci/Scrapling

Python · ★ 98,113 · 🍴 10,278 · 📈 1,655 stars today

🕷️ The BEST Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scrapling_dev

中文介绍 号称最强的 Python 网页抓取框架,从单请求到全站爬取全能处理。内置智能适配、动态渲染与反爬策略,适合数据采集、市场调研与自动化测试场景。

vercel-labs/agent-desktop

Rust · ★ 17,032 · 🍴 1,202 · 📈 2,173 stars today

Computer use CLI for AI agents

中文介绍 Vercel 实验室推出的命令行工具,为 AI 智能体提供电脑操控能力。代理可直接执行桌面操作,适合自动化测试、界面交互与无人值守任务。

max-sixty/worktrunk

Rust · ★ 11,561 · 🍴 439 · 📈 546 stars today

Worktrunk is a CLI for Git worktree management, designed for parallel AI agent workflows

中文介绍 专为并行 AI 智能体设计的 Git worktree 管理 CLI。自动隔离分支、简化切换流程,让多个代理并行开发互不干扰,适合多智能体协作与代码审查场景。

lz4/lz4

C · ★ 12,400 · 🍴 1,636 · 📈 23 stars today

Extremely Fast Compression algorithm

中文介绍 极速压缩算法库,以接近内存带宽的速度进行无损压缩。适用于实时数据传输、嵌入式系统与日志处理等对延迟敏感的场景,提供 C 及多种语言绑定。

google/skills

Python · ★ 22,426 · 🍴 1,895 · 📈 211 stars today

Agent Skills for Google products and technologies

中文介绍 Google 发布的 Agent Skills 合集,涵盖搜索、地图、日历等产品能力。为 AI 智能体提供标准化接口,简化与 Google 服务的集成,适合构建生产级助手应用。

vercel-labs/agent-browser

Rust · ★ 45,970 · 🍴 3,151 · 📈 150 stars today

Browser automation CLI for AI agents

中文介绍 Vercel 实验室推出的浏览器自动化 CLI,专为 AI 智能体设计。支持页面导航、元素交互与截图,可驱动真实浏览器执行任务,适合网页测试与数据采集。

scenario-labs/skills

Python · ★ 3,125 · 🍴 274 · 📈 667 stars today

Get production-ready images, video, audio, and 3D from any AI agent: skills that pick the right model, price before spending, and keep characters and brands consistent through the Scenario MCP, plus expert teams that drive Blender, Maya, ZBrush, Unreal, and Unity.

中文介绍 让 AI 智能体生成生产级图像、视频、音频与 3D 资产。通过 Scenario MCP 自动选模型、预算前置,并保持角色与品牌一致性,适合创意产业与内容制作。

ConardLi/garden-skills

JavaScript · ★ 14,196 · 🍴 1,652 · 📈 175 stars today

ConardLi's open-source Skills collection, featuring web design, knowledge retrieval, image generation, and more.

中文介绍 ConardLi 开源技能合集,涵盖网页设计、知识检索、图像生成等方向。提供可插拔的技能模块,帮助开发者快速为 AI 助手扩展专业能力,适合个性化工作流。

microsoft/VibeVoice

Python · ★ 56,047 · 🍴 6,306 · 📈 223 stars today

Open-Source Frontier Voice AI

中文介绍 微软开源的前沿语音 AI 框架,集成了实时语音识别与合成能力。支持低延迟对话与多语言交互,适合构建语音助手、呼叫中心机器人等场景,强调开放与可定制。

floci-io/floci

Java · ★ 30,479 · 🍴 3,333 · 📈 779 stars today

Light, fluffy, and always free - The AWS Local Emulator alternative

中文介绍 轻量、免费且开源的 AWS 本地模拟器替代品。可离线模拟 S3、DynamoDB 等核心服务,帮助开发者无云开销测试应用,适合本地开发、CI 流水线与教学环境。

JiaFeiMiao-K-Cat/DiscImageStudio

C# · ★ 7,350 · 🍴 263 · 📈 984 stars today

中文介绍 暂无描述信息,但据项目名推测为磁盘镜像处理工具,可能支持 ISO、IMG 等镜像的创建、编辑或提取。适合系统备份、数据恢复或光盘复制场景。

Turns Matter: Advancing Streaming Video RL via Turn-Level Credit

👍 124

Streaming video understanding requires models to repeatedly decide whether to respond or continue observing. Recently, Group Relative Policy Optimization (GRPO) has been applied to this setting, with existing approaches emphasizing reward design while retaining a single trajectory-level advantage ac

中文介绍 针对流式视频理解中GRPO奖励设计未区分轨迹内各回合贡献的问题,提出回合级信用分配方法,将优势分解至每个回合,改善长时交互场景下的训练稳定性与模型性能。

From Pixels to Actions - Towards Native Vision-Language-Action Policies

👍 56

Vision-language-action models are typically built on a simple premise: perception is first compressed by visual encoders, and action reasoning begins only afterwards. Yet manipulation often hinges on precisely the visual signals most vulnerable to this abstraction, e.g., local geometry, boundaries,

中文介绍 指出视觉编码器预压缩会丢失操纵任务所需的关键局部几何与边界信息,提出原生视觉-语言-动作策略,将视觉特征直接融入动作推理,减少信息瓶颈,提升操作精度。

FrontierPhysics: Evaluating Agents for End-to-End Physics Research

👍 48

AI agents are playing an increasingly important role in physics research. However, existing physics benchmarks only evaluate fixed questions, isolated sub-tasks, or paper reproduction with given solutions, and rarely check research quality alongside correctness. We introduce FrontierPhysics, an end-

中文介绍 构建端到端物理研究基准FrontierPhysics,评估智能体从假设到论文的完整科研流程,不仅检查结果正确性,还评估研究质量,覆盖多子领域,揭示现有智能体在开放研究中的不足。

What Does Multi-Turn Reinforcement Learning Really Improve in LLM Agents?

👍 34

While multi-turn agentic reinforcement learning (RL) excels on long-horizon tasks, its performance gains remain highly sensitive to interaction constraints. Evaluating across sampling breadth K and interaction depth T via Pass@(K,T), we show that trajectory-level RL advantages depend on sufficient t

中文介绍 基于Pass@(K,T)度量,系统分析多轮智能体强化学习中采样宽度与交互深度的影响,发现轨迹级优势依赖足够的任务探索,且性能提升对交互约束高度敏感,为训练设计提供指导。

Halley: A Predictive Control Plane for Optical AI Data Center Networks

👍 31

A frontier AI model is trained in an AI data center, on tens of thousands of accelerators joined by a dedicated network, and every training step ends with all of them exchanging data at once. Between racks, that network is optical, and each of its four layers, from lasers to switches, runs a fixed o

中文介绍 提出Halley,一种预测性控制平面,用于AI数据中心光网络,利用全局训练步信息动态调度光交换,降低网络拥塞和时延,提升大规模训练的效率与稳定性。

PyRUA-DataFarm: LLM Agent as a Cost-Efficient Teleoperator for Real-Ready Robot Data

👍 24

Teleoperation remains the default way to collect robot demonstrations, but every episode costs robot and operator time, and every new task starts from scratch. We ask whether a coding agent can take over the teleoperator's job. In our DataFarm, the agent receives a data request and one calibration e

中文介绍 提出PyRUA-DataFarm,用代码智能体替代人工遥操作,依据数据请求和校准示例自动生成机器人操作代码,低成本收集真实世界演示数据,新任务无需从零开始,显著减少人工成本。

StreamDuo: A Live Commentator and a Research Agent for Unseen Games

👍 23

Streaming video LLMs can now commentate live gameplay, but they are trained and tested on the same games. We study cross-game streaming commentary, in which a model trained on one game commentates games it has never seen, and find that current commentators transfer their timing but not their knowled

中文介绍 研究跨游戏流式解说场景,发现现有模型虽能迁移时间节奏但缺乏知识迁移。提出StreamDuo框架,结合实时解说员与研究智能体,对未见游戏进行有效评论与知识注入。

EmoReward: Self-Evolving Reward Calibration for Emotional Video Generation RL

👍 22

Group Relative Policy Optimization (GRPO) has advanced generative model post-training in verifiable domains such as code and mathematics, but evaluating video quality, especially emotional expression, often depends on subjective aesthetics and has no uniquely correct target. Existing approaches typi

中文介绍 针对情绪视频生成中GRPO因主观审美而缺乏唯一正确目标的问题,提出EmoReward,自进化奖励校准机制,通过反馈逐步调整奖励信号,提升生成视频的情绪表达准确性与一致性。

On the Generalization Limitation of On-Policy Distillation for Post-RL Consolidation

👍 16

Multi-teacher on-policy distillation (MOPD) extends on-policy distillation (OPD) to consolidate capabilities from multiple teachers. Although prior work reports near-perfect in-distribution (IND) distillation performance, we find that this can mask incomplete capability consolidation beyond the supe

中文介绍 发现多教师在线策略蒸馏在分布内看似完美,但掩盖了能力整合不足的问题。系统研究其泛化局限性,证明需扩展评估范围才能确保跨任务能力真正被蒸馏,为后训练整合提供警示。

TAC-CoT: Tool-Empowered Implicit Chain-of-Thought via Interleaving Explicit Tokens

👍 14

LLM-based agentic systems are increasingly combined with CoT reasoning to solve tasks that require accurate reasoning, external knowledge, and multi-step deliberation. Yet, explicit CoT incurs substantial generation overhead, while implicit CoT, which replaces textual rationales with compact continu

中文介绍 提出TAC-CoT,将显式CoT令牌与隐式推理交替穿插,利用工具增强推理能力,在保持隐式高效性的同时提升多步任务的准确性,降低生成开销,平衡效率与性能。

MUON-RT Reusing the Teacher's Trajectory for Momentum-Free Muon for LLM Self-Distillation

👍 14

Self-distillation methods for large language models with an exponential moving average (EMA) teacher already retain a history of student updates. Muon nevertheless stores a separate momentum buffer, increasing persistent training memory. We propose Muon with Reused Teacher (Muon-RT), which reuses th

中文介绍 提出Muon-RT,复用EMA教师的轨迹作为优化器动量,消除Muon的独立动量缓冲区,在LLM自蒸馏中减少持久内存占用,同时保持收敛速度和性能。

SpecOPD: Branch-Parallel On-Policy Distillation for Speculative Decoding

👍 14

Speculative decoding (SD) accelerates large language model inference by reducing the sequential cost of token-by-token generation. Advanced SD methods reduce proposal cost with parallel drafters that predict entire blocks in one forward pass, but overall speedup still depends on acceptance length. O

中文介绍 提出SpecOPD,分支并行在线策略蒸馏,用于投机解码,训练多个并行草稿分支以优化接受长度,减少token顺序生成开销,加速LLM推理同时保持输出质量。

VideoMemBench: How Well Do MLLMs Remember in Video Conversations?

👍 13

Multimodal large language models (MLLMs) can now process hour-long videos, making extended multi-turn dialogue over a single video a natural mode of use. Sustaining such a dialogue requires more than answering each question in isolation: the model must retrieve and reuse information established earl

中文介绍 提出VideoMemBench,评估多模态LLM在长视频多轮对话中的记忆能力,测试模型能否检索并复用早期信息,发现现有模型在长时间对话中记忆退化严重,为视频对话系统提供基准。

Human-to-Robot Transfer via Bridging Representation and Action Space in VLAs

👍 13

Pretrained Vision-Language-Action (VLA) policies provide a strong starting point for deployment, but adapting them to a target robot and environment still requires costly embodiment-specific demonstrations. Human demonstrations are far easier to collect, yet directly leveraging them is challenging d

中文介绍 针对VLA策略需特定于具体机器人的演示数据问题,提出桥接表示与动作空间的方法,利用人类演示数据对齐到目标机器人,减少对昂贵本体演示的依赖,提升泛化与部署效率。

Every Token Needs Its Own Teacher: Multi-teacher Adaptive On-policy Self-distillation

👍 13

On-policy Self-distillation (OPSD) has proven effective for post-training large language models by constructing a teacher conditioned on privileged references. However, existing approaches keep the teacher fixed either throughout training by consistently exposing it to the full reference, or within

中文介绍 提出多教师自适应在线策略自蒸馏,每个令牌使用不同教师,按需动态调整蒸馏强度,避免固定教师导致的学习偏差,增强LLM后训练中的能力整合与分布外泛化。

Dory1.0-4B: A Looped Normalized-GPT

👍 11

We introduce Dory1.0-4B, a 4-billion-parameter looped (latent-recurrent) language model with elastic depth: a single checkpoint serves different inference compute budgets. Dory1.0-4B is built on a normalized Transformer (nGPT) backbone with interleaved sliding-window and full-attention layers, organ

中文介绍 发布Dory1.0-4B,一种40亿参数的循环潜在递归语言模型,基于归一化Transformer并交织滑窗与全注意力,支持弹性深度,单个检查点可适应不同推理算力预算。

Hybrid Computer Use: A Unified Programmatic, Semantic, and Visual Interface for General Agents

👍 10

Computer-use agents increasingly tackle tasks ranging from everyday desktop activities to long-horizon professional workflows. Yet interfaces centered on screenshots and mouse-and-keyboard actions can impose a human-interaction prior, requiring repeated screen interpretation while leaving the comput

中文介绍 提出Hybrid Computer Use统一接口,结合程序化、语义与视觉操作,让智能体直接调用计算机功能而非仅模拟鼠标键盘,提升长程任务效率并减少屏幕解读负担。

PyRUA-REAL: Robot Exploration and Adaptive Learning through RSI in the Real World

👍 10

Coding agents powered by models such as GPT-6-Astra connect spatial reasoning with executable robot behavior, raising the question of how such agents can produce lasting improvements in robot capabilities from physical interaction. In this work, we study embodied recursive self-improvement (RSI) thr

中文介绍 通过GPT-6-Astra驱动的编码智能体在真实机器人上实现具身递归自我改进,使智能体从物理交互中持续学习并提升能力,探索自主改进机器人技能的新路径。

The SIMPLE Law for Energy Cost of LLM Inference

👍 9

Generative AI is billed per token, yet tokens are not equally expensive. Each token enters the KV-cache and raises the cost of those that follow, so energy grows non-linearly along a sequence whose tokens look identical. No analytical model states what a token costs, and the most accurate SOTA model

中文介绍 提出SIMPLE定律,解析描述LLM推理中能量成本与序列位置的关系,每token的KV缓存成本累积导致能耗非线性增长,为推理计费与优化提供精确模型。

LoopVQ: Sub 1-bit KV Cache Quantization for Looped Transformers

👍 9

Looped transformers are a parameter-efficient alternative to standard transformers for language modeling. By reapplying the same layers across multiple loop steps, recent looped language models match the performance of non-looped models several times their size. While this architecture effectively r

中文介绍 提出LoopVQ,面向循环Transformer的KV缓存量化方法,在保持模型性能的同时实现亚1比特量化,显著降低循环语言模型的显存占用,提升长序列推理效率。

Building a Real Agent OS for Teams: Lessons from Ankole

@AgentBullLab · 20 粉丝 · 1.4M 阅 · 501 赞 · 19 转

1. From Single-User Harnesses to Collective Cognition On a solo developer's laptop, tools like Claude Code, Codex, and OpenClaw can already work autonomously for hours. Drop that same single-user

中文介绍 分享从单人智能体到团队智能体系统的演进经验,以 Ankole 为例。指出 Claude Code、Codex 等工具在开发者笔记本上可自主工作数小时,但扩展到团队需要构建「智能体操作系统」,实现集体认知与协作编排。

How to Use Grok Bot to Build a Team of AI Trading Agents

@RohOnChain · 84.8K 粉丝 · 405.8K 阅 · 525 赞 · 72 转

I will break down the EXACT framework to build a 24/7 AI trading team with Grok Bot that researches the market, finds strategies, backtests & trades on its own. Let's get straight to it. Bookmark This

中文介绍 提供用 Grok Bot 构建 24/7 AI 交易团队的完整框架:市场研究、策略发现、回测到自动交易。教程形式,含具体步骤,面向加密货币/量化交易场景。

XPENG and Academic Partners Release GroundingPI, a Pretrained Model for Embodied Perception

@XPENG_Global · 432.1K 粉丝 · 349.3K 阅 · 501 赞 · 30 转

XPENG and Academic Partners Release GroundingPI, a Pretrained Model for Embodied Perception, Outperforming GPT-6 Astra in Overall Perception Benchmarks XPENG, in collaboration with research teams from

中文介绍 小鹏与高校团队发布具身感知预训练模型 GroundingPI,在整体感知基准上超越 GPT-6 Astra。强调模型在机器人/具身智能场景的感知能力提升。

The Future Is the Agentic Internet: Moving Beyond Personal Agents to Team Agents

@Teamily_AI · 9.4K 粉丝 · 257.8K 阅 · 534 赞 · 53 转

Dear Teamily AI Customers, Ecosystem Partners, and Investors, Over the past few months, Teamily AI has grown rapidly, and we're grateful to everyone who has joined us on this journey. In this letter,

中文介绍 致客户信阐述「智能体互联网」愿景:从个人智能体走向团队智能体。强调多智能体协作网络将成为未来范式,并分享公司近期增长与生态进展。

Rented Brains

@HeyElsaAI · 80.6K 粉丝 · 107.7K 阅 · 500 赞 · 39 转

Three companies own how the world thinks. Here is the way out. A HeyElsa report introducing Infero · October 2026 Part one: how we got here AI, in plain words Forget the jargon. An AI model is a very

中文介绍 报告指出三家巨头控制着全球思维模式,提出出路在于新模型 Infero。拆解 AI 本质,反思知识集中化问题,主张通过替代性模型分散智能控制权。

Toolmakers

@getpeid · 835.7K 粉丝 · 50.0K 阅 · 539 赞 · 74 转

Every few generations, a technology arrives that changes how everyone lives and works. Fire. The printing press. Electricity. The internet. Now, intelligence. It's the most important technology of our

中文介绍 将 AI 定位为继火、印刷术、电力、互联网之后改变一切的技术。类比历代工具革命,强调智能是「最重要的技术」,探讨工具制造者与使用者的角色。

Agent State Machines: Building Deterministic Scaffolding for Agents

@JoshARosen · 3.6K 粉丝 · 32.1K 阅 · 522 赞 · 39 转

If you're not using some type of agent state machine, you're completely at the mercy of the nondeterministic nature of models. Are you really okay with models deciding what actions to take and when

中文介绍 主张用智能体状态机构建确定性脚手架,对抗模型不可预测性。若不用状态机,智能体决策完全受模型非确定性摆布,无法控制动作时机与流程。

Critical thinking when agents do the work

@addyosmani · 476.7K 粉丝 · 29.5K 阅 · 680 赞 · 71 转

Back in the day, critical thinking was taught as approaching new info with humble curiosity and a healthy dose of skepticism - take in different views, form your own opinion. Still true, but

中文介绍 反思智能体代劳后的批判性思维:传统上以谦逊好奇和怀疑审视新信息,如今需适应新场景——如何在智能体输出基础上保持独立判断,避免过度依赖自动化。

Anthropic's $2 Trillion IPO: Who Gets Rich From the AI We Helped Build?

@ActionModelAI · 56.1K 粉丝 · 5.3K 阅 · 503 赞 · 412 转

Anthropic could be heading toward one of the biggest IPOs in history. Founders, employees and investors could make billions. But what about the people whose knowledge helped make AI valuable in the

中文介绍 探讨 Anthropic 可能迎来的 2 万亿美元 IPO 财富分配:创始人、员工、投资者或成亿万富翁,但质疑数据提供者——那些帮助构建 AI 价值的知识贡献者能否获益。

How Primus zkTLS Makes Web Data Verifiable for AI Agents

@Louisfabulous2 · 2.5K 粉丝 · 4.1K 阅 · 584 赞 · 38 转

AI agents are becoming more capable, but intelligence alone isn't enough to make them trustworthy, especially when money is involved. Imagine an AI agent checking Binance for an asset's current price

中文介绍 介绍 Primus zkTLS 如何让 AI 智能体验证网页数据。以智能体查询币安价格为例,说明零知识 TLS 使数据来源可验证,解决智能体涉及金钱交易时的信任问题。

Cult.fit builds operational muscle with Codex

Cult.fit is using Codex to help teams across functions build tools, move faster, and change how work gets done.

中文介绍 Cult.fit利用OpenAI的Codex,帮助团队跨职能构建工具、加快工作流程并改变工作方式。

Australian universities are creating AI-native graduates

Australian universities are expanding access to ChatGPT and bringing AI into teaching and learning to prepare students for an AI-enabled workforce.

中文介绍 澳大利亚多所大学扩大ChatGPT使用范围,将AI融入教学与学习,为培养适应AI职场的毕业生做准备。

GPT‑6 Cyber: Advancing the frontier of aligned cybersecurity agents

Introducing GPT‑6 Cyber, OpenAI’s cybersecurity model for finding exploitable vulnerabilities and developing fixes, trained to stay within authorized scope.

中文介绍 OpenAI发布网络安全模型GPT-6 Cyber,专注于发现可利用漏洞并开发修复方案,训练时限制在授权范围内。

Merge builds product and business tools with OpenAI technology

Merge uses GPT-6 Astra, ChatGPT, and Codex to review product code and support marketing. Its teams also use AI to plan executive dinners and prepare sales updates.

中文介绍 Merge使用GPT-6 Astra、ChatGPT和Codex进行产品代码审查与营销支持,并用AI规划高管晚宴和销售更新。

[AINews] It's Swapping Season

November 4, 2026 — OpenAI’s Linux kernel security work, Kimi K3.1 Pro’s open weights, and Ricursive’s Ripple-65B.

中文介绍 2026年11月4日AI新闻:OpenAI的Linux内核安全工作、Kimi K3.1 Pro开源权重、Ricursive的Ripple-65B模型。

Setting a new standard for responsible AI infrastructure

OpenAI introduces its National Community Compact and Compute Campus Policy Blueprint to protect local resources, create jobs, and guide responsible AI infrastructure development.

中文介绍 OpenAI推出国家社区契约和计算园区政策蓝图,以保护地方资源、创造就业并引导负责任AI基础设施建设。

Flipsnack grows product signups with conversion optimization

Flipsnack focused its ChatGPT Ads campaigns on getting people to try its product, growing registrations while lowering the cost per signup.

中文介绍 Flipsnack通过ChatGPT广告活动优化转化,聚焦吸引用户试用产品,在降低每注册成本的同时提升注册量。

Franklin Templeton uses OpenAI to support investment research

Starting with equity research, Franklin Templeton is developing AI workflows to help portfolio managers compare research views and examine investment risks.

中文介绍 Franklin Templeton从股票研究入手,开发AI工作流,帮助投资组合经理比较研究观点并评估投资风险。

Texas was the new epicenter of the AI data center buildout. That might be kaput.

Texas is the fastest-growing data center market in the US and ground zero for Big Tech’s largest AI buildouts. But the blue wave that swept the Lone Star State may jeopardize that position. Democrat James Talarico won the US Senate seat against Ken Paxton. Gina Hinojosa won the race for governor aga

中文介绍 得克萨斯州曾是全美增长最快的AI数据中心市场,但民主党人James Talarico赢得参议院席位后,蓝色浪潮或危及该地位。

OpenAI to acquire Pierre Computer Company, Inc.

OpenAI will acquire Pierre Computer Company, expanding Codex’s tools for building, understanding, and reviewing code.

中文介绍 OpenAI将收购Pierre Computer Company,以扩展Codex在构建、理解和审查代码方面的工具。

市场总览

美股整体承压,SPY 跌 4.41%(RSI 33.5),QQQ 跌 6.49%(RSI 26.9 超卖),NVDA 跌 9.26% 与 TSLA 跌 10.77% 领跌,科技股动能转弱。加密市场同样走低,BTC 跌 5.38% 至 76077 美元(RSI 38.8),ETH 跌 7.58% 至 2238 美元,恐慌贪婪指数 28 处于恐慌区,总市值 2.58T 美元(24h -7.41%),BTC 主导率 59.2%。中概股逆势偏强,京东涨 3.28%(RSI 69.4),拼多多涨 2.19%(RSI 61.5),阿里巴巴涨 2.64%,均获 MACD 金叉支持。商品外汇方面,黄金期货跌 0.95% 但近 5 日涨 4.51%(RSI 47.1),原油跌 0.83%(RSI 42.9),美元兑人民币微跌 0.07%,美元指数微涨 0.13%(RSI 50.1)。VIX 暴涨 69.66% 至 28.18,反映市场恐慌急升;10Y 美债收益率跌至 4.84%(RSI 34.6)。整体技术面风险偏好下降,多数资产呈偏空或中性状态。

今日关注

NVDA Nvidia (NVDA)
偏下行

NVDA 现价 181.60 美元,RSI 14 为 25.4,处于超卖区;价格低于 SMA20(213.68)及 SMA50(220.64),且低于 SMA200(204.66),呈空头排列。MACD 为 -7.2216,信号线 -4.2415,柱状图负值扩大,显示下行动能增强。近 1 日跌幅 9.26%,近 5 日跌幅 15.01%,距 52 周高点 -25.38%,技术面整体偏空。

QQQ Nasdaq 100 ETF (QQQ)
偏下行

QQQ 现价 647.05 美元,RSI 14 为 26.9,明确超卖;价格低于 SMA20(712.4)与 SMA50(722.27),但仍高于 SMA200(680.71)。MACD -14.5519 低于信号线 -8.3324,柱状图负值走阔,显示下行趋势强化。近 1 日下跌 6.49%,近 5 日下跌 9.55%,技术指标整体呈偏空信号。

AAPL Apple (AAPL)
偏上行

AAPL 现价 344.17 美元,RSI 14 为 53.0,处于中性偏强区间;价格高于 SMA20(343.1)、SMA50(335.46)及 SMA200(298.34),呈多头排列。MACD 为 2.2923,信号线 3.0269,柱状图负值但绝对值小于信号,显示短期动量偏正。近 1 日仅跌 1.53%,近 5 日跌 1.36%,相对抗跌,技术面偏上行。

JD 京东 (JD)
偏上行

JD 现价 28.64 美元,RSI 14 为 69.4,接近超买但尚未进入;价格高于 SMA20(26.46)与 SMA50(26.87),且贴近 SMA200(28.56)。MACD 为 0.0562,信号线 -0.2592,柱状图转正,形成金叉。近 1 日上涨 3.28%,近 5 日上涨 11.74%,动能明显偏强,技术面呈上行倾向。

^VIX VIX 恐慌指数 (^VIX)
中性

^VIX 现价 28.18,RSI 14 为 69.6,接近超买;价格高于 SMA20(18.29)、SMA50(16.7)及 SMA200(18.35),显示波动率显著抬升。MACD 1.5035 高于信号线 0.8788,柱状图正,形成金叉。近 1 日大涨 69.66%,近 5 日涨 63.84%,但处于高位,中性对待,等待回归均值。

全部资产

^VIX

VIX 恐慌指数

$28.18 +69.66%
5 日
+63.84%
距 52w 高
-20.2%
RSI(14)
69.6
趋势
中性
SMA 20 / 50 / 200
18.29 / 16.70 / 18.35
MACD / 信号
1.503 / 0.879
MACD 金叉 (4 天前)

^TNX

10Y 美债收益率 (%)

$4.84 -1.16%
5 日
-3.05%
距 52w 高
-9.8%
RSI(14)
34.6
趋势
中性
SMA 20 / 50 / 200
5.05 / 5.03 / 4.56
MACD / 信号
-0.053 / -0.024

DX-Y.NYB

美元指数 DXY

$101.68 +0.13%
5 日
-0.19%
距 52w 高
-1.2%
RSI(14)
50.1
趋势
多头
SMA 20 / 50 / 200
101.94 / 101.08 / 99.70
MACD / 信号
0.103 / 0.216
接近 52 周高多头排列

SPY

S&P 500 ETF

$733.00 -4.41%
5 日
-4.73%
距 52w 高
-6.2%
RSI(14)
33.5
趋势
中性
SMA 20 / 50 / 200
765.53 / 766.40 / 730.15
MACD / 信号
-4.530 / -1.796

QQQ

Nasdaq 100 ETF

$647.05 -6.49%
5 日
-9.55%
距 52w 高
-15.2%
RSI(14)
26.9
趋势
中性
SMA 20 / 50 / 200
712.40 / 722.27 / 680.71
MACD / 信号
-14.552 / -8.332
RSI 超卖

AAPL

Apple

$344.17 -1.53%
5 日
-1.36%
距 52w 高
-3.2%
RSI(14)
53.0
趋势
多头
SMA 20 / 50 / 200
343.10 / 335.46 / 298.34
MACD / 信号
2.292 / 3.027
多头排列

MSFT

Microsoft

$535.93 -3.58%
5 日
+1.09%
距 52w 高
-4.2%
RSI(14)
54.4
趋势
多头
SMA 20 / 50 / 200
531.63 / 516.13 / 439.45
MACD / 信号
4.702 / 5.104
多头排列

NVDA

Nvidia

$181.60 -9.26%
5 日
-15.01%
距 52w 高
-25.4%
RSI(14)
25.4
趋势
中性
SMA 20 / 50 / 200
213.68 / 220.64 / 204.66
MACD / 信号
-7.222 / -4.242
RSI 超卖

GOOGL

Alphabet

$282.57 -2.20%
5 日
-0.44%
距 52w 高
-30.8%
RSI(14)
28.0
趋势
空头
SMA 20 / 50 / 200
320.79 / 334.05 / 339.74
MACD / 信号
-16.547 / -11.838
死叉(SMA50↓SMA200) (5 天前)RSI 超卖空头排列

TSLA

Tesla

$343.69 -10.77%
5 日
-14.78%
距 52w 高
-31.1%
RSI(14)
35.8
趋势
空头
SMA 20 / 50 / 200
381.65 / 372.07 / 384.32
MACD / 信号
3.794 / 6.745
MACD 死叉 (今天)空头排列

META

Meta

$605.08 -3.96%
5 日
-3.11%
距 52w 高
-22.4%
RSI(14)
34.9
趋势
中性
SMA 20 / 50 / 200
668.10 / 674.14 / 632.13
MACD / 信号
-23.454 / -16.343
加密恐慌贪婪
28
恐慌
加密总市值
$2.58 T
-7.41% / 24h
BTC 主导率
59.2%
ETH 10.6%
24h 成交量
$104.8 B
活跃币 22,809

BTC-USD

Bitcoin

$76,077.53 -5.38%
5 日
-1.75%
距 52w 高
-29.2%
RSI(14)
38.8
趋势
中性
SMA 20 / 50 / 200
79,932.41 / 81,990.73 / 73,192.11
MACD / 信号
-1,252.128 / -996.369

ETH-USD

Ethereum

$2,238.82 -7.58%
5 日
-0.62%
距 52w 高
-38.8%
RSI(14)
39.8
趋势
中性
SMA 20 / 50 / 200
2,340.44 / 2,503.00 / 2,156.67
MACD / 信号
-57.014 / -61.334
MACD 金叉 (1 天前)

SOL-USD

Solana

$90.19 -5.55%
5 日
-5.41%
距 52w 高
-47.4%
RSI(14)
29.8
趋势
中性
SMA 20 / 50 / 200
99.58 / 108.41 / 88.96
MACD / 信号
-4.278 / -3.645
RSI 超卖

BABA

阿里巴巴 (BABA)

$108.07 +2.64%
5 日
+8.57%
距 52w 高
-40.3%
RSI(14)
55.3
趋势
空头
SMA 20 / 50 / 200
105.09 / 108.43 / 124.98
MACD / 信号
-1.884 / -2.475
MACD 金叉 (1 天前)空头排列

PDD

拼多多 (PDD)

$83.95 +2.19%
5 日
+5.89%
距 52w 高
-39.7%
RSI(14)
61.5
趋势
中性
SMA 20 / 50 / 200
80.44 / 79.86 / 90.58
MACD / 信号
0.390 / -0.078
MACD 金叉 (3 天前)

JD

京东 (JD)

$28.64 +3.28%
5 日
+11.74%
距 52w 高
-16.5%
RSI(14)
69.4
趋势
中性
SMA 20 / 50 / 200
26.46 / 26.87 / 28.56
MACD / 信号
0.056 / -0.259
MACD 金叉 (3 天前)

0700.HK

腾讯控股 (0700.HK)

HK$427.40 -0.70%
5 日
+11.42%
距 52w 高
-35.4%
RSI(14)
59.7
趋势
中性
SMA 20 / 50 / 200
403.83 / 422.05 / 474.30
MACD / 信号
-4.040 / -9.141
MACD 金叉 (3 天前)

GC=F

黄金期货

$4,074.60 -0.95%
5 日
+4.51%
距 52w 高
-27.1%
RSI(14)
47.1
趋势
空头
SMA 20 / 50 / 200
4,080.70 / 4,230.42 / 4,501.49
MACD / 信号
-53.454 / -65.302
MACD 金叉 (1 天前)空头排列

CL=F

WTI 原油期货

$91.27 +0.83%
5 日
-12.34%
距 52w 高
-23.6%
RSI(14)
42.9
趋势
中性
SMA 20 / 50 / 200
98.04 / 95.48 / 87.16
MACD / 信号
-0.362 / 1.123
MACD 死叉 (4 天前)

USDCNY=X

美元 / 人民币

¥6.67 -0.07%
5 日
+0.11%
距 52w 高
-7.3%
RSI(14)
39.1
趋势
空头
SMA 20 / 50 / 200
6.68 / 6.70 / 6.79
MACD / 信号
-0.010 / -0.011
MACD 金叉 (今天)接近 52 周低空头排列
风险提示

本报告仅基于公开行情数据的技术指标解读,不构成任何投资建议或买卖信号。过去走势不代表未来表现,市场波动可能随时变化,投资者应结合自身风险承受能力独立判断。

Paraguay’s President, Santiago Peña, Quits Amid Ueno Bank Scandal

The president of the South American nation, Santiago Peña, has apologized for failing to prevent its worst financial crisis in decades.

中文摘要 巴拉圭总统圣地亚哥·佩尼亚因乌埃诺银行丑闻辞职,并为未能阻止该国数十年来最严重的金融危机致歉。

Beau Lamarre-Condon trial live updates: prosecution begins cross-examination of double murder accused

Lamarre-Condon returns to witness box for day two of his evidence. He has pleaded not guilty to murdering the couple in Sydney in 2024 Beau Lamarre-Condon tells double murder trial shooting of Jesse Baird wasn’t intentional but ‘it was my fault’ Get our breaking news email, free app or daily news po

中文摘要 博·拉马尔-康登双重谋杀案庭审进入第二天,控方开始交叉盘问。他否认2024年在悉尼杀害杰西·贝尔德和卢克·戴维斯,但称射击并非故意,"是我的错"。

Liberal bus crashes into shop awning on its first outing of Victorian election campaign

Opposition leader Jess Wilson grimaced as the ‘Jess Express’, carrying Liberals and reporters, backs into the structure in suburban Melbourne Get our breaking news email, free app or daily news podcast The Victorian Liberal party’s first offical day on the campaign trail has hit a snag, after its ca

中文摘要 维州选举竞选首日,自由党领袖杰斯·威尔逊的竞选巴士"杰斯快车"在墨尔本郊区倒车时撞上商店遮阳篷,威尔逊面露苦相。

Gaza turns discarded cooking pots into seawalls

As winter approaches, Palestinians in Gaza are turning damaged cooking pots into makeshift seawalls to avoid flooding.

中文摘要 随着冬季临近,加沙的巴勒斯坦人将废弃的烹饪锅改造成临时海堤,以防洪水侵袭。

Beau Lamarre-Condon tells double murder trial shooting of Jesse Baird wasn’t intentional but ‘it was my fault’

Former NSW police officer gives evidence for a second day after pleading not guilty to murdering Baird and Luke Davies in early 2024 Beau Lamarre-Condon trial live updates: former police officer continues evidence in double murder trial Beau Lamarre-Condon has said he is “guilty of manslaughter” but

中文摘要 前新州警官博·拉马尔-康登在双重谋杀案庭审作证第二天表示,射杀杰西·贝尔德并非故意,但"是我的错"。他否认2024年初谋杀贝尔德和卢克·戴维斯。

Democratic socialist defeats incumbent Karen Bass in Los Angeles mayor's race

The results mean America's two most populous cities - New York and Los Angeles - will be led by democratic socialists.

中文摘要 民主社会主义者击败现任市长卡伦·巴斯,赢得洛杉矶市长选举。至此,美国人口最多的两座城市纽约和洛杉矶均由民主社会主义者领导。

Peru airport officials discover 3,000 tarantulas smuggled inside dirty socks

Three suitcases’ worth of the spiders were on their way to Germany, where they could fetch an estimated €1.5m More than 3,000 tarantulas were seized at an airport in Lima on their way to Germany, Peruvian authorities said. The spiders were hidden inside dirty socks contained in three suitcases carri

中文摘要 秘鲁利马机场查获逾3000只藏在脏袜子中的狼蛛,这些蜘蛛被藏在三个行李箱内,正运往德国,估计价值约150万欧元。

Trump signs order preventing future presidents from changing Oval Office

The order comes after Trump made a series of big changes the Oval Office decor and launched landmark construction projects across Washington.

中文摘要 特朗普签署命令,禁止未来总统更改椭圆形办公室装饰,此前他对办公室进行了大刀阔斧的改造并启动了华盛顿的地标建设项目。

Turkiye’s parliament ratifies Mecca defence agreement

Parliament's approval lays the groundwork for further Turkish involvement in Saudi Arabia's conflict with the Houthis.

中文摘要 土耳其议会批准《麦加防务协议》,为土耳其进一步介入沙特与胡塞武装的冲突铺平道路。

Trump met Russian envoy after he appeared with Melania Trump and Ukrainians

Trump again said a major Russia-Ukraine deal is 'very close', but gave no specifics.

中文摘要 特朗普在俄罗斯特使与梅拉尼娅·特朗普及乌克兰人一同露面后会晤了该特使,再次表示一项重大的俄乌协议"即将达成",但未透露细节。

Global tech stocks plunge as US midterms add to AI slowdown fears

Investors have reacted to possible tougher scrutiny of data centre construction and the AI industry.

中文摘要 全球科技股大幅下挫,因美国中期选举结果加剧市场对AI行业及数据中心建设可能面临更严格审查的担忧。投资者对AI相关股票的估值和监管风险感到不安。

China to Cut Diesel Exports on Tight Stocks, Energy Aspects Says

China’s state-run refiners are expected to reduce diesel exports this month because of tight inventories, although they could relax those curbs should domestic demand soften, according to Energy Aspects.

中文摘要 能源咨询机构Energy Aspects表示,由于库存紧张,中国国有炼油厂预计将在本月减少柴油出口。但若国内需求走软,这些限制措施可能会有所放松。

Asian tech stocks sink

‘Complete carnage’ as Nikkei 225 falls below 60,000 points for first time since May with Seoul’s Kospi under 5,000 points

中文摘要 亚洲科技股暴跌,日经225指数自5月以来首次跌破60,000点,首尔Kospi指数也跌至5,000点下方,市场形容为“完全屠杀”。

Hedge Fund Samlyn Dumps All Stocks But Financials, Cuts Staff

Rob Pohly’s Samlyn Capital is dismissing senior analysts after selling off its exposure to nearly every industry except global financials, as it pivots to focus on that sector.

中文摘要 对冲基金Samlyn Capital抛售几乎所有行业的股票持仓,仅保留全球金融股,并解雇了资深分析师。该基金由Rob Pohly管理,正转向专注于金融板块。

John Woods Sees AI Selloff as Positioning-Driven

John Woods says the sharp pullback in AI-exposed stocks is a positioning-driven correction rather than a sign of deteriorating market fundamentals. (Source: Bloomberg)

中文摘要 John Woods表示,近期AI相关股票的急剧回调是仓位驱动型的修正,而非市场基本面恶化的迹象。他认为这是投资者减仓所致。

Wall Street Traders Wrong-Footed After Bullish Equity Calls

Wall Street’s biggest trading desks spent weeks saying the stock-market rally would resume once the election was done. On Thursday, that call backfired spectacularly.

中文摘要 华尔街各大交易台此前数周一直押注美股涨势将在选举后重拾升势,但周四这一预期遭遇惨败。市场走势令许多看涨交易员措手不及。

Copper Moves Lower as US Election Undermines AI and Tariff Trade

Copper slumped to its lowest since July as a political shakeup in Washington threatened to blunt two key bullish drivers: the rapid buildout of AI infrastructure, and the risk of US import tariffs on the metal.

中文摘要 铜价跌至7月以来最低。美国政治局势变化削弱了两大看涨因素:AI基础设施的快速建设,以及美国对铜征收进口关税的风险。选举结果冲击了此前押注这两条主线的交易。

Taiwan’s Currency Weakens Past 33 Per US Dollar on AI Jitters

The Taiwan dollar weakened past the key psychological level of 33 per US dollar for the first time since April 2025, as a broad retreat in artificial intelligence shares weighed on the island’s tech-heavy equity market.

中文摘要 台币兑美元自2025年4月以来首次跌破33的心理关口,因人工智能股票普遍回落拖累台湾以科技股为主的大盘。市场避险情绪上升。

US lawmakers summon Jensen Huang over Chinese spying concerns

Top Republicans ask Nvidia boss to brief congressional committees after reports he was targeted by China

中文摘要 美国共和党高层要求英伟达CEO黄仁勋向国会委员会简报。此前有报道称黄仁勋被中国间谍机构列为目标,引发了对华安全担忧。

China’s 10-Year Yield Hits Record Intraday Low on Haven Demand

China’s benchmark 10-year sovereign yield dropped to an all-time intraday low, driven by haven demand and a recent shift in the policy rate that was regarded as de facto easing.

中文摘要 中国10年期国债收益率跌至盘中历史新低,受避险需求推动,且近期政策利率调整被视为事实上的宽松举措。市场资金面充裕。

Putin Transfers Japan Tobacco Assets to Local Firm in Latest Asset Seizure

Japan said it has protested Russia’s seizure of Japan Tobacco International’s local assets, the latest development in an escalating diplomatic spat between the two nations.

中文摘要 俄罗斯总统普京将日本烟草国际在当地的资产移交给一家本地公司。日本对此表示抗议,称这是两国间不断升级的外交争端的最新发展。

该源今日无内容。