JavaScript · ★ 29,910 · 🍴 6,021 · 📈 2,265 stars today
A spy satellite simulator in your browser, except the data is real. Live open source spatial intelligence on a photorealistic 3D globe.
中文介绍 浏览器中的间谍卫星模拟器，数据全部来自真实开源情报。它在逼真的 3D 地球上叠加实时空间情报，适合关注地缘政治、开源情报（OSINT）和地理空间分析的用户，用作可视化侦察工具。
TypeScript · ★ 1,807 · 🍴 552 · 📈 504 stars today
Open-source AI sales OS — self-hosted CRM with native AI agents + WhatsApp (WAHA). Open alternative to Kommo, Octadesk & Intercom for any business that sells by chat. MCP-ready, multi-tenant, LGPD.
中文介绍 开源自托管的 AI 销售操作系统，内置原生 AI 智能体并集成 WhatsApp（基于 WAHA），可替代 Kommo、Octadesk、Intercom。支持 MCP、多租户与 LGPD 合规，面向所有通过聊天完成销售的企业，用作对话式 CRM。
JavaScript · ★ 65,406 · 🍴 10,738 · 📈 217 stars today
Extracted system prompts from Anthropic - Claude Fable 5.1, Opus 5, Claude Design, Claude Code. OpenAI - ChatGPT GPT-6-Astra, Codex. Google - Gemini 3.8 Flash, 3.1 Pro, Antigravity. xAI - Grok, Grok Bot, Cursor, Kimi and more! Updated regularly.
中文介绍 收集并公开各大厂商大模型的系统提示词（system prompt）提取结果，涵盖 Anthropic 的 Claude、OpenAI 的 ChatGPT 与 Codex、Google 的 Gemini、xAI 的 Grok 等。适合提示词工程师与研究者分析模型行为、对齐策略和指令设计。
TypeScript · ★ 3,080 · 🍴 211 · 📈 209 stars today
User friendly sideloader
中文介绍 面向普通用户的越狱侧载工具（sideloader），用于把第三方应用安装到 iOS 设备上。相比命令行工具更易上手，适合不熟悉技术细节、希望在 iPhone 或 iPad 上安装非商店应用的用户。
Batchfile · ★ 33,209 · 🍴 2,540 · 📈 65 stars today
中文介绍 基于 zapret 的流量处理方案，主要用于绕过网络限制、恢复 Discord 和 YouTube 的访问。通过 DPI 规避技术处理数据包，适合网络环境受限、需要稳定访问这两个服务的用户。项目说明较少，具体用法参见仓库文档。
Python · ★ 5,128 · 🍴 401 · 📈 262 stars today
🤖📐专为数学建模设计的 Agent & skills ,自动完成数学建模，生成一份完整的可以直接提交的论文。 An Agent Designed for Mathematical Modeling ,Automatically complete mathmodel and generate a complete paper ready for submission.
中文介绍 专为数学建模竞赛设计的 AI Agent 与技能集，可自动完成建模流程并生成一份完整、可直接提交的论文。它把问题分析、模型选择、求解与写作串成流水线，适合参加数模比赛的学生和需要快速产出建模报告的研究者。
C# · ★ 15,922 · 🍴 1,953 · 📈 227 stars today
Smart PVR for newsgroup and bittorrent users.
中文介绍 面向新闻组和 BitTorrent 用户的智能 PVR 工具，用于自动追剧。它监控剧集更新、抓取资源、重命名整理并通知下载器，是自建媒体库用户搭配下载工具使用的经典自动化组件。
TypeScript · ★ 2,501 · 🍴 310 · 📈 376 stars today
Open Source AI trading agent that operates autonomously across 1000+ markets - Polymarket, Kalshi, Binance, Hyperliquid, Solana DEXs, 5 EVM chains. Scans for edge, executes instantly, manages risk while you sleep. Agent commerce protocol for machine-to-machine payments. Self-hosted. Built on Claude.
中文介绍 开源 AI 交易智能体，可在 1000 多个市场上自主运行，覆盖 Polymarket、Kalshi、Binance、Hyperliquid、Solana DEX 及 5 条 EVM 链。它能扫描套利机会、即时执行交易并管理风险，适合量化交易与预测市场玩家。
Java · ★ 33,213 · 🍴 2,014 · 📈 136 stars today
Browse media content with your own rules on Android TV
中文介绍 Android TV 上的第三方媒体浏览客户端，让用户按自定义规则浏览内容。作为 YouTube 非官方前端，它提供更灵活的界面与播放控制，适合在电视盒子和智能电视上追求自由观影体验的用户。
Python · ★ 137,629 · 🍴 20,234 · 📈 230 stars today
100+ AI Agents, Agent Skills and RAG Apps - Free and Open Source.
中文介绍 收录 100 多个免费开源的 AI 智能体、Agent 技能与 RAG 应用示例的合集。内容覆盖多种 LLM 使用范式，适合开发者学习大模型应用搭建、快速寻找可复用模板并作为项目起步参考。
Go · ★ 1,402 · 🍴 104 · 📈 355 stars today
Network stack research tool. TCP tunnel with pluggable transports.
中文介绍 网络栈研究工具，提供支持可插拔传输层（pluggable transports）的 TCP 隧道。可用于研究流量伪装、协议混淆与网络穿透，适合网络安全研究者、翻墙与抗审查技术开发者进行实验。
C · ★ 4,912 · 🍴 547 · 📈 237 stars today
Graphics Creation Tools
中文介绍 Armory 引擎配套的图形创作工具，专注于 3D 纹理与材质绘制。它支持在模型表面直接绘制 PBR 贴图，面向游戏美术、3D 设计师和独立开发者，用作美术资产制作流程中的纹理绘制环节。
Python · ★ 3,596 · 🍴 552 · 📈 113 stars today
claude-red is a curated library of offensive security skills designed for the Claude skills system. Each skill is a structured SKILL.md file that primes Claude with expert-level methodology for a specific attack surface — from SQLi to shellcode, EDR evasion to exploit development.
中文介绍 面向 Claude 技能系统构建的攻防安全技能库，每个技能是一个结构化的 SKILL.md 文件，为 Claude 注入特定领域的专家级方法论。适合渗透测试与红队人员借助大模型辅助安全评估工作。
Python · ★ 7,288 · 🍴 824 · 📈 210 stars today
YuE2: frontier music generation with symbolic planning, zero-shot covers, and agentic music editing.
中文介绍 YuE2 音乐生成模型，支持符号规划、零样本翻唱（zero-shot covers）与智能体式音乐编辑。它在前沿音乐生成基础上加入结构化编排能力，适合音乐创作者和研究者生成、改编与编辑歌曲片段。
Rust · ★ 7,236 · 🍴 256 · 📈 54 stars today
Worktrunk is a CLI for Git worktree management, designed for parallel AI agent workflows
中文介绍 面向并行 AI 智能体工作流的 Git worktree 管理 CLI。它简化多工作树的创建、切换与清理，让多个 Agent 同时在不同分支上独立工作而不互相干扰，适合用 AI 编码代理的开发者。
Go · ★ 23,452 · 🍴 3,059 · 📈 189 stars today
Fully autonomous AI Agents system capable of performing complex penetration testing tasks
中文介绍 全自主 AI 智能体系统，能够执行复杂的渗透测试任务。它把侦察、漏洞发现与利用等环节交由 Agent 自动编排，适合安全团队和红队用于自动化安全评估与渗透测试。
👍 444
Automating empirical research is a long-standing direction of AI. Recent automatic research (AutoResearch) agents bring this goal within reach, as modern LLMs show the capability to independently implement solutions and learn from the execution outcomes. Behind these gains, post-training (especially
中文介绍 针对自动科研（AutoResearch）智能体在长程实验中的扩展难题，提出以世界模型支撑后训练的扩展方案，使智能体能够预测执行结果并从反馈中学习，从而减少对真实环境的高成本交互，提升自主实现与迭代研究的效率。
👍 231
We introduce NCP-ArchPreview, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP). Alongside NTP, the model learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and mo
中文介绍 提出潜空间语言模型 NCP-ArchPreview，在标准 next-token prediction 之外引入 Next Concept Prediction（NCP），让模型预测跨多个 token 的离散概念，从而学到显式的多粒度语义结构，推动自回归预训练超越 token 级建模。
👍 92
Large language model (LLM)-based multi-agent systems (MAS) achieve strong performance by employing specialized multiple agents, yet their performance depends on the prompt design of each agent. For MAS prompt optimization, textual gradient methods that guide prompt updates using natural-language fee
中文介绍 面向基于 LLM 的多智能体系统（MAS）的提示词设计问题，提出 AgentGrad：以干预手段引导的提示优化方法，用自然语言反馈作为文本梯度指导各智能体的提示更新，缓解人工调参困难，提升多智能体协同性能。
👍 91
Large Vision-Language Models (LVLMs) have achieved strong performance on diverse visual tasks, yet their ability to reconstruct and reason about the 3D structure of the scene depicted in 2D images -- referred to as spatial intelligence -- remains limited. Existing approaches attempt to address this
中文介绍 针对 LVLM 空间智能不足的问题，提出 SpatialBlock：基于合成积木堆叠问题构造训练与评测数据，要求模型在二维图像上重建并推理三维结构，从而增强视觉语言模型对场景空间关系的理解能力。
👍 53
Agent usage is shifting toward long-horizon tasks such as coding and scientific discovery, among which terminal tasks are especially important. We introduce T1, a Mixture-of-Experts model of 122B total trained with reinforcement learning, operating a real shell in a cloud sandbox for up to 300+ tool
中文介绍 提出 T1，一个 122B 总参数的 Mixture-of-Experts 终端智能体，通过强化学习训练，可在云端沙箱中操作真实 shell 并完成 300 步以上工具调用，面向编程与科学发现等长程任务，验证了 RL 在终端环境中的可扩展性。
👍 47
Large Language Model (LLM) agents are turning language into real-world effects, making safety necessary against both indirect prompt injections and direct harmful requests. System-level safety harnesses add an enforcement layer beyond model-level defenses, but existing harnesses are usually designed
中文介绍 针对 LLM 智能体面临间接提示注入与直接有害请求的安全威胁，提出 EvoSafeHarness，通过演化方式为不同模型和领域自动构建系统级安全防护层，在模型级防御之外增加可执行的约束层，提升智能体实际部署安全性。
👍 37
Recent advances in wearable sensing enable continuous monitoring of physiological and behavioral signals, yet existing benchmarks rarely evaluate whether AI systems can reason over a real user's longitudinal wearable record. We introduce WearableQA, a benchmark comprising 4,084 10-option multiple-ch
中文介绍 提出 WearableQA：基于真实用户长期可穿戴传感数据的健康推理基准，含 4,084 道十选一多选题，覆盖生理与行为信号，用于评估 AI 系统能否在个体纵向记录上进行时序推理与健康判断，填补现有基准的空白。
👍 30
Reducing audio-encoder depth lowers the inference cost of speech large language models, but removing complete blocks perturbs the embeddings consumed by the decoder and can cause deletion and premature end-of-sequence errors. We introduce X-AuT, a progressive framework that selects layer combination
中文介绍 为降低 Speech LLM 推理成本，提出 X-AuT：通过跨尺度蒸馏对音频编码器进行渐进式压缩，选择层组合而非直接删除整块，避免嵌入扰动导致的漏词与过早结束序列问题，在保持性能的同时减少编码器深度。
👍 26
Mainstream robotic policies often adopt a Markovian formulation, but many complex real-world manipulation tasks are inherently non-Markovian, requiring long-horizon memory beyond the current observation. Existing memory mechanisms often rely on language summaries, growing visual windows, or their co
中文介绍 针对复杂操作任务的非马尔可夫特性，提出以「记忆即计划」为核心的世界-动作建模框架：将长程记忆作为可规划表示而非语言摘要或增长视觉窗口，使机器人策略能够利用超出当前观测的历史信息进行长程决策。
👍 23
SWE-Bench Pro has emerged as a standard benchmark for evaluating software engineering agents on challenging repository-level tasks. However, our analysis work show that its evaluation is undermined by two sources of unreliability: reward hacking, enabled by leakage of gold solutions or hidden evalua
中文介绍 分析指出 SWE-Bench Pro 评测存在奖励作弊风险，源于 gold solution 泄露与隐藏评测信息等不可靠因素，为此提出 SWE-Bench Pro Verified，通过更严格的验证流程构建可信基准，用于公平评估软件工程智能体。
👍 20
Sequential memory agents process long documents by reading chunks one after another while maintaining a compact memory state, coupling document traversal to reasoning depth. This coupling introduces sensitivity to evidence placement and ties inference latency linearly to document length. We introduc
中文介绍 针对顺序记忆智能体阅读长文档时推理深度与文档长度耦合的问题，提出 PARSER：并行读取文本块、分层深度推理，解耦文档遍历与推理过程，降低对证据位置的敏感性，并使推理延迟不再随文档长度线性增长。
👍 19
We introduce MetroLLM-Bench, a 955-case benchmark for testing language models as the policy layer of a transit kiosk. It covers six real metro systems, ranging from 37 to 414 stations, and eleven categories that include routing, fare calculation, disruptions, accessibility, and adversarial input. In
中文介绍 提出 MetroLLM-Bench：含 955 个案例的基准，用于测试语言模型作为地铁服务终端策略层的能力，覆盖六条真实地铁系统（37 至 414 站）及路径规划、票价计算、故障、无障碍与对抗输入等十一类任务，考察实际可用性。
👍 19
While research on recursive self-improvement (RSI) has predominantly automated model training pipelines, reliable autonomous development demands a missing pillar: post-hoc monitoring and auditing to understand what models learn and ensure safe alignment. Mechanistic interpretability tools are essent
中文介绍 提出 SAEScientist-Bench，用于评估 AI 智能体能否自主开展 SAE 可解释性研究。该基准填补递归自我改进研究中事后监控与审计的空白，考察智能体在理解模型所学特征与安全对齐方面的自主研究能力。
👍 18
Temporal reasoning benchmarks for Video-LLMs are often mediated by language, leaving room for linguistic shortcuts from option wording, answer correlations, or language priors. To reduce such shortcuts, we introduce TempCloze, a video cloze benchmark for evaluating visual temporal reasoning in Video
中文介绍 针对 Video-LLM 时序推理评测易受语言线索干扰的问题，提出 TempCloze 视频完形填空基准，要求模型补全被遮挡的中段视频内容，减少选项措辞与语言先验带来的捷径，更纯粹地评估视觉时序推理能力。
👍 18
Code world models represent worlds as executable programs, but this representation alone does not determine how to construct a complex world. We introduce Recursive Code World Models (RCWM), a framework for reconstructing complex 3D worlds in code from a single reference image. RCWM couples a Recurs
中文介绍 提出递归代码世界模型 RCWM：以可执行程序表示世界，通过递归场景程序从单张参考图像重建复杂三维世界，将递归分解与代码生成结合，解决代码世界模型「如何构建复杂世界」的表示与构造问题。
👍 17
On-Policy Self-Distillation (OPSD) has emerged as a popular paradigm for large language model (LLM) self-improvement, allowing models to act as their own teachers by leveraging privileged information such as ground-truth solutions. However, recent findings indicate that OPSD can severely degrade the
中文介绍 针对 On-Policy Self-Distillation 在 LLM 自提升中可能严重损害性能的问题，提出 Negative Self-Distillation：让模型从自身错误中学习，通过避开缺陷而非模仿教师来改进推理，从而在自蒸馏范式中提升推理能力。
👍 17
Quantization has been widely adopted in LLM training and inference to reduce cost and improve efficiency. However, low-bit quantization of the attention module often introduces large errors at very low bit-widths, causing performance degradation. Existing methods mainly rely on smoothing techniques
中文介绍 针对 LLM 注意力模块在极低比特量化下误差大、性能下降的问题，提出 HyQuant 混合精度量化方法，突破传统平滑技术思路，对注意力不同部分采用差异化比特分配，在降低训练与推理成本的同时保持模型精度。
👍 17
AI research agents combine prior knowledge, public sources, and experimental feedback to produce useful results. The Discovery Certification Protocol (DCP) turns claims about these results into executable recovery and feedback tests. Gate 1 validates useful improvement on sealed evaluation. Gate 2 g
中文介绍 针对 AI 科研智能体成果难以验证的问题，提出发现认证协议 DCP，将发现主张转化为可执行的复现与反馈测试：第一关在密封评测集上验证有效改进，第二关检验反馈可复现性，从而审计智能体发现的真实性。
👍 16
Reasoning language models have made substantial advances on a variety of complex tasks, yet their capabilities remain overwhelmingly English-centric: models primarily reason in English regardless of the language they are prompted in. This is inaccessible for non-English-speaking users, risks losing
中文介绍 针对推理模型以英语为中心、非英语用户难以受益的问题，研究数据混合策略对多语言推理泛化的作用，通过构建跨语言训练数据配比，使模型能够在被提示语言中直接用该语言推理，提升非英语场景下的推理表现。
👍 15
Neurosymbolic systems rely on mathematical solvers to guarantee reasoning correctness, yet solvers are fundamentally blind to whether a formal translation maintains strict reference-equivalence to a designated formalization. We formalize this vulnerability as Verdict-Preserving-Unfaithfulness (VPU):
中文介绍 指出神经符号系统中求解器无法判断形式化翻译是否保持与原题严格等价，并将该漏洞形式化为 Verdict-Preserving-Unfaithfulness（VPU），提出生成式奖励模型来评估自动形式化质量，弥补仅依赖求解器判定的不足。
@IAmPascio · 124.8K 粉丝 · 812.5K 阅 · 881 赞 · 140 转
The superintelligence, the singularity, AGI, whatever you call it, the one we're so scared of is, ironically, also the one we most likely won't see coming. This essay is going to probably be a pretty
中文介绍 一篇观点长文：作者认为我们恐惧的超级智能或AGI，恰恰最可能以我们意料之外的形式到来，甚至「由一群孩子构成」。属于哲思型讨论，非技术教程，信息多为作者个人推演，实际可验证内容有限。
@mayukh_panja · 9.8K 粉丝 · 311.6K 阅 · 501 赞 · 75 转
There seems to be a widespread impression that OpenAI has somehow “solved the Navier–Stokes equations,” and that this represents some enormous breakthrough for fluid dynamics and engineering. It
中文介绍 驳斥「OpenAI 已解决纳维-斯托克斯方程」的流行误解，指出该说法在流体力学和工程学界被夸大传播，实际成果与真正解决该千禧年难题相去甚远。属于对 AI 科研新闻的纠错澄清。
@mbeisen · 73.0K 粉丝 · 113.7K 阅 · 528 赞 · 68 转
Over the last few months, the cancer-curing capabilities of LLMs have improved dramatically, to the point that they have cured previously untreatable types of cancer and are on the cusp of developing
中文介绍 讨论 LLM 在癌症生物学上的「严重错位」：作者称近几个月 LLM 的治癌能力大幅提升，已治愈此前无法治疗的癌症类型，并逼近自主开发新疗法。观点争议性强，把 AI 能力表述推到极高，需谨慎看待。
@EXM7777 · 142.6K 粉丝 · 94.6K 阅 · 501 赞 · 56 转
I'm basically giving you the full system i'd use to automate content creation for ANY business... every tool in the chain, every step in order, and where it breaks because a content agent is just a
中文介绍 一份「自动化内容创作」完整课程：作者给出可直接套用于任意业务的系统，逐条列出工具链、执行顺序，并指出内容 agent 在哪些环节会崩。适合想搭建内容流水线的从业者参考，强调实战而非概念。
@ForwardEditor · 11.7K 粉丝 · 57.6K 阅 · 511 赞 · 32 转
Hey, I’m Brian. Let’s get you started with vibe editing. 1. What I highly suggest having DaVinci Resolve Studio - $295 USD. It’s a one-time purchase, not a subscription. In my experience, Studio
中文介绍 「vibe editing」入门无废话指南。作者 Brian 建议使用 DaVinci Resolve Studio（295 美元一次性买断而非订阅），按步骤带新手进入 AI 辅助视频剪辑工作流。工具选型和上手路径讲得比较具体。
@OpenAIDevs · 421.3K 粉丝 · 51.6K 阅 · 677 赞 · 54 转
We shipped GPT-6 Astra. Builders took it from there: Explore 2,234 modeled anatomical parts in 3D: A Manhattan recreation built in Unreal Engine: Six Van Gogh paintings become a walkable
中文介绍 OpenAI 官方开发者账号展示社区用 GPT-6 Astra 做的作品合集：包括 2,234 个可 3D 探索的解剖部件、用 Unreal Engine 还原的曼哈顿场景、以及把六幅梵高画作变成可行走空间。属于产品发布后的用例展示。
@Av1dlive · 25.6K 粉丝 · 49.2K 阅 · 512 赞 · 53 转
i'm going to show you exactly how to master OpenAI's Images 2.5 to create app designs, websites, product shots.....14 tricks. 44 curated examples. the prompts, the results, and how to fix what goes
中文介绍 一份掌握 OpenAI 图像模型 Images 2.5 的完整教程：覆盖 App 设计稿、网站、产品图等场景，包含 14 个技巧、44 个精选范例，并演示出问题后如何修。偏向 prompt 与实操技巧合集。
@shannholmberg · 36.1K 粉丝 · 46.0K 阅 · 510 赞 · 44 转
marketing engineers build the systems behind the marketing. learn how to connect your knowledge, data and agents into workflows that research, plan and run campaigns. I started bringing more of my
中文介绍 介绍「营销工程师」这一角色的入门路径：核心是把知识、数据和 agent 连成可自动研究、规划并跑campaign的工作流。作者分享自己如何逐步把营销工作交给系统，偏职业转型与工作流搭建方向。
@AndrewYNg · 1.9M 粉丝 · 43.5K 阅 · 722 赞 · 105 转
When you’re skilled at AI Engineering, your best work won’t be merely implementing a product that someone else spec’ed out. Instead, you will actively shape the build. Before modern AI tools
中文介绍 吴恩达提出「AI 工程技能图谱」：当真正掌握 AI 工程时，你的价值不只是实现别人写好的产品规格，而是主动塑造 build 本身。属于对 AI 工程能力框架的定义性论述，非具体教程。
@DataRepublican · 957.2K 粉丝 · 37.3K 阅 · 846 赞 · 155 转
NYC and LA are restricting student AI. We used it to expose what school missed—and repair two years of gaps in six weeks. Part 1 of 3 in a series. Between my husband and I, three of our parents were
中文介绍 系列文章第 1 篇：在纽约、洛杉矶限制学生用 AI 的背景下，作者用 AI 帮儿子在六周内补完两年 Saxon 数学的缺口。记录具体方法与过程，属于家庭场景下 AI 教育实践的案例分享，共三部分。
@skeptrune · 17.6K 粉丝 · 16.6K 阅 · 507 赞 · 38 转
I made the switch from B2B SaaS web development to inference about a month ago and wanted to share my learnings for others that might be interested in doing a similar thing. After all, inference may
中文介绍 作者一个月前从 B2B SaaS Web 开发转向推理工程，分享转型学习路径，并整理出一份「推理工程技能图谱」，供有类似转岗意向的人参考。属于职业路径与技能清单类内容。
@Alaya_AI · 321.1K 粉丝 · 10.6K 阅 · 504 赞 · 450 转
As autonomous vehicles, robotics, and spatial computing platforms evolve, AI model training is shifting away from isolated 2D image annotation toward complex multimodal sensor fusion. Today’s frontier
中文介绍 Alaya AI 介绍其面向自动驾驶、机器人和空间计算的多模态传感器同步能力：随着模型训练从孤立 2D 图像标注转向复杂多模态传感器融合，其方案聚焦传感器对齐与前沿训练需求。偏产品技术宣传。
Perplexity uses Astra to write communications, change software, and monitor production systems, and checks in much less frequently than with earlier models.
中文介绍 Perplexity 采用 Astra 模型编写通讯内容、修改软件并监控生产系统，与使用早期模型相比，人工检查频率明显降低。
Before co-founding Kepler, Vinoo Ganesh led Spark at Palantir and built Project Frontline — a pioneering program for Forward Deployed Engineers. He takes us through the best practices of FDEs.
中文介绍 前 Palantir Spark 负责人、Kepler 联合创始人 Vinoo Ganesh 曾主导 Palantir 前线部署工程师项目 Frontline，他在访谈中分享该岗位的最佳实践。
We agree with Sebastian: this should have been DeepSeek v5
中文介绍 DeepSeek 发布 v4.1-Flash，采用 763B 参数、每 token 激活 8B、16B 的新颖因果编码器-解码器架构，并支持视觉能力，被形容为「鲸鱼回归」，作者认为其应被命名为 v5。
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? Join MIT Technology Review executive editor Niall Firth for a conversation with senior AI editor Will Douglas Heaven and AI
中文介绍 MIT Technology Review 举办圆桌讨论，由执行编辑 Niall Firth 与资深 AI 编辑 Will D 对谈，探讨顶尖 AI 实验室员工所提出的先进 AI 可能毁灭人类的说法是否成立，抑或只是危言耸听。
GPT‑6 Astra improves Devin’s ability to test software and show that it works, with the goal of helping engineers review less code and ship more.
中文介绍 Cognition 借助 GPT-6 Astra 提升 Devin 测试软件并验证其可运行的能力，目标是让工程师减少代码审查量、加快交付。
Learn how OpenAI evolved Habitat from a Python library into a globally distributed storage platform serving 1 billion ChatGPT users and 22M requests per second.
中文介绍 OpenAI 将 Habitat 从 Python 库发展为全球分布式存储平台，目前服务 10 亿 ChatGPT 用户，每秒处理 2200 万次请求。
中文介绍 TLDR AI 当日内容涉及 Agents API、Cognition SWE-2 以及 Muse 共享智能体等话题。
César de la Fuente’s lab uses Codex and ChatGPT to search living and extinct genomes for antimicrobial candidates to fight drug-resistant infections.
中文介绍 研究员 César de la Fuente 的实验室利用 Codex 与 ChatGPT 在现存及已灭绝生物的基因组中搜寻抗菌分子候选物，以对抗耐药性感染。
Meet the Data agent in ChatGPT Work. Connect company data, uncover insights, and build interactive dashboards with AI using natural language.
中文介绍 ChatGPT Work 推出数据智能体，可连接企业数据、发掘洞察，并通过自然语言让 AI 构建交互式仪表板。
On July 22, 2026, a transmission line fault in Ashburn, Virginia—the heart of the world’s largest data center cluster—knocked more than 3 gigawatts of load off the grid in seconds. And it wasn’t the first time. Two years earlier, a single failed surge arrester dropped roughly 60 Virginia facilities
中文介绍 2026 年 7 月 22 日，弗吉尼亚州阿什本——全球最大数据中心集群所在地——一条输电线路故障在数秒内令电网失去逾 3 吉瓦负荷；两年前也曾因单个避雷器故障损失约 60 兆瓦，凸显 AI 供电的架构问题。
Introducing ChatGPT for Financial Services, combining built-in financial data and GPT-6 Astra for research, modeling, and client-ready materials.
中文介绍 OpenAI 推出面向金融服务的 ChatGPT，结合内置金融数据与 GPT-6 Astra，用于研究、建模及生成可直接交付客户的资料。
OpenAI and GSA will offer eligible federal, state, local, and tribal governments $0 license fees, 50% off usage, and expanded cyber defense support.
中文介绍 OpenAI 与美国总务管理局合作，向符合条件的联邦、州、地方及部落政府提供 0 美元许可费、使用费五折优惠及扩展的网络安全防御支持。
a quiet day
中文介绍 当日 AI 领域较为平静，无重大事件。
GPT‑Live‑1 brings natural, full-duplex voice conversations to the API, with stronger instruction following, custom voices, and telephony support.
中文介绍 GPT-Live-1 在 API 中提供自然全双工语音对话，具备更强的指令遵循能力、自定义语音及电话支持。
Build and launch cloud agents with the Agents API, a managed service powered by the Codex harness for orchestration, long-running sessions, and tool use.
中文介绍 OpenAI 推出 Agents API，这是由 Codex 框架驱动的托管服务，用于构建和部署云端智能体，支持编排、长时间运行会话及工具调用。
美股方面，SPY 与 QQQ 分别收于 764.29 与 714.88，单日 +0.85%/+0.87% 但 5 日仍为 -1.15%/-0.39%，RSI14 均在 50.3/50.8 的中性区，MACD 读数低于信号线，指数层面呈高位震荡；META、AAPL、MSFT 维持多头排列，而 NVDA 出现 MACD 死叉、GOOGL 跌破 SMA20。加密方面，恐慌贪婪指数报 61（贪婪），总市值 2.66 万亿美元、24 小时 -2.61%，BTC 主导率 58.2%、ETH 11.6%；BTC 现价 77236.56 低于 SMA20 的 78586.04，RSI14 为 55，ETH 与 SOL 则保持多头排列但 MACD 读数低于信号线，整体为贪婪情绪下的高位整理。中概股整体偏弱，BABA、PDD、0700.HK 均呈空头排列且 RSI14 分别只有 36.6、30.3、39.7，JD 回到中性。商品与外汇分化明显：WTI 原油 RSI14 达 71.5 进入超买且多头排列，10Y 美债收益率 RSI14 为 73.7 同处超买并接近 52 周高，黄金中性，美元/人民币 RSI14 仅 14.7 深度超卖、空头排列，VIX 则单日 -11.21% 至 15.84。
现价 332.27 高于 SMA20 的 316.63 与 SMA50 的 317.97，多头排列成立；RSI14 为 62.8，动量偏强但未进入超买区；MACD 3.227 高于信号线 1.9023，红柱延续；单日 +1.75%，距 52 周高点仅 -3.57%。
现价 648.03 明显高于 SMA20 的 586.47，5 日涨幅 +6.12% 为样本中最高；MACD 14.0186 对信号线 4.1412，红柱显著扩张；RSI14 达 67.2，接近超买阈值，动量读数在美股大盘股中最强，但距 52 周高点仍有 -18.05%。
现价 100.05 高于 SMA20 的 88.52、SMA50 的 83.23 与 SMA200 的 79.59，多头排列完整，5 日涨幅 +9.58%；但 RSI14 已至 71.5 进入超买，单日回落 -2.37%，MACD 4.1926 仍高于信号线 2.7079，属超买状态下的强动量读数。
现价 77.81 低于 SMA20 的 84.32、SMA50 的 85.65 与 SMA200 的 97.3，呈空头排列；RSI14 仅 30.3，逼近超卖边缘；MACD -2.2718 低于信号线 -1.3506，绿柱持续；5 日跌 -4.68%，距 52 周高点 -44.19%，技术面为样本中最弱之一。
现价 6.7 低于 SMA20 的 6.72、SMA50 的 6.75 与 SMA200 的 6.85，空头排列；RSI14 仅 14.7 处深度超卖，为全样本最低；MACD -0.0118 已下穿信号线 -0.0112 形成死叉，且贴近 52 周低点，指标面延续弱势下行结构。
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
本报告仅为基于公开行情数据的技术指标客观解读，不构成任何投资建议。过去走势不代表未来表现，指标读数可能随价格快速变化，仅供技术指标解读参考。
Follow today’s news live Get our breaking news email, free app or daily news podcast Andrew Hastie says he “feels sorry” Barnaby Joyce and that he and One Nation MP David Farley probably have “buyers remorse” after joining the right wing party. Hastie has been feuding with Pauline Hanson this week o
中文摘要 澳大利亚自由党议员安德鲁·哈斯蒂表示，对加入一国党的同僚感到「买家懊悔」，称波琳·汉森「不是我曾以为的那个人」。工党政府则为其社媒算法「选择退出」方案辩护，能源部长克里斯·鲍文就中东局势影响燃油价格发声。
Houthis accuse Saudi Arabia of launching 129 strikes in 48 hours as fears grow over impact on Riyadh’s oil industry
中文摘要 沙特阿拉伯与也门胡塞武装在曼德海峡附近互相发动袭击。胡塞方面指控沙特在48小时内发动129次空袭，外界担忧冲突升级将冲击沙特石油产业。
A 15-year-old boy was miraculously rescued after spending two days clinging to his capsized boat in Alaska’s Bering Sea.
中文摘要 一名15岁男孩在阿拉斯加白令海翻船后，紧抓船体漂流两天后获救。
The Philippine Coast Guard says 41 more bodies have been recovered from the M/V June Aster, with 13 people still missing
中文摘要 菲律宾渡轮「M/V June Aster」火灾死亡人数升至76人。菲海岸警卫队称新发现41具遗体，仍有13人失踪。
Ben Shelton faces Alexander Zverev in the US final, hoping to end a 23-year drought in Grand Slams for American men.
中文摘要 美国网球公开赛男单决赛由本·谢尔顿对阵亚历山大·兹维列夫。谢尔顿力争为美国男单终结23年大满贯冠军荒。
The cool Kazakh lifts the trophy after frustrating Sabalenka in a 6-4, 5-7, 6-2 win in New York as the new world No 1.
中文摘要 哈萨克斯坦选手叶莲娜·雷巴金娜以6-4、5-7、6-2击败阿丽娜·萨巴伦卡，夺得2026年美国网球公开赛女单冠军。
Arsenal maintain perfect start to the season with their fourth win, thanks to goals from Guimaraes and Saka.
中文摘要 阿森纳主帅阿尔特塔对桑德兰一役的点球判罚表示不满，但球队仍凭借吉马良斯和萨卡的进球取胜，取得赛季开局四连胜。
The United States has been officially neutral on an issue that was a source of decades of violence in Northern Ireland until a 1998 peace deal.
中文摘要 特朗普表示爱尔兰统一将是「极好的」，打破美国在该问题上长期保持中立的惯例。自1998年和平协议以来，北爱尔兰问题一直是敏感议题。
Protesters in about 20 cities, including Berlin, Hamburg and Munich, demand the banning of the Alternative for Germany.
中文摘要 德国约20个城市、包括柏林、汉堡和慕尼黑爆发数万人游行，抗议极右翼政党德国选择党（AfD），示威者要求取缔该党。
Hunger and disease have tormented Yemen for years. Now, fighting between the Saudi-backed government and Iran-backed Houthis is pushing it to the brink.
中文摘要 也门民众因沙特支持的政府与伊朗支持的胡塞武装交战而再次逃离家园。多年饥荒与疾病之后，新一轮战斗使也门局势濒临崩溃。
The first naval battle between two Unmanned Surface Vessels (USVs) has taken place in the Black Sea.
中文摘要 黑海发生首次两艘无人水面艇（USV）之间的海战。
Even so, finding unity among the loose grouping of emerging economies will be difficult as the war in the Middle East strains their own interests.
中文摘要 随着油价上涨，金砖国家领导人峰会有伊朗参会，这是美国所没有的。但中东战争使各成员国利益分歧加大，这一松散的新兴经济体集团难以达成统一立场。
Nobody could have predicted that today's main global headline would be Trump's desire to see a unified Ireland.
中文摘要 特朗普关于爱尔兰统一的表态可能意在向大西洋彼岸的听众喊话。BBC评论称，当天全球头条是特朗普表达希望看到爱尔兰统一。
The call comes amid growing concerns that AI models may become able to inflict serious damage worldwide.
中文摘要 Anthropic首席执行官达里奥·阿莫代伊呼吁放缓人工智能开发步伐。此际外界日益担忧AI模型可能具备造成全球严重损害的能力。
Ebola has reached a seventh DR Congo province after an infected man travelled through Rwanda and Uganda.
中文摘要 埃博拉疫情已扩散至刚果民主共和国第七个省，一名感染者曾途经卢旺达和乌干达。刚果政府仍坚称病例数正在下降。
Consumer group Which? has called the sale of jack mackerel, a cheaper, less nutritious species of fish, an example of "skimpflation".
中文摘要 英国消费者组织Which？指出，价格更低、营养更差的「竹荚鱼」正替代鲭鱼登陆更多超市货架，并将其视为「缩水式通胀」的典型案例。
Warring tech bosses unite behind warnings that humans could lose control of the technology
中文摘要 OpenAI的奥尔特曼与xAI的马斯克罕见同调，声援Anthropic首席执行官达里奥·阿莫代伊提出的放缓AI研发呼吁，警告人类可能失去对人工智能技术的控制。
SpaceX is set to get a larger weighting in the Nasdaq 100 later this month, a change that could trigger billions of dollars of buying by passive funds tied to the benchmark.
中文摘要 SpaceX将在本月晚些时候的指数再平衡中获得更高的纳斯达克100权重，此举可能促使跟踪该基准的被动基金买入数十亿美元。
About-face comes a day after a filing revealed plans to sell up to 50mn shares by the end of October
中文摘要 甲骨文创始人拉里·埃里森取消75亿美元的公司股票出售计划，此前一天申报文件披露拟在10月底前最多出售5000万股。
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 Fmr. Ambassador to NATO Nick Burns, Council on Foreign Relations Ray Takeyh, Future of Life Institu
中文摘要 彭博周末节目由David Gura、Christina Ruffini和Lisa Mateo在纽约直播主持，梳理周末头条，嘉宾包括前驻北约大使Nick Burns及外交关系委员会Ray Takeyh。
Advances in autonomous agents and bitter rivalry between Anthropic and OpenAI have pushed once-fringe fears into the mainstream
中文摘要 自主智能体的进步以及Anthropic与OpenAI之间的激烈竞争，将一度边缘化的人类灭绝担忧推向主流讨论。
RBI rejects group’s appeal against forced listing, setting the stage for what could become India’s biggest IPO
中文摘要 印度央行驳回塔塔集团反对强制上市的申诉，要求塔塔之子（Tata Sons）推进上市，此举可能催生印度史上最大规模IPO。
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 Jeff Mason as they play and check out the quiz for yourself at Bloomberg.com (Source: Bloomberg)
中文摘要 彭博推出互动新闻测验节目Pointed，由Lisa Mateo、Christina Ruffini与Jeff Mason参与作答，不仅考察新闻知识，也考验答题者的信心。
Leaders of emerging powers including Russia and Iran projected a united front at a weekend summit in New Delhi, with a joint statement that targeted a US-launched war in the Middle East, Washington’s tariff campaign and its sanctions policy.
中文摘要 俄罗斯、伊朗等新兴国家领导人在新德里周末峰会上展现团结姿态，联合声明抨击美国在中东发动的战争、关税行动及制裁政策。
Joining David Gura on Bloomberg This Weekend, Stanford Spogli Institute for International Studies Senior Fellow Francis Fukuyama says the Iran war risks repeating lessons from Iraq and Afghanistan, arguing that military power alone is unlikely to produce lasting political change. In discussing his n
中文摘要 斯坦福大学学者弗朗西斯·福山表示，伊朗战争可能重演伊拉克和阿富汗的教训，认为仅靠军事力量难以带来持久政治变革。
PNC Chief Investment Officer Amanda Agati is on Bloomberg This Weekend and says the Fed faces a difficult decision as energy-driven inflation pushes markets toward expectations for higher rates, warning that further tightening could risk a policy error. Agati also tells hosts David Gura and Christin
中文摘要 PNC首席投资官Amanda Agati表示，能源驱动的通胀令美联储决策困难，市场预期利率上升，进一步紧缩可能带来政策失误风险。
US forces have redirected 100 commercial vessels over the past 60 days since the blockade against Iran resumed, the Central Command says in an update on X.
中文摘要 美国中央司令部称，自对伊朗封锁恢复以来的60天内，美军已累计改变100艘商业船只航向。
39 回复 · 程序员 节点
12 回复 · Apple 节点
7 回复 · Apple 节点
9 回复 · 程序员 节点
17 回复 · Apple 节点
7 回复 · Apple 节点
10 回复 · Apple 节点
11 回复 · 程序员 节点
6 回复 · 程序员 节点
47 回复 · 程序员 节点
该源今日无内容。