Python · ★ 110,864 · 🍴 16,802 · 📈 2,221 stars today
利用 AI 大模型和自动化工作流，根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.
中文介绍 基于 AI 大模型与自动化工作流，输入主题或关键词即可一键生成高清短视频。整合脚本生成、素材匹配、配音和剪辑等环节，大幅降低短视频制作门槛。适合内容创作者、营销人员和自媒体运营者，用于快速批量产出视频素材。
Python · ★ 30,258 · 🍴 2,338 · 📈 804 stars today
Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
中文介绍 面向 AI Agent 的自进化上下文数据库，统一管理 Agent 记忆、知识 RAG 与技能。通过上下文持续演化和整合，解决 Agent 在长期任务中记忆分散、知识检索和工具能力割裂的问题。适合构建复杂 Agent 应用的开发者，用于提高智能体的上下文利用效率。
TypeScript · ★ 2,721 · 🍴 322 · 📈 795 stars today
local multi-agent harness
中文介绍 本地多智能体运行框架，提供在一台机器上搭建和调度多个 AI Agent 的机制。适合开发者本地调试多 Agent 协作、编排和测试场景，无需依赖云服务即可快速迭代。项目定位轻量灵活，便于实验和折腾。
Python · ★ 29,872 · 🍴 3,541 · 📈 766 stars today
817 structured cybersecurity skills for AI agents · Mapped to 6 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF & MITRE F3 (Fight Fraud) · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 29 security domains ·
中文介绍 为 AI 智能体准备的 817 个结构化网络安全技能，覆盖 MITRE ATT&CK、NIST CSF 2.0、D3FEND 等六大框架，并遵循 agentskills.io 标准。可将安全知识与工具操作打包成 Agent 可调用的技能，适用于安全自动化、威胁检测分析和渗透测试辅助等场景。
Rust · ★ 26,460 · 🍴 3,422 · 📈 80 stars today
Production-grade Rust-native trading engine with deterministic event-driven architecture
中文介绍 生产级 Rust 原生交易引擎，采用确定性事件驱动架构。提供高性能、低延迟的回测和实盘交易能力，支持多种资产类别和交易所接入。适合量化交易团队和机构开发者，用于策略回测、实时交易系统搭建与高频交易场景。
Shell · ★ 223,902 · 🍴 19,257 · 📈 1,894 stars today
Skills for Real Engineers. Straight from my .agents directory.
中文介绍 面向真实工程师的 AI Agent 技能集，来自作者日常使用的 .agents 目录。包含一系列可复用的开发工作流技能，帮助 Claude Code 等 AI 编程工具更好地执行工程任务，如代码生成、重构和项目维护。适合前端或全栈工程师提升 AI 辅助开发效率。
Shell · ★ 274,311 · 🍴 24,558 · 📈 557 stars today
An agentic skills framework & software development methodology that works.
中文介绍 一套实用的 Agent 技能框架与软件开发方法论，将复杂任务拆解为可复用的 Agent 技能，并提供配套协作流程。目标是让 AI Agent 在真实工程中稳定工作，减少失控和返工。适合采用 Agent 编程的开发者团队，用于规范 AI 辅助开发的执行方式。
Python · ★ 19,854 · 🍴 1,693 · 📈 472 stars today
LLM inference server with continuous batching & SSD caching for Apple Silicon — managed from the macOS menu bar
中文介绍 面向 Apple Silicon 的 LLM 推理服务器，支持连续批处理和 SSD 缓存，并通过 macOS 菜单栏应用进行管理。能在本地高效运行大模型推理，减少内存压力、提升吞吐。适合 Mac 用户本地部署 LLM、做实验或搭建私有 AI 服务。
JavaScript · ★ 65,831 · 🍴 12,745 · 📈 198 stars today
Open-source AI job search: scan job portals, evaluate listings with a structured A-F rubric into a 1.0-5.0 score, tailor your CV, track applications — runs locally in your AI coding CLI (Claude Code, Codex, OpenCode, Antigravity…)
中文介绍 开源 AI 求职助手：自动扫描招聘网站，按结构化的 A-F 评分标准将职位评为 1.0-5.0 分，还能定制简历并跟踪投递进度。全部在本地 AI 编程 CLI（如 Claude Code）中运行，保护隐私。适合求职者用来筛选机会和优化申请材料。
TypeScript · ★ 111,896 · 🍴 6,632 · 📈 128 stars today
High performance self-hosted photo and video management solution.
中文介绍 高性能自托管照片与视频管理方案。提供类似 Google Photos 的备份、分类、检索和分享体验，数据完全由自己掌控。支持移动端和 Web 端，适合注重隐私的家庭用户、摄影爱好者或团队用于私有相册与媒体库管理。
Rust · ★ 4,578 · 🍴 85 · 📈 1,397 stars today
中文介绍 Amadeus Protocol 的 node 仓库。仓库描述为空，根据名称推断该项目是协议网络节点实现，用于运行和接入 Amadeus 网络。具体技术栈、功能与使用方式需查阅官方文档。适合想要搭建节点参与网络的开发者。
Python · ★ 13,163 · 🍴 659 · 📈 63 stars today
Draw pretty maps from OpenStreetMap data! Built with osmnx +matplotlib + shapely
中文介绍 用 Python 从 OpenStreetMap 数据绘制精美地图。基于 osmnx、matplotlib 和 shapely，可定制街道、建筑、水域等图层风格，生成极具设计感的地图插画。适合数据可视化爱好者、设计师和开发者，用于制作城市地图海报或创意地图作品。
TypeScript · ★ 16,227 · 🍴 810 · 📈 430 stars today
中文介绍 GenLayer 项目脚手架或样板工程，用于快速初始化基于 GenLayer 网络的去中心化应用或智能合约开发项目。仓库无详细描述，具体功能与目录结构需查看 README。适合准备在 GenLayer 上开发的 Web3 开发者。
👍 374
Long-horizon agents can fail even when their underlying models can solve the constituent steps. They may lose track of mutable state, fail to reactivate lessons from earlier executions, skip known procedures, or stop prematurely. We bet on harness scaling to improve the execution system around an ag
中文介绍 针对长程智能体在底层模型可解时仍因状态丢失、未复用经验、过早停止而失败的问题，StateM 采用 harness scaling 扩展智能体执行系统，而非微调模型。在 Terminal-Bench 2.1 上取得 95.3% 原始准确率，单次运行成本约 15 美元。
👍 142
Skills have emerged as a practical and effective approach for enhancing LLM agents at inference time through structured packages of knowledge. However, existing evaluations largely measure whether skills improve aggregated task success, leaving a more fundamental question underexplored: \textbf{When
中文介绍 系统研究 agent skills（结构化知识包）在推理时增强 LLM 智能体的生效条件，指出现有评测只看总体成功率，掩盖了技能何时有效、何时失效的问题。通过分解任务与技能的交互揭示失效的机制性原因，为技能的设计与使用提供指导。
👍 92
Reinforcement Learning (RL) has been promising in single-turn LLM fine-tuning. However, long-horizon agentic reasoning introduces increasingly branching interactions and sparse rewards, exposing several limitations of RL: its heavyweight backpropagation-based training stack makes it impractical to f
中文介绍 针对 RL 在长程智能体微调中因反向传播训练栈过重、难以处理分支交互和稀疏奖励的问题，提出 Agentic ESOpt，采用免反向传播的进化策略优化替代 RL 后端，在极低 GPU 资源下实现长程智能体的策略微调。
👍 60
Frontier open-weight models are increasingly available, but serving them still largely assumes datacenter infrastructure. We present FreeToken, an edge-native MoE serving system that treats a personal machine not as a small GPU, but as a unified, elastic inference platform. FreeToken co-designs the
中文介绍 将个人电脑视为统一弹性推理平台，提出边缘原生 MoE 服务系统 FreeToken，采用带宽自适应执行机制，协同设计调度、显存与通信，使开放权重 MoE 模型在边缘设备上高效运行，摆脱对数据中心基础设施的依赖。
👍 58
Scientific discovery often involves optimising expensive-to-evaluate objectives over vast, structured, and open-ended hypothesis spaces, such as molecules, protein sequences, and computer programs. Generative models such as large language models (LLMs) provide expressive priors over such spaces, but
中文介绍 面向分子、蛋白质序列、程序等开放假设空间中的高代价目标优化，提出 Large Discovery Models，以经验数据为基础进行模型驱动的开放式搜索，利用生成式先验引导探索，提升科学发现的样本效率与开放性。
👍 52
Artificial superintelligence (ASI) requires AI to move beyond mastering existing knowledge toward exploring the unknown, creating new knowledge, and turning new ideas into verifiable results. However, the capabilities of today's AI systems are still largely built on learning, compressing, and applyi
中文介绍 提出 ASI-Bench，面向人工智能超级智能能力的评测基准，衡量 AI 从掌握已知知识走向探索未知、创造新知识并将想法转化为可验证结果的能力，弥补现有模型仅靠学习、压缩和应用知识进行评测的局限。
👍 44
Although Large Vision-Language Models (VLMs) have significantly advanced embodied navigation, their direct deployment remains challenging, as existing methods often force VLMs into unnatural action spaces that misalign with their 2D pre-training priors, compounded by rigid reasoning schedules and in
中文介绍 针对 VLM 直接用于具身导航时被强制映射到与 2D 预训练先验不符的动作空间、且推理调度僵硬的问题，提出 Embodied-Navigator，通过指认、思考、记忆和对齐四个环节，使 VLM 在自然动作空间中完成高效导航。
👍 42
We present MOSS-VL, an open vision-language model family that treats real-time interaction -- perceiving while it speaks -- as a first-class capability. It is co-designed across the stack: the language decoder attends to vision only through gated cross-attention, so the model can naturally see incom
中文介绍 提出开源视觉语言模型家族 MOSS-VL，将边看边说的实时交互作为一等能力。语言解码器仅通过门控交叉注意力访问视觉信息，从架构到训练全栈协同设计，支持流式视觉输入与语音生成同步，实现低延迟多模态交互。
👍 39
Creative agents still lack an effective way to learn from high-quality human films, limiting their ability to produce cinematic-grade videos. A key challenge is the absence of a structured video representation that is both faithful to film content and directly usable for agentic reasoning and manipu
中文介绍 面向智能体学习高质量人类影片的需求，提出 AVA-Encoder，学习一种既忠实于影片内容、又可直接用于智能体推理与操作的视频表征，弥补现有视频表示在结构化程度和可操作性上的缺失，为电影级视频生成与创作智能体提供基础。
👍 26
AI has long assisted scientific research, but the rapid advance of LLMs and agentic scaffolds is reshaping the landscape; a single system can now carry whole-stage research from an initial hypothesis all the way to final published paper, which is a paradigm now referred to as AutoResearch. Existing
中文介绍 针对 AutoResearch（从假设到成稿的全流程自动科研）缺少细粒度诊断的问题，在 100 个真实前沿研究任务上进行端到端评测，系统分析智能体在科研各阶段的失败模式，为自动科研系统的能力边界和改进方向提供实证依据。
👍 19
Frontier large language models (LLMs) safety evaluation has largely treated harmful generation as an attack outcome rather than as an object of analysis. Consequently, little is known about the harmful outputs produced during model misbehavior, partly because large-scale, high-quality collections of
中文介绍 把有害生成本身当作分析对象，提出 HarmProfile 刻画前沿 LLM 在越狱或异常行为中产生的有害输出分布。通过大规模高质量有害样本与统计分析，揭示模型不当行为的内容结构与模式，为安全评估从攻击结果转向有害分布建模提供工具。
👍 16
Modern agents operate inside agent harnesses that manage tools, context, and control flow, making the harness a critical part of the agent system. Our original Agent Lightning introduced a disaggregated architecture that connects arbitrary agents to RL training through an LLM endpoint proxy, an appr
中文介绍 提出 Agent Lightning v1.0，在原有解耦架构上推进 harnessed agentic RL：通过 LLM endpoint proxy 将任意智能体接入 RL 训练，使工具、上下文和控制流组成的外部 harness 成为可训练系统的一部分，统一优化智能体与环境交互。
👍 15
As text-to-image generative models advance, they raise critical safety concerns, particularly the generation of Not-Safe-For-Work (NSFW) content such as violence and nudity, further exacerbated by red-teaming adversarial attacks. Existing defenses predominantly operate under white-box assumptions, r
中文介绍 针对文生图模型被红队对抗攻击生成 NSFW 内容的问题，提出 DiSCO，通过分布引导的对比提示优化进行防御，不依赖白盒假设，在保持生成质量的同时抑制暴力、色情等不安全内容的产生。
👍 13
Instruction-based general video editing seeks to unify diverse editing operations within a single, intuitive interface. Existing approaches often rely on resource-intensive conditioning, using either heavyweight branches or costly source concatenation. Is there any efficient way to model editing int
中文介绍 从二值证据视角出发，在生成式精修网络中建模编辑指令与源视频的关系，提出 GRNEdit，避免重型条件分支或高成本源拼接，实现高效、统一的通用视频编辑，兼顾编辑质量与计算开销。
👍 12
Vision encoders are a critical component of vision-language models, and scaling their capacity effectively improves performance. However, dense scaling increases compute cost and inference latency. Mixture-of-Experts (MoE) architectures offer a compelling alternative, having enabled efficient scalin
中文介绍 针对视觉编码器密集缩放带来的高计算量和推理延迟，提出 MoE 视觉编码器 MoE-ViE，用混合专家架构高效扩展视觉编码能力，在图像与视频理解任务中以较低成本获得接近稠密缩放的效果。
👍 11
Memory is becoming core infrastructure for long-horizon LLM agents, yet existing evaluations offer limited guidance on which memory substrate, namely the underlying medium in which memory is represented and stored, should be used under different operating regimes. We present a controlled harness eva
中文介绍 系统评估记忆智能体中不同 memory substrate（记忆表示与存储的底层介质）在多种运行机制下的适用性。通过受控 harness 评测，比较不同记忆介质对长程 LLM 智能体任务表现的影响，为选择记忆底层提供实证指导。
👍 10
Despite recent advances in unified multimodal models for multi-reference image generation, existing benchmarks remain organized around predefined task types (e.g., "subject composition"), which are ill-suited to this combinatorial setting and lead to fragmented coverage, uncontrolled complexity, and
中文介绍 针对多参考图像生成现有基准按预定义任务类型组织、难以覆盖组合空间的不足，提出 TRACE-Bench，将多参考生成能力分解为可组合的原子维度进行细粒度诊断评测，以受控复杂度评估统一多模态模型的真实能力与失败模式。
👍 9
Large-scale image generation has benefited from advances in data scale, quality, rebalancing, and recaptioning, yet conventional pipelines typically optimize task-specific datasets in isolation. A central challenge is not only how to curate each task-specific corpus, but also how to organize heterog
中文介绍 提出以能力为中心的数据设计范式，将异构任务语料组织为协同演化的能力集合，而非孤立优化各任务数据集，用于通用图像生成。在规模、质量、重平衡与重标注基础上，强调跨任务数据相互促进，提升生成器综合泛化能力。
👍 8
Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks. Yet existing benchmarks largely rely on researcher-selected tasks, leaving uncertain whether such progress extends to the work that real-world users actually dema
中文介绍 针对现有基准使用研究者自选任务、难以反映真实用户需求的问题，提出 StartupBench，基于市场验证的端到端工作流评测通用智能体，检验 LLM agent 在真实业务场景中的复杂任务执行能力。
👍 7
Large language models are increasingly deployed through agent harnesses that manage tools, extensions, persistent state, permissions, and external actions. Existing safety benchmarks mainly target individual attack mechanisms or a limited subset of operational settings, making it difficult to compar
中文介绍 面向 agent harness 全生命周期提出安全评测基准 HarnessRisk，覆盖工具、扩展、持久状态、权限与外部动作等环节，系统评估各类安全风险，弥补现有基准只关注单一攻击机制或局部设置的不足，支持不同 harness 的安全比较。
@nateherk · 25.8K 粉丝 · 1.3M 阅 · 558 赞 · 75 转
9 of my favorite Grok Bot hacks. I have been getting my hands dirty with Grok Bot, and the default setup is only the starting point. Creating a few bots is easy. The useful part is teaching them your
中文介绍 Grok Bot实战一周总结，精选9个实用技巧。默认配置只是起点，创建机器人本身不难，关键在于把个人工作流和习惯教给机器人，越定制越有用。
@sairahul1 · 135.5K 粉丝 · 938.1K 阅 · 867 赞 · 115 转
You are running a one-person company. You find the leads. Write the outreach. Make the images. Answer the replies. Count the week. Everything waits for you. Nothing gets handed off. And by the time
中文介绍 一人公司如何用Grok Bot减负：从线索挖掘、外联触达、视觉素材到回复跟进，所有重复性工作都可交给机器人，把创始人从琐事中解放出来，聚焦核心业务。
@XFreeze · 260.4K 粉丝 · 163.1K 阅 · 589 赞 · 93 转
Grok Bot is one of the most viral AI product launches right now... it keeps going viral as more people discover how insanely useful it is Just think about what you’re getting: You can have multiple
中文介绍 Grok Bot是当下最火的AI产品之一，入门方法很简单：关键是可以同时配置多个机器人，各自负责不同任务，实用性远超预期，值得上手一试。
@milesdeutscher · 676.5K 粉丝 · 142.9K 阅 · 515 赞 · 65 转
Elon has officially done it. He's created the most powerful agentic software in the world. In this guide, I'm going to lay out everything you need to know about the new Grok Bot. I've been ruthlessly
中文介绍 作者称马斯克打造了全球最强智能体软件Grok Bot，并给出终极指南。基于大量实测，覆盖功能、用法与技巧，适合想全面了解这款产品的人。
@mattyp · 49.7K 粉丝 · 141.9K 阅 · 764 赞 · 67 转
Grok Bot has a single output: a message This might seem backwards. Power-user tools are built on complexity and breadth, right? Chat can’t possibly have my workflow. It’s a special workflow. I’m a
中文介绍 观点帖：Grok Bot唯一的输出就是一条消息，看似简陋，实则回归本质。作者认为聊天界面足以承载复杂工作流，不需要追求工具的功能堆砌。
@JensenHuang · 1.0M 粉丝 · 110.9K 阅 · 1.9K 赞 · 234 转
Land, power and shell: The next critical resource for AI factories. AI factories are the defining infrastructure of the AI era—where compute transforms energy and data into intelligence that powers
中文介绍 黄仁勋谈AI工厂：算力将能源和数据转化为智能，而土地、电力和厂房（shell）是AI工厂的关键资源，将成为AI时代下一个争夺焦点，重要性堪比算力本身。
@0xCodez · 32.7K 粉丝 · 92.3K 阅 · 508 赞 · 48 转
Every AI tool you have used so far waits for you. You open it, you ask, it answers, you close it. The work still happens in your hands - the model just talks you through it. Follow my Substack to get
中文介绍 全教程：10步教你用Grok Bot Agents实现生活自动化。对比传统AI工具只会被动回答，Grok Bot能主动执行任务，真正把工作从你手中接过去。
@PiCoreTeam · 4.3M 粉丝 · 74.1K 阅 · 850 赞 · 222 转
Starting August 24, 2026, Pi App Studio will update how app creation and editing are priced. Until now, Pi has heavily subsidized the cost of creating and editing apps in App Studio for all creators.
中文介绍 Pi App Studio发布定价调整公告：自2026年8月24日起，应用创建和编辑将采用新计费标准。过去Pi一直为创作者提供高额补贴，此次调整将影响后续使用成本。
@_avichawla · 73.4K 粉丝 · 54.7K 阅 · 525 赞 · 48 转
Everything you need to understand, set up, and get real work out of Grok Bot. It covers what a Bot actually is, how the shared computer works underneath, how it compares to Hermes Agent, and how to
中文介绍 Grok Bot大师课，系统讲解Bot到底是什么、底层共享计算机如何运作，并与Hermes Agent对比，教你如何正确配置并让Grok Bot真正产出成果。
@GoogleAI · 2.4M 粉丝 · 49.3K 阅 · 516 赞 · 53 转
Turn your spreadsheets into live, interactive mini-apps with Sheets canvas and Gemini 3.7 Flash — no coding required. How does it work? Let’s dive into it: Most AI UI tools produce prototypes that
中文介绍 Google AI推出Sheets canvas与Gemini 3.7 Flash组合，无需编程即可把电子表格变成实时交互的迷你应用，区别于多数只出静态原型的AI UI工具。
@queenpresh_x · 3.3K 粉丝 · 15.5K 阅 · 516 赞 · 113 转
Your real-world data can help train better AI, and @KGeN_IO gives you a reason to contribute it. AI is getting smarter, faster, and more capable every day. But behind every impressive model is
中文介绍 讨论AI训练数据从哪来：真实世界数据对AI进化至关重要。@KGeN_IO平台通过激励机制鼓励用户贡献个人数据，让普通人也能参与训练更好的AI。
@Oracle_Vault · 21.6K 粉丝 · 11.5K 阅 · 2.8K 赞 · 597 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 ORACLE披露：2026年自主AI智能体已成为预测市场最有效策略之一，Polymarket上超30%活动来自算法/AI钱包。其验证的AI机器人24小时回报高达200%。
@Oracle_AI_Vault · 23.6K 粉丝 · 7.0K 阅 · 3.0K 赞 · 511 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 ORACLE验证AI机器人在Polymarket活跃交易，24小时内回报最高200%。2026年自主AI智能体已成预测市场利器，超30%交易量来自算法与AI钱包。
OpenAI reaffirms Zero Data Retention for eligible API customers and previews Private Safety Processing for advanced AI safety without compromising data privacy.
中文介绍 OpenAI重申对符合条件的API客户提供零数据保留（ZDR），并预览“私有安全处理”功能，旨在不损害数据隐私的前提下推进高级AI安全。
中文介绍 Hugging Face发布LFM2.5的Q4_0检查点，该版本通过量化感知蒸馏得到，旨在降低模型部署的资源需求。
the Memory crunch continues - Moore’s Law reversed to 2007 levels
中文介绍 据Latent Space报道，内存价格过去12个月上涨500%，内存紧缩持续，相当于摩尔定律倒退回2007年水平。
Replit introduces Free Mode, powered by GPT-5.6 Luna, so anyone can turn ideas into working software without worrying about token costs.
中文介绍 Replit推出免费模式，由GPT-5.6 Luna模型驱动，用户无需担心使用费用即可将想法转化为可用软件，从而扩大软件创建工具的覆盖面。
中文介绍 TLDR AI简报涵盖三大动态：GLM-5.3接口发布、Cerebras推出新芯片，以及OpenAI网络安全相关工作放缓。
ChatGPT Ads is expanding to 31 European markets. Learn how advertisers can reach people as they explore, compare options, and make decisions.
中文介绍 OpenAI宣布ChatGPT Ads扩展至欧洲31个市场，广告主可在用户探索、比较选项和做出决策的过程中触达受众。
Glean CEO Arvind Jain explains why model routing helps control AI costs for organizations, and how human feedback loops at scale improve its routing systems.
中文介绍 Glean CEO Arvind Jain表示，前沿模型成本与开源权重流行推动模型路由需求；模型路由帮助组织控制AI成本，大规模人类反馈循环持续改进路由系统。
OpenAI launches an initiative to strengthen democratic oversight of AI in national security, supporting government institutions with tools, training, and expertise.
中文介绍 OpenAI启动新倡议，旨在加强国家安全领域AI的民主监督，为政府机构提供工具、培训和专业支持。
中文介绍 Hugging Face博客文章（来自IBM研究）探讨AI智能体实际所需的内存规模，分析智能体应用中的内存需求与优化问题。
OpenAI and CodeAI are partnering to help students build AI literacy, think critically about AI, and develop the skills to use and shape it responsibly.
中文介绍 OpenAI与CodeAI合作，为学生提供AI素养教育，培养批判性思考能力，以及负责任地使用和塑造AI所需的技能。
OpenAI is strengthening monitoring, alignment, and security for frontier AI models. See how new safeguards are guiding the pace of model development.
中文介绍 OpenAI宣布加强前沿AI模型的监控、对齐与安全防护，以新防护措施指导模型开发节奏，应对网络关键能力时代的挑战。
ChatGPT for Teens helps teens learn, think critically, and use AI with confidence, with stronger built-in protections, healthy-use features, and additional controls for parents.
中文介绍 OpenAI推出ChatGPT for Teens，面向青少年学习场景，内置更强保护、健康使用功能及家长控制选项，帮助青少年自信地学习与使用AI。
AI companies like Anthropic and OpenAI regularly publish reports on how people are using products like Claude and ChatGPT, but they only release the data they want us to see, AI researchers say. “There is no independent source to corroborate it,” says Anka Reuel, a computer science PhD candidate at
中文介绍 MIT Tech Review指出，Anthropic、OpenAI等公司定期发布AI使用报告，但研究人员认为这些数据经过筛选，缺乏独立来源佐证。计算机科学家Anka Reuel表示，“没有独立来源可以证实”。
The AI industry’s boldest promise right now is that AI will soon improve itself, with almost no need for human oversight. LLMs can already write code, generate synthetic data for training, and optimize the computer chips they run on. Forecasts of explosive AI progress predict that what researchers c
中文介绍 MIT Tech Review分析认为，AI的递归自我改进可能不会很快到来。尽管大语言模型已能编写代码、生成训练合成数据并优化芯片，但爆炸式AI进展的预测可能过于乐观。
Asana used OpenAI Codex to replace an outdated testing system in two weeks, completing work expected to take five years for about $12K.
中文介绍 Asana使用OpenAI Codex在两周内替换了过时的测试系统，完成了预计需五年完成的工程工作，成本约1.2万美元。
整体市场技术面分化：美股 SPY 现价 769.06 站上 20/50/200 日均线，多头排列，RSI 57.8 中性偏强；QQQ 微跌 0.2%，RSI 51.5，动量不明显。科技股中 AAPL 5 日 +4.82% 偏强，MSFT 出现 MACD 死叉，META 跌破各日均线且 RSI 37.1，技术面最弱。加密板块显著领涨，恐慌贪婪指数 62 为贪婪，加密总市值 2.47T USD、24 小时 +8.09%；ETH 单日 +18.27%、RSI 82.7 超买，BTC RSI 74.4 超买，动能强但有过热信号。中概股多数反弹，BABA 和 PDD 站上 20 日线但 MACD 死叉，京东 5 日 -7.15%，技术面中性偏弱。商品外汇方面，黄金 RSI 70.4 超买，WTI 原油多头排列；美元/人民币 RSI 20.4 超卖，DXY RSI 29.7 超卖。VIX 跌 6% 至 14.89，权益市场情绪稳定。整体呈风险偏好回升、部分资产过热的分化状态。
ETH 单日上涨 18.27%，5 日累计 20.53%，现价 2266.69 高于 SMA20(1905.23)、SMA50(1863.12)及 SMA200(2003.34)，MACD 柱为 39.79 且快线位于信号线上方，呈金叉状态；RSI 82.7 进入超买区。动能强劲但短期有过热迹象，52 周高点回撤仍达 54.24%。技术 setup 偏上行。
BTC 现价 69665.66，单日 +7.71%、5 日 +10.62%，站上 SMA200(68997.07)，MACD 金叉且柱值为 404.01，远高于信号线 7.53；RSI 74.4 超买。加密恐慌贪婪指数 62 为贪婪，总市值 2.47T、24 小时 +8.09%。但价格距 52 周高点 -44.8%，整体仍处修复性上行态势。
META 现价 546.03，低于 SMA20(581.76)、SMA50(594.19)与 SMA200(625.08)，呈空头排列；MACD -11.90 低于信号 -7.69，继续死叉，5 日下跌 5.67%，RSI 37.1 偏弱。距 52 周高点回撤 30.95%。各项指标显示中期动能仍向下，技术状态偏下行。
USDCNY 现价 6.72，低于 SMA20(6.75)、SMA50(6.77)与 SMA200(6.88)，空头排列；RSI 20.4 进入超卖区，同时接近 52 周低点，MACD -0.0104 低于信号 -0.009，柱为负。1 日 -0.32%、5 日 -0.33%。人民币相对美元偏强，美元兑人民币技术 setup 偏下行。
MSFT 现价 484.31，仍高于 SMA20(463.04)、SMA50(416.57)与 SMA200(431.67)，RSI 63.5 处于正常偏强区间；但 MACD 23.05 已低于信号线 24.76，形成死叉，且 5 日累计下跌 1.65%。多空信号并存，趋势标记为中性，故 stance 为中性。
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
本报告仅供技术指标解读参考，不构成任何投资建议；报告内容基于公开行情数据的客观指标读数，不对未来涨跌做预测。过去走势不代表未来表现，市场存在不确定性，读者应独立判断。
It comes after a 60-day ceasefire expired on Monday, with no sign of a diplomatic or military off-ramp to the conflict.
The couple will remain non-working royals. It is unclear where they will live.
Greens senator David Shoebridge argues blame for debt rests with major parties for selling off Australian assets. Follow today’s news live Get our breaking news email, free app or daily news podcast ‘Outrageous’: Greens leader says sending Israeli ambassador home should be on the table Greens leader
Recently released body camera footage shows a deck collapsing during a warrant operation in Huntsville, Missouri.
Victorian court of appeal also to consider whether triple murder and attempted murder convictions should be quashed Erin Patterson argues mushroom murder convictions should be quashed due to ‘catastrophic’ failure of justice Get our breaking news email, free app or daily news podcast McWilliams says
A landslide triggered by the quake had blocked access to the village where Paulina Pobi had lived, leaving her buried under the collapsed remains An 84-year-old woman has been discovered alive in the flattened remains of her home three days after a powerful earthquake hit eastern Indonesia, authorit
The UK's foreign secretary condemned the move as "an unacceptable and destructive act".
Australia's foreign minister says she will summon the Israeli ambassador over 'insulting' Zomi Frankcom decision.
Retirement of longtime Stars and Stripes publisher comes amid the Trump administration's push for editorial control.
Nearly 270,000 Israelis have left the country over three years, citing political uncertainty and security concerns.
After a deadline extension, the latest U.S. tariff threats seemed headed toward resolution on Wednesday, but questions lingered about what Canada would give up, and what it would get in exchange.
Prime Minister Carney hailed "significant progress" in trade talks as the two countries seek to hammer out final details ahead of a new weekend tariff deadline.
The collapse was caused by a landslide, while local officials say the death toll is likely to rise as recovery efforts continue More than 100 people have died after an artisanal goldmine at a small village in western Central African Republic (CAR) caved in due to a landslide, local officials and aid
English footballer Harry Kane won the title of ‘Europe’s best goalscorer’, awarded the Golden Shoe for the second time.
Rising costs fuel anger, but President Bola Tinubu’s machinery and opposition divisions may shape the race.
It comes after a 60-day ceasefire expired on Monday, with no sign of a diplomatic or military off-ramp to the conflict.
中文摘要 特朗普威胁称，任何帮助伊朗的国家将面临「巨大的经济后果」。此前为期60天的停火协议周一到期，冲突仍无外交或军事解决迹象。
Ebola is spreading rapidly beyond the main epicenter of Democratic Republic of Congo’s outbreak, overwhelming treatment capacity in some areas and putting mounting pressure on efforts to trace and contain infections.
中文摘要 刚果（金）埃博拉疫情已从主要疫区伊图里省扩散至北基伍省，病例持续增加。部分地区治疗能力不堪重负，对追踪和遏制感染构成压力。
Australia’s unemployment rate unexpectedly rose in July, supporting the Reserve Bank’s view that the labor market is likely to gradually loosen and prompting traders to pare bets on another interest-rate increase.
中文摘要 澳大利亚7月失业率意外升至4.5%，就业岗位减少。数据支持澳大利亚央行关于劳动力市场逐步放松的看法，交易员削减对再次加息的押注。
India’s market regulator banned a Mauritius-based unit of JPMorgan Chase & Co. from the capital market over alleged manipulative trading in the South Asian nation’s newly introduced auction-based system for share price discovery.
中文摘要 印度市场监管机构禁止摩根大通旗下毛里求斯部门进入资本市场，因其被指在该国新推出的拍卖式股价发现机制中从事操纵性交易。
China’s central bank used the daily reference rate to signal a desire for slower currency gains, after the yuan climbed to its strongest against the dollar in over three years.
中文摘要 中国央行通过设定弱于预期的每日中间价，释放放缓人民币升值步伐的信号。此前人民币兑美元汇率升至三年多来最强水平。
It comes after the interest rate on 30-year bonds reached the highest level in almost 20 years.
中文摘要 美国联邦政府国债总额突破40万亿美元，在过去十年间翻了一番。此前30年期国债收益率已达到近20年来的最高水平。
A group of Democratic lawmakers are urging Federal Reserve Chairman Kevin Warsh to disclose any conversations he has had with President Donald Trump since taking over the central bank in May.
中文摘要 一群民主党议员敦促美联储主席凯文·沃什公开披露其与唐纳德·特朗普总统之间的所有通话记录，这些通话发生在沃什5月接任美联储之后。
JPMorgan Chase & Co. strategists warned that markets may view the US Treasury’s surprise effort to curb long-term borrowing costs as lacking credibility, potentially pushing up the term premium and yields over time.
中文摘要 摩根大通策略师警告，市场可能认为美国财政部为遏制长期借贷成本而采取的意外债券回购举措缺乏可信度，长期或推高期限溢价和国债收益率。
Canada’s preliminary trade deal with the US caught Mexican officials and business leaders off guard. Now, Mexico wants one of its own, as questions grow over whether months of painstaking engagement can deliver comparable tariff relief from President Donald Trump.
中文摘要 美国与加拿大的初步贸易协议令墨西哥官员和商界领袖措手不及。墨西哥如今也寻求达成类似协议，但外界质疑数月的艰苦谈判能否从特朗普政府获得同等关税减免。
Global funds see any rise in Treasury 10-year yields toward 5% spurring outflows from emerging Asian bonds, as higher rates in the US will prompt investors to favor dollar assets over regional debt.
中文摘要 全球基金认为，美国10年期国债收益率若升至5%附近，将引发亚洲新兴市场债券资金外流，因美国利率走高促使投资者更青睐美元资产而非地区债券。
Australia’s prudential regulator will monitor lending practices across the country’s banks in an effort to ensure slowing credit growth does not lead to riskier loans.
中文摘要 澳大利亚审慎监管机构将审查全国银行的贷款行为，以确保信贷增长放缓不会促使银行转向风险更高的贷款，从而维护金融体系稳定。
CK Hutchison Holdings Ltd. commenced international arbitration proceedings against Panama, seeking more than $1.5 billion in damages over the loss of its investments in two ports on the country’s strategic canal.
中文摘要 长江和记实业有限公司已正式对巴拿马提起国际仲裁，称因其在巴拿马运河两个港口的投资损失，要求赔偿金额超过15亿美元。
8 回复 · 程序员 节点
31 回复 · Apple 节点
12 回复 · 程序员 节点
13 回复 · 程序员 节点
29 回复 · Apple 节点
11 回复 · Apple 节点
15 回复 · Linux 节点
28 回复 · Apple 节点
7 回复 · Apple 节点
12 回复 · Apple 节点
该源今日无内容。