Go · ★ 34,626 · 🍴 2,461 · 📈 3,290 stars today
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 阿里开源的代码审查工具，采用混合架构：确定性流水线负责规则检测，LLM Agent 补充语义分析，可输出精确到行的评论，并内置 NPE、线程安全等多语言规则集，适合在大规模仓库中接入 CI 做自动化评审。
JavaScript · ★ 10,498 · 🍴 561 · 📈 3,606 stars today
A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings
中文介绍 Cloudflare 推出的编码 Agent 技能，用于执行多阶段安全审计：先分阶段排查代码，再对结论做独立验证，最终输出机器可读的审计结果，适合在 CI 或本地让 AI Agent 自动跑安全审查的团队。
JavaScript · ★ 95,814 · 🍴 10,144 · 📈 680 stars today
Production-grade engineering skills for AI coding agents.
中文介绍 面向 AI 编码 Agent 的生产级工程技能集合，把代码审查、测试、调试等成熟工程实践封装成 Agent 可直接调用的技能，帮助 Claude Code、Cursor 等工具在真实项目中减少幻觉、提升交付质量。
TypeScript · ★ 4,070 · 🍴 288 · 📈 1,350 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，让 AI Agent 直接使用你已登录的真实浏览器，无需重新登录或中断当前工作。由 CLI 与浏览器扩展组成，可接入任何支持 shell 的 AI Agent，适合需要操作已有账号页面的自动化场景。
Rust · ★ 4,916 · 🍴 304 · 📈 940 stars today
Turn your coding agents into research agents
中文介绍 把编码 Agent 改造成研究 Agent，让 Claude Code 等工具在写代码之外还能检索文献、整理论据并完成调研任务，适合研究人员或工程团队做技术选型与资料收集。
TypeScript · ★ 145,845 · 🍴 23,618 · 📈 538 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 官方的终端内智能编码工具，能理解整个代码库，自动完成重复性任务、解释复杂代码并处理 git 工作流，开发者可在命令行中直接与它协作，无需切换编辑器。
Java · ★ 78,445 · 🍴 8,679 · 📈 912 stars today
Ghidra is a software reverse engineering (SRE) framework
中文介绍 美国国家安全局开源的软件逆向工程框架，提供反汇编、反编译、图形化分析与脚本扩展能力，广泛用于二进制漏洞分析、恶意样本研究和固件逆向，是 IDA 之外的主流免费替代方案。
Python · ★ 24,542 · 🍴 2,939 · 📈 287 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，主要面向在 Claude Cowork 中工作的知识工作者，提供文档处理、信息整理等任务的现成插件，安装后可直接扩展 AI 助手的办公能力。
Go · ★ 26,157 · 🍴 3,548 · 📈 1,123 stars today
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.
中文介绍 腾讯开源的 LLM 知识平台，把原始文档转换成可检索问答的 RAG、能自主推理的 Agent，以及可自我维护的 Wiki，适合企业搭建私有知识库与智能问答系统。
Swift · ★ 6,124 · 🍴 286 · 📈 738 stars today
Tinycast — a tiny, fully native macOS launcher, hotkeys, and clipboard history.
中文介绍 轻量的原生 macOS 启动器，集成全局热键与剪贴板历史记录，追求体积小、启动快、完全原生，适合希望用键盘快速唤起应用、检索历史复制内容的 Mac 用户。
Go · ★ 25,251 · 🍴 4,073 · 📈 153 stars today
eBPF-based Networking, Security, and Observability
中文介绍 基于 eBPF 的云原生网络、安全与可观测性平台，用内核层可编程能力替代传统 iptables 做服务间连通、网络策略与流量监控，是 Kubernetes 集群的主流 CNI 方案之一。
TypeScript · ★ 54,819 · 🍴 6,823 · 📈 665 stars today
The open-source AI voice studio. Clone, dictate, create.
中文介绍 开源 AI 语音工作室，支持声音克隆、语音听写与音频创作，把语音合成、转写等能力整合到统一界面，适合内容创作者和开发者快速生成或处理音频。
JavaScript · ★ 261,117 · 🍴 39,085 · 📈 1,173 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 AI 编码 Agent 的性能优化系统，通过技能、直觉、记忆与安全机制，配合研究优先的开发流程，提升 Claude Code、Codex、Cursor 等工具在复杂任务中的表现。
Python · ★ 50,791 · 🍴 4,828 · 📈 327 stars today
We write your reusable computer vision tools. 💜
中文介绍 Roboflow 维护的计算机视觉工具库，提供检测框绘制、区域统计、轨迹跟踪等可复用组件，兼容主流检测与分割模型，帮助开发者快速搭建视觉 demo 与后处理流程。
C · ★ 35,721 · 🍴 3,748 · 📈 872 stars today
Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
中文介绍 纯 C 编写、零依赖的推理引擎，通过把 MoE 专家权重从磁盘流式加载，让消费级硬件也能运行前沿混合专家大模型，适合显存有限但想本地跑大模型的开发者。
Python · ★ 3,408 · 🍴 352 · 📈 386 stars today
A smarter, self-hosted AI assistant — multi-user, multi-agent.
中文介绍 腾讯云推出的自托管 AI 助手，支持多用户与多 Agent 协作，可在自有服务器部署以保障数据私有，适合企业内部搭建可共享的智能助理与自动化工作流。
TypeScript · ★ 7,511 · 🍴 1,119 · 📈 469 stars today
Ever® Gauzy™ - Open Business Management Platform (ERP/CRM/HRM/ATS/PM) - https://gauzy.co
中文介绍 开源业务管理平台，把 ERP、CRM、HRM、招聘（ATS）与项目管理整合到一套系统中，支持自托管，适合中小企业用统一平台替代多套零散 SaaS 工具。
TypeScript · ★ 68,541 · 🍴 7,413 · 📈 381 stars today
Autonomous coding agent as an SDK, IDE extension, or CLI assistant.
中文介绍 自主编码 Agent，可作为 SDK、IDE 扩展或 CLI 助手使用，能读写文件、执行命令并逐步完成开发任务，开发者可按需选择集成方式接入现有工作流。
Go · ★ 14,805 · 🍴 1,485 · 📈 204 stars today
Secure environments for developers and their agents
中文介绍 为开发者和其 AI Agent 提供安全隔离的开发环境，通过远程工作区统一下发配置与权限，让代码和凭据不落在本地终端，适合需要合规与集中管理的工程团队。
TypeScript · ★ 204,943 · 🍴 60,742 · 📈 319 stars today
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
中文介绍 fair-code 的工作流自动化平台，原生支持 AI 能力，可在可视化画布中搭建流程并嵌入自定义代码，提供 400 多个集成，支持自托管或云端，适合做内部自动化与 AI 编排。
👍 294
Language models may need to internalize information that arrives over time and retain it through many subsequent updates. To study this challenge, we introduce long-horizon memorization, a setting in which a model learns 100 query-answer tasks through continual supervised fine-tuning without retaini
中文介绍 提出「长时程记忆」设定：模型在不保留原始数据的前提下，通过持续监督微调依次学习 100 个问答任务，考察信息随时间到达并被后续多次更新保留的能力。研究显示不同持续学习机制可组合使用，为长时程知识内化提供分析框架。
👍 113
Foundation models, alongside advances in learned game-world models, are reshaping AI across the game lifecycle. Beyond playing games, recent systems model players and game dynamics, support design and development, adapt player-facing experiences at runtime, and evaluate resulting artifacts. Yet thes
中文介绍 综述基础模型时代的游戏 AI：除对局本身外，系统还能建模玩家与游戏动态、辅助设计与开发、在运行时自适应调整玩家体验并评估生成产物。文章梳理这些方向的研究现状，指出游戏全生命周期中基础模型带来的新问题与机遇。
👍 102
Realtime spoken interaction demands deep reasoning, prompt responses, and fluid turn-taking. We present StepAudio 3 Realtime, an audio-language foundation model organized around a continuous listen-converse-think-act loop. Deep Perception captures rich acoustic cues to interpret user intent, while S
中文介绍 StepAudio 3 Realtime：面向实时语音交互的音频-语言基础模型，围绕「听-说-思-行」连续循环组织。其中 Deep Perception 捕捉丰富声学线索以理解用户意图，模型兼顾深度推理、快速响应与流畅的轮次转换。
👍 88
Recursive self-improvement (RSI) enables AI systems to turn experience and feedback into persistent changes that improve both their capabilities and the process of future improvement. We first use the Headroom-Closed Index (HCI) to reveal the problems of existing LLMs, then introduce the RSI concept
中文介绍 先用 Headroom-Closed Index（HCI）揭示现有 LLM 的局限，进而提出递归自我改进（RSI）概念：让 AI 系统把经验与反馈转化为持久改变，在提升自身能力的同时改进未来的改进过程，朝真正自主演化的智能迈进。
👍 76
We introduce StepAudio 3 Music, a large-scale, long-form music generation model that supports explicit musical planning and open-domain text-controlled generation. The StepAudio Music Tokenizer represents audio as a 50-Hz stream from a 65536-entry single codebook, using semantically informed self-su
中文介绍 StepAudio 3 Music：大规模长时音乐生成模型，支持显式音乐规划与开放域文本控制。其 Music Tokenizer 以 50Hz 单码本（65536 项）表示音频，并采用语义信息引导的自监督方式训练，实现可控的长曲目生成。
👍 70
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 ScienceIDE 把全球科学代码仓库转化为 agent 可学习的环境。针对工具链碎片化、领域约定隐式、正确性标准专门化等障碍，将数十年科研代码中的可执行模型、方法与工具转为可靠学习经验，缓解科学代码难以利用的问题。
👍 59
When we speak of recursive self-improvement (RSI), are we speaking of a phenomenon, a mechanism, or a prospect? Towards autonomous and evolving intelligence, RSI is being claimed at many scales, while no single framework that formally describes these emerging instances exists. Its counterpart in the
中文介绍 递归自我改进（RSI）在多个尺度上被广泛主张，却缺乏统一的形式描述。本文提出 Generalized Agent Iteration 框架，把迭代策略改进与递归自我改进纳入同一形式体系，为自主演化智能提供可分析的理论基础。
👍 48
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 现有置信度估计器共享同一设计前提，难以满足可信部署需求。本文提出「经验置信度」估计，让模型从推理到 agent 任务中积累的经验判断输出正确的概率，以支撑何时发布、升级或重试的决策。
👍 44
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 ProgramDistill：面向编码 agent 的新基准，要求 agent 从可运行的交互式 Web 应用推断目标行为，并在不完整的应用中实现。它把现成软件转化为可验证、带参考指引的软件工程任务，更贴近真实 Web 开发场景。
👍 39
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 Agora 把 Git 作为多个自动研究 agent 的共享记忆。单个 AutoResearch 循环可无人值守改进训练配置，但多个循环各自从零开始只会带来重复搜索；共享研究进展后，多 agent 协作才更可能转化为新发现而非重复劳动。
👍 30
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 EvolveTrade：面向自演化 LLM 交易 agent 的经验驱动策略精炼方法。针对部署前固定的手写工具使用策略难以适应市场变化的问题，让 agent 依据经验调整证据收集、工具调用与信号验证的方式，实现策略持续演化。
👍 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 HypoEvolve 用遗传算法驱动多 agent LLM 发现科学假设。系统通过批评、比较与修订进行演化搜索，并系统考察不同形式的 agent 协作对假设发现效果的影响，为科学假设自动生成提供可分析的设计依据。
👍 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 提出面向 LLM 偏好对齐的零阶范式。针对直接偏好对齐中似然位移的问题，该方法不再依赖似然信息，而以零阶优化从偏好对中提取对齐信号，在保持计算与内存效率的同时改善偏好对齐效果。
👍 20
We introduce and release ScienceBuddy, an interactive scientific research workspace that brings continually improving scientific agents into researchers' everyday workflows. ScienceBuddy supports researchers in carrying out scientific tasks while transforming their requests, feedback, and execution
中文介绍 ScienceBuddy 是交互式科研工作空间，把持续自我改进的科学 agent 嵌入研究者日常流程。它支持科研任务执行，同时将用户的请求、反馈与执行过程转化为改进信号，实现「递归中的递归」式自我提升。
👍 17
Embodied navigation requires agents to interpret visual observations, accumulate spatial knowledge, and execute actions to follow instructions or locate objects. Training-based methods face generalization challenges, while training-free methods exploit multimodal large language models (MLLMs) but of
中文介绍 HarnessVLN：以统一的 agent harness 整合免训练具身导航。基于训练的方法泛化受限，免训练方法虽借助 MLLM 却难以协调视觉观测、空间知识积累与动作执行；该框架统一这些环节，完成指令跟随与目标定位任务。
👍 15
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 探索用 VLM agent 实现机器人的上下文学习。有限示教无法覆盖机器人将遇到的全部任务与情境，因此在部署时从上下文中学习成为泛化的关键。该方法让 VLM agent 借助上下文快速适应陌生环境，向人类般的灵活适应靠近。
👍 10
Recent work extends recursive self-improvement (RSI) to agent harnesses for long-horizon coding and terminal tasks, enabling agents to improve execution mechanisms from experience. However, generalizable harness RSI remains challenging. First, evolving harnesses on evaluation benchmarks or their sub
中文介绍 ModularRSI：面向 agent harness 的模块化、可泛化递归自我改进。已有工作能让 agent 从经验改进执行机制，但在评测基准或其子集上演化容易过拟合、难以迁移；该方法通过模块化设计提升 harness 自我改进的通用性。
👍 9
Intelligent systems that act in the world require image understanding that is both comprehensive and spatially grounded. Current vision-language models (VLMs) can generate fluent and detailed image captions, but reliably associating them with image pixels remains challenging. Existing methods that c
中文介绍 PANORAMA：通过掩码提议选择实现全景接地的图像描述。当前 VLM 能生成流畅详细的描述，却难以可靠地与像素对应；该方法借助掩码提议的筛选，把文本描述与图像区域精确关联，兼顾描述的全面性与空间定位准确性。
👍 9
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 针对长上下文 MoE 训练的显存峰值问题：只要任一组件的峰值分配超过设备显存，训练即失败，因此目标是同时压低所有峰值而非平均占用。现有并行方案仍留下四个不受限的峰值且随规模增长，本文提出将其一并压平的方法。
👍 9
We demonstrate that training LLMs with RL does not improve performance equally across a dataset. RL shows large improvements on easy problems that an LLM is already good at solving, but small improvements on hard problems. We call this the Matthew Effect in RL for LLMs, after the phenomenon of cumul
中文介绍 揭示 RL 训练 LLM 中的「马太效应」：性能提升在数据上分布不均，模型本就擅长的简单问题提升显著，难题提升有限。作者据此提出「永不放弃」的训练思路，以改善模型在难题上的表现。
@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 微软 AI 负责人 Mustafa Suleyman 就「模型福祉」发声：他先明确当前 AI 没有意识、不会感受或遭受痛苦，再针对「人类或将对模型负有照护义务」这一逐渐兴起的运动阐述自己的看法。
@LingoAITech · 411.4K 粉丝 · 156.5K 阅 · 524 赞 · 227 转
From the Semantic Web to the Human-Centric Agentic Economy Data3.0 + Finance3.0: Sovereign Data, Personal AI, Open Agents and Programmable Value Abstract Artificial intelligence is changing the role
中文介绍 一篇论述 AI 原生 Web3 架构的长文：从语义网走向以人为中心的 agent 经济体，提出 Data3.0 加 Finance3.0 框架，涵盖数据主权、个人 AI、开放 agent 与可编程价值。
@jack · 12.1M 粉丝 · 153.5K 阅 · 2.2K 赞 · 304 转
the frontier is the edge of what we know. no company owns what comes next. i want more people to be able to advance it. i favor open releases that people can examine, use, and improve together without
中文介绍 Jack Dorsey 主张开放前沿：前沿即人类认知的边界，不该被任何公司独占。他支持开放发布，让更多人能够检视、使用并共同改进模型，而不是把下一步交给少数机构。
@zachlloydtweets · 20.4K 粉丝 · 123.1K 阅 · 500 赞 · 47 转
The software factory approach (a closed agentic loop that runs in the cloud) is growing in popularity, but it can be daunting to adopt. In this post I’ll go through “crawl, walk, run” steps for making
中文介绍 讲「软件工厂」模式的落地路径：把在云端持续运行的闭环 agent 循环，按「爬—走—跑」三阶段逐步引入团队，降低直接采用这种封闭 agent 循环的门槛与风险。
@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 17 人全员与 Hermes 协作——这是 NousResearch 开源的 AI agent harness，跑在德国一台物理机器上，作者分享小团队整建制部署与使用的实际做法。
@GoogleCloudTech · 1.3M 粉丝 · 43.8K 阅 · 505 赞 · 60 转
When developers first work on harness engineering for agentic coding systems, they often fall into the same trap: they run common end-to-end benchmarks like Terminal-Bench and DeepSWE, watch a
中文介绍 谷歌云讲 harness engineering：开发者评估 AI 编码 agent 时常踩同一个坑，只跑 Terminal-Bench、DeepSWE 这类端到端基准。文章给出评估、迭代与防护 agent 的工程方法。
@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。他称当下 AI 能力已相当可观、创新节奏极快，人类正逼近 AGI，新机构正是在这一转折节点上设立。
@GoogleAIStudio · 203.4K 粉丝 · 35.5K 阅 · 506 赞 · 51 转
New Gemini Audio models are available for developers to build more intelligent conversational experiences via the Gemini API and Google AI Studio. Today, we released new Gemini Live models in the
中文介绍 Google AI Studio 上线新的 Gemini 音频模型，开发者可通过 Gemini API 与 AI Studio 调用 Gemini Live 实时语音模型和 Transcribe，构建更智能的实时对话体验。
@GoogleAIStudio · 203.4K 粉丝 · 35.1K 阅 · 585 赞 · 47 转
Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking are our most advanced live dialogue models yet. Major upgrades in intelligence and parallel reasoning make it more intuitive to collaborate and
中文介绍 Google 发布 Gemini 3.8 Live 与 Gemini 3.8 Live Extended Thinking，称其为迄今最强的实时对话模型，在智能水平与并行推理上有明显升级，人机协作更自然。
@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 一篇观点短文：AI agent 的关键变化不是聊天消失，而是聊天不再是产品的全部。助手只负责解释任务，agent 系统则要真正把任务推进下去——调用工具、执行并交付结果。
A dash of cold water keeps the foomers away.
中文介绍 该期 AINews 汇总多条 AI 新闻，包括 Yegge 关停 Gas Town 项目、Databricks 的 Astra 成本上涨 60%，并称业界对 AI 的过热预期出现降温迹象。
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，提供前沿智能能力、律所定制工作流、可连接的法律数据源，以及针对机密客户工作的法律级管控措施。
中文介绍 TLDR AI 当日简报涵盖三项内容：Claude 与 Cowork 合并、ChatGPT 推出赞助式智能体，以及被称为「harness tax」的相关话题。
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，讨论为超智能「承保」、支持可被追责的 AI 智能体，并涉及其 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 发布一套用于追踪、调查和披露模型失准的框架，同时公布六份关于模型出现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，在美国 10 座城市为 1000 名老年人提供免费实操 ChatGPT 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 广告新体验，包括赞助智能体（Sponsored Agents）、面向营销人员的工具，以及与 HubSpot 和 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 文章指出 AI 热潮正演变为材料挑战：半导体与数据中心在性能上接近物理极限，支撑 AI 基础设施的材料与运行其上的算法同等关键。
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 ChatGPT Work 与 Codex 的分析功能，帮助团队了解 AI 使用情况与支出、识别培训需求，并将 AI 采用与业务成果关联。
congrats to TypeSafe!
中文介绍 AINews 介绍 Jev，一种只负责决策、分类、路由和评分的「系统一模型」，据称比小型前沿大模型快 100 倍以上、成本低 200 倍以上。
New OpenAI Economic Research shows how workers use AI beyond traditional roles and which new activities become recurring parts of their work.
中文介绍 OpenAI 经济研究新报告显示，员工在传统岗位职责之外使用 AI，其中部分新活动正逐渐成为其工作中的常态化组成部分。
中文介绍 TLDR AI 当日简报涵盖三项话题：Jev 模型、Periodic Neon 项目以及 Gemini 3.8 Live 的发布。
Good Start Labs trained an AI on a railroad game — and one version improved at financial research. The difference was the training design.
中文介绍 Good Start Labs 用一款铁路游戏训练 AI，其中一个版本在金融研究任务上有所提升，差异来自训练设计，以此探讨游戏技能能否迁移到现实工作。
Listen to the session or watch below Employees at the world’s leading AI labs are saying there’s a real possibility that advanced AI could destroy humanity. Are they right? Or is this more scaremongering and hype? Watch a conversation unpacking AI extinction fears: where they come from, whether they
中文介绍 MIT 科技评论圆桌讨论聚焦 AI 灭绝风险：全球顶尖 AI 实验室员工称先进 AI 确有可能毁灭人类，讨论围绕该说法是否成立，还是危言耸听与炒作。
中文介绍 DeepMind 发布两款新模型：Gemini 3.8 Live 与 Gemini 3.8 Live Extended Thinking。
美股方面，SPY 现价 762.6、单日 +1.13%，RSI14 为 50，价格略低于 SMA20（764.35）但高于 SMA50（759.53）与 SMA200（715.87）；QQQ 单日 +1.73%、RSI 53，维持多头排列。个股分化明显：AAPL RSI 65.4、META RSI 74 已超买，NVDA、MSFT 价格仍在均线上方，但 MSFT 的 MACD 7.6351 低于信号线 10.8055，红柱转弱。中概股整体最弱，BABA、PDD、0700.HK 均为空头排列，RSI 分别为 37.2、34.3、40.1，JD RSI 30.4 逼近超卖。加密方面，恐慌贪婪指数 50 属中性，总市值 2.63 万亿美元、24 小时 -1.73%，BTC 主导率 58.1%、ETH 11.3%；BTC RSI 51 且价格低于 SMA20，ETH、SOL 虽保持多头排列但 MACD 柱回落。商品与外汇中，WTI 原油 RSI 65.8、多头排列且价格远高于 SMA200（80.47）；黄金 RSI 46.8、MACD -3.2106 低于信号线 24.3913；美元/人民币 RSI 25.6 超卖并接近 52 周低；美元指数 100.24 与 10Y 美债收益率 4.95（RSI 65.3）均接近 52 周高。VIX 15.44 单日 -12.82%，空头排列。整体指标面呈股票偏强、中概与黄金偏弱的分化格局。
现价 337，单日 +1.38%、5 日 +3.19%，同时站上 SMA20（321.4）、SMA50（319.74）与 SMA200（286.08），呈多头排列，距 52 周高仅 -2.2%。RSI14 为 65.4，偏强但未进入超买区；MACD 5.2001 高于信号线 3.5139，红柱延续，动量结构完整，指标状态偏上行。
现价 682.31，5 日累计 +5.89%、单日 +1.34%，价格明显高于 SMA20（608.61）与 SMA50（606.69）及 SMA200（624.03）。RSI14 达 74，已进入超买区间；MACD 23.979 大幅高于信号线 14.2832，红柱扩张，短线动量强，但距 52 周高仍为 -13.72%，超买读数需留意。
现价 108.54，低于 SMA20（114.04）、SMA50（117.94）与 SMA200（133.54），形成空头排列。RSI14 为 37.2，位于中性偏弱区域；MACD -3.2744 低于信号线 -2.6445，绿柱维持。距 52 周高 -43.67%，5 日变动仅 -0.02% 属弱势横盘，均线与动量均未修复。
现价 26.64，单日 -0.97%、5 日 -1.41%，跌破 SMA20（28）、SMA50（29.61）与 SMA200（29.05）三条均线。RSI14 低至 30.4，逼近超卖阈值但未触发反转信号；MACD -0.8129 低于信号线 -0.7383，绿柱延续。距 52 周高 -27.73%，且信号列表为空，趋势读数偏下行。
现价 76400.38，单日 +0.33%、5 日 -1.13%，低于 SMA20（78013.74），但仍高于 SMA50（72149.11）与 SMA200（70336.81），并录得 SMA50 上穿 SMA200 金叉。RSI14 为 51，处中性；MACD 878.63 低于信号线 1622.27，红柱收缩。距 52 周高 -39.46%，方向读数不明朗。
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进行客观描述，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，请结合自身风险承受能力独立判断。
Follow the day’s news live Get our breaking news email, free app or daily news podcast Australia needs to slow down its artificial intelligence use and create better security laws, the companies creating the technology have told the federal government. In a submission to an ongoing inquiry, OpenAI s
中文摘要 澳大利亚联邦政府收到科技企业提交的意见，OpenAI 警告需放缓人工智能应用、完善安全立法以应对全球安全威胁；总理阿尔巴尼斯将在美国会见苹果公司首席执行官蒂姆·库克。
Clashes between Iranian-backed Houthi fighters and Saudi-backed government forces have been reported near oil-rich Marib Province.
中文摘要 也门战事升级，伊朗支持的胡塞武装向石油资源丰富的马里卜省推进，与沙特支持的也门政府军发生冲突。
Educator talks of ‘moral obligation’ and urges ‘every wealthy white celebrity’ to make similar donation YouTube star Ms Rachel has said she is matching rapper Macklemore’s $1m donation to organizations supporting Palestinians, calling on other public figures to follow suit. Macklemore, who was fired
中文摘要 YouTube 明星 Ms Rachel 表示将等额匹配说唱歌手 Macklemore 向支持巴勒斯坦组织捐赠的 100 万美元，并呼吁更多富裕白人名人效仿。
Russians vote for the first time since 2022. Can opposition parties challenge Putin-backed United Russia in this vote?
中文摘要 俄罗斯举行自 2022 年以来的首次选举投票，外界关注其他政党能否挑战普京支持的统一俄罗斯党。
Lawyers criticised mass arrests under emergency powers that allowed authorities to detain people without charge.
中文摘要 特立尼达和多巴哥宣布结束国家紧急状态。此前紧急状态赋予当局无需起诉即可拘留人员的权力，大规模逮捕引发律师批评。
The ruling comes amid legal battle over Trump’s changes to the Kennedy Center, including his effort to rename it.
中文摘要 美国一名法官裁定，对肯尼迪表演艺术中心进行任何实体改造前须提前 30 天通知。相关诉讼涉及特朗普对该中心的改动，包括更名计划。
The BBC understands Iranian President Masoud Pezeshkian and Foreign Minister Abbas Araghchi are among the delegates allowed to enter the US.
中文摘要 美国将允许伊朗代表团赴纽约出席联合国会议。据 BBC 了解，伊朗总统佩泽希齐扬和外长阿拉格齐均在获准入境人员之列。
The deal, which needs approval from Congress, comes as Riyadh seeks Washington's help in its war with Yemen's Houthis.
中文摘要 特朗普政府批准向沙特阿拉伯出售 F-35 战机，交易尚需国会批准。沙特正寻求美国在其与也门胡塞武装的战争中提供帮助。
A new generation of cross-cultural comics is taking regional humor global.
中文摘要 报道聚焦新一代跨文化喜剧演员，他们正将带有地域特色的幽默推向全球舞台。
Djibouti has appealed to the international community to help provide humanitarian assistance as thousand flee Yemen.
中文摘要 也门难民人数激增，吉布提呼吁国际社会提供人道援助；有也门人讲述前往吉布提的逃亡旅程，数千人逃离也门。
Iran's delegation to UNGA will be smaller this year, with travel restrictions and limits on luxury goods purchases.
中文摘要 美国批准伊朗高层领导人签证，允许其出席联合国大会。伊朗今年代表团规模缩小，并受旅行限制及奢侈品购买限制。
The CDC and officials from the US state differ over how four deaths should be classified as cases continue to spread.
中文摘要 美国宾夕法尼亚州就麻疹死亡病例的认定与疾控中心产生分歧，已请求 CDC 协助。双方对四例死亡是否应计为麻疹病例意见不一，疫情仍在扩散。
Move ends the Security Council’s watchdog panel on Iran sanctions.
中文摘要 俄罗斯和中国在联合国安理会投否决票，否决延长监督伊朗制裁的授权，安理会相关监察小组就此终止。
UN Interim Security Force for Abyei strengthens security on the ground after days of deadly fighting.
中文摘要 苏丹与南苏丹边境地区爆发冲突，至少 26 人死亡。联合国阿卜耶伊临时安全部队在连日激战后加强当地安全部署。
Why does Brazil get to speak first at UNGA?
中文摘要 该视频解释为何巴西在联合国大会一般性辩论中总是第一个发言。
Deal comes amid escalation in Middle East conflict — but transfer likely to face opposition in US Congress
中文摘要 特朗普政府批准向沙特阿拉伯出售F-35战机，交易金额约240亿美元。此举正值中东冲突升级之际，但该笔转让预计将在美国国会遭遇反对。
Asian stocks and bonds were set to rise, tracking a Wall Street rally as a pullback in oil prices eased inflation concerns and helped revive appetite for risk. The yen was steady ahead of the Bank of Japan’s rate decision.
中文摘要 油价回落缓解通胀担忧，带动华尔街股市反弹，亚洲股市与债券有望跟随上涨；日元在日本央行利率决议公布前保持稳定。
Oil dropped for a third day as supply concerns eased and traders looked to the next round of diplomacy that’ll shape the US-Iran war.
中文摘要 原油价格连续第三个交易日下跌，因供应担忧缓解，交易员正关注将影响美国与伊朗战争走向的下一轮外交谈判。
Developer launches system to track and report AI model misconduct
中文摘要 OpenAI披露其人工智能模型出现新的「令人担忧」行为，并推出用于追踪和报告模型不当行为的系统。
Kai-Fu Lee, CEO of 01.AI, says he suspects American AI companies may already have developed models that are "really great, but also really scary." He discusses this on "Bloomberg: The China Show" on the sidelines of the BNP Paribas Global Markets APAC Conference. (Source: Bloomberg)
中文摘要 零一万物（01.AI）首席执行官李开复表示，他怀疑美国人工智能公司可能已开发出「非常出色、也非常可怕」的模型。他在法国巴黎银行全球市场亚太会议期间作出上述表态。
Macerich's President and CEO, Jack Hsieh, joins the program shortly after presenting at the Bank of America real estate conference. The segment is framed around Macerich and the broader theme of "tracking the shopping mall resurgence." He speaks with Romaine Bostick on "The Close." (Source: Bloomber
中文摘要 美国购物中心运营商Macerich总裁兼首席执行官Jack Hsieh表示，消费者仍在支出，但选择更为挑剔；他在美国银行房地产会议后发言，并谈及购物中心回暖趋势。
Liz Ann Sonders, Charles Schwab chief investment strategist, says there is no one asset allocation that makes sense for every investor. She says the firm is neutral on equities, less favorable on fixed income, and more favorable on commodities. (Source: Bloomberg)
中文摘要 嘉信理财首席投资策略师Liz Ann Sonders表示，不存在适合所有投资者的资产配置。公司对股票持中性立场，对固定收益较不看好，更看好大宗商品。
Vincent Reinhart, Chief Economist over at BNY Investments, discusses the Bank of Japan’s upcoming rate decision due Friday. He speaks with Romaine Bostick on "The Close." (Source: Bloomberg)
中文摘要 纽约梅隆投资首席经济学家Vincent Reinhart讨论日本央行将于周五公布的利率决议，并谈及美联储的加息前景。
DoubleLine Capital chief executive Jeffrey Gundlach warned that the next US downturn could trigger a debt crisis that sends long-term Treasury yields sharply higher — defying decades of conventional wisdom that bonds will always serve as safe haven during times of economic strife.
中文摘要 DoubleLine Capital首席执行官Jeffrey Gundlach警告，美国下一轮经济衰退可能引发债务危机，推动长期美国国债收益率大幅上升，与债券在危机中充当避风港的传统认知相悖。
The State Department notified Congress that it plans to sell Saudi Arabia up to 48 F-35 warplanes in a first-time sale that could be worth as much as $24.3 billion. The sale of the Lockheed Martin Corp. fighter jet to Saudi Arabia has been in the works for a number of years. President Donald Trump s
中文摘要 美国国务院通知国会，计划向沙特阿拉伯出售最多48架F-35战机，交易价值最高达243亿美元；这是美国首次向沙特出售洛克希德·马丁公司生产的该型战机。
Jeni Britton, founder of Jeni’s Splendid Ice Creams and now Founder and CEO of Floura, discussed her shift from ice cream to fiber-based snacks made from produce trim like apple cores, watermelon rinds, mango and pineapple rinds. Britton highlighted Starbucks as an important validation point for the
中文摘要 Jeni's Splendid Ice Creams创始人Jeni Britton现创办并出任Floura首席执行官，利用苹果核、西瓜皮以及芒果、菠萝皮等果蔬边角料生产纤维零食，并提到星巴克的重要性。
For decades, the Japanese yen has been one of the world’s favored currencies for investors looking to borrow cheaply and put that money into assets offering higher returns.
中文摘要 报道解释套利交易机制：数十年来日元一直是投资者借入低成本资金、再投向高收益资产的首选货币之一，如今其吸引力正在下降。
3 回复 · 程序员 节点
5 回复 · Apple 节点
3 回复 · 程序员 节点
15 回复 · 程序员 节点
10 回复 · 程序员 节点
4 回复 · 程序员 节点
5 回复 · 程序员 节点
6 回复 · Apple 节点
15 回复 · Apple 节点
29 回复 · Apple 节点
该源今日无内容。