Top Project
microsoft/AI-For-Beginners
Jupyter Notebook · ★ 56,169 · 🍴 11,259 · 📈 869 stars today
12 Weeks, 24 Lessons, AI for All!
中文介绍 这个项目提供了为期12周的AI学习课程，包含24节课程，旨在为所有人普及人工智能知识。为什么重要：这表明了在技术教育中，AI知识普及的趋势和需求日益增长。
2026-08-01
Morning Brief
2026-08-01
今日，GPT-5.6价格大幅降价引发了商业模式与市场反应的深刻变革。
今日最重要的 5 件事
-
1
OpenAI大幅调价引发行业关注
为什么重要 GPT-5.6的降价将大幅降低AI工具的使用门槛，推动更多企业和个人用户接受人工智能技术。
-
2
TED Benna推出新退休计划
为什么重要 新退休计划为员工提供更灵活的资金使用方式，有助于解决当下金融压力，提高员工的生活质量。
-
3
基辅空袭致9人遇难
为什么重要 这一事件加剧了乌克兰地区的紧张局势，同时引发国际社会对人道主义问题的关注。
-
4
油价因中东局势回升
为什么重要 油价波动可能影响全球经济复苏和投资者情绪，需引起市场的注意。
-
5
西班牙首相呼吁欧盟会议应对移民危机
为什么重要
市场状态
Risk-Off
- 3 个跟踪资产处于多头趋势，7 个处于空头趋势。
- SPY 当前为 bullish，5 日 +1.10%，RSI 53.0。
- 加密恐惧贪婪指数为 27（恐慌）。
- 加密总市值 24h 变化为 +0.42%。
观察 观察 BTC 在 5 日 -1.08% 后能否企稳。
招聘信号
- 升温 AI产品管理 随着OpenAI推动GPT-5.6的商业化落地，市场对AI产品管理岗位的需求将持续上升。
- 升温 机器人技术与协作 Gemini Robotics的最新进展表明对机器人协作与应用的需求正在快速增加。
- 稳定 金融科技战略 新退休计划显示出金融科技在职场福利领域的应用潜力，但需求持平。
论文趋势
X 讨论主题
Most Practical
paperswithbacktest/awesome-systematic-trading
Python · ★ 12,056 · 🍴 1,505 · 📈 529 stars today
A curated list of awesome libraries, packages, strategies, books, blogs, tutorials for systematic trading.
中文介绍 这是一个整理好的系统化交易资源列表，包括库、策略、书籍和教程，提供给想要深入这一领域的交易者。为什么重要：它反映了系统化交易在金融市场中的逐渐普及，并支持交易策略的开发与优化。
Most Surprising
usekaneo/kaneo
TypeScript · ★ 5,467 · 🍴 472 · 📈 778 stars today
🎯 All you need. Nothing you don't. Open source project management that works for you, not against you.
中文介绍 该项目是一个开源项目管理工具，致力于优化用户的工作流程，提供所需的功能并剔除多余的元素。为什么重要：这种轻量级的工具强调了在项目管理中效率与灵活性的重要性，适应不同团队的需求。
查看更多 (12)
zhaoxuya520/reverse-skill
PowerShell · ★ 11,518 · 🍴 1,752 · 📈 1,360 stars today
Reverse Engineering / Authorized Penetration Testing / Security Research Skill Router Pack AI-powered routing + On-demand toolchain bootstrapping + Self-evolving knowledge base Supports Claude Code, Kiro, Cursor, Cline, and other AI coding clients 逆向/渗透/安全技能路由包 - AI 自动路由 + 按需自举工具链 + 自动进化经验库 | 支持 Cla
中文介绍 项目专注于逆向工程、授权渗透测试和安全研究，提供了AI驱动的技能路由和自我演化的知识库。为什么重要：它表明了安全领域使用AI技术辅助渗透测试和研究的趋势，提升了效率和效果。
microsoft/generative-ai-for-beginners
Jupyter Notebook · ★ 113,972 · 🍴 61,182 · 📈 104 stars today
21 Lessons, Get Started Building with Generative AI
中文介绍 这个项目提供了21节课程，旨在帮助初学者了解生成式AI的基础并开始构建应用。为什么重要：它反映了生成式AI教育重要性的上升，促进了新技术的普及与应用。
github/copilot-sdk
Java · ★ 10,218 · 🍴 1,381 · 📈 145 stars today
Multi-platform SDK for integrating GitHub Copilot Agent into apps and services
中文介绍 该项目提供一个多平台SDK，方便将GitHub Copilot Agent整合到应用和服务中。为什么重要：表明了智能辅助工具在软件开发中的广泛应用，提升了开发效率。
github/gh-stack
Go · ★ 723 · 🍴 32 · 📈 67 stars today
GitHub Stacked PRs
中文介绍 GitHub Stacked PRs允许用户在GitHub上管理和处理多个关联的Pull Request，提高开发流程的灵活性。为什么重要：这一功能提升了开源项目的维护效率，支持多任务协作。
huggingface/speech-to-speech
Python · ★ 10,041 · 🍴 1,218 · 📈 1,275 stars today
Build local voice agents with open-source models
中文介绍 该项目允许用户构建本地语音助手，使用开源模型实现语音转语音功能。为什么重要：它推动了语音交互技术的普及，为构建智能助手提供了便利。
abus-aikorea/voice-pro
Python · ★ 11,588 · 🍴 1,705 · 📈 53 stars today
Gradio WebUI for creators and developers, featuring key TTS (Edge-TTS, kokoro) and zero-shot Voice Cloning (E2 & F5-TTS, CosyVoice), with Whisper audio processing, YouTube download, Demucs vocal isolation, and multilingual translation.
中文介绍 提供了Gradio WebUI，帮助创作者和开发者使用TTS和零样本语音克隆技术，配合音频处理功能。为什么重要：这一工具简化了语音生成和处理工作流，提升了用户体验及创作效率。
iv-org/invidious
Crystal · ★ 21,492 · 🍴 2,425 · 📈 361 stars today
Invidious is an alternative front-end to YouTube
中文介绍 Invidious是YouTube的替代前端，提供无广告的视频观看体验。为什么重要：这一项目强调了用户对隐私保护与无干扰内容消费的需求，推动了视频平台的创新方向。
ansible/ansible
Python · ★ 70,034 · 🍴 24,259 · 📈 26 stars today
Ansible is a radically simple IT automation platform that makes your applications and systems easier to deploy and maintain. Automate everything from code deployment to network configuration to cloud management, in a language that approaches plain English, using SSH, with no agents to install on rem
中文介绍 Ansible是一个简单的IT自动化平台，可以简化应用和系统的部署与维护。为什么重要：它显示了基础设施自动化和DevOps流程优化的趋势，助力简化操作复杂性。
microsoft/TRELLIS.2
Python · ★ 9,812 · 🍴 1,191 · 📈 167 stars today
Native and Compact Structured Latents for 3D Generation
中文介绍 该项目致力于开发用于3D生成的紧凑结构潜变量，提升生成模型的表现。为什么重要：它推动了3D建模和生成技术的发展，适用于虚拟现实和游戏创建等领域。
TencentCloud/TencentDB-Agent-Memory
TypeScript · ★ 10,045 · 🍴 970 · 📈 250 stars today
TencentDB Agent Memory is a team-level memory hub for AI Agents — turning conversations, docs, and code into four reusable memory assets (Chat Memory, Skill, LLM-Wiki, Code-Graph) that are governed, shared, and equipped across agents and frameworks.
中文介绍 该项目是一个团队级的记忆中心，旨在将交互内容转化为可重复使用的记忆资产。为什么重要：强调了AI Agent在知识管理和交互历史保存中的辅助作用，推动了智能系统的提升。
NomaDamas/k-skill
JavaScript · ★ 6,673 · 🍴 790 · 📈 64 stars today
한국인을 위한 스킬 모음집 - 에이전트를 한국인으로
中文介绍 为韩国用户提供的技能集合，旨在帮助特定用户群体更好地利用AI代理技术。为什么重要：这一定制化的解决方案反映了针对不同文化和地区的AI应用需求正在增长。
bytedance/deer-flow
Python · ★ 78,583 · 🍴 10,731 · 📈 204 stars today
An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.
中文介绍 这是一个开源的超级代理框架，能够处理复杂的任务，并结合多个工具及技能。为什么重要：这一框架强调了在复杂任务中的智能化处理和自动化的趋势，提高了AI的应用范围与能力。
AskChem: Claim-Centered Infrastructure for Chemistry Literature Synthesis
👍 287
Chemistry literature synthesis often requires assembling specific findings scattered across many publications, yet existing literature-search systems primarily return ranked document lists. As a result, scientists and AI agents need to locate relevant information, verify their provenance, and assemb
中文介绍 针对化学文献的综合问题，论文提出了 AskChem 这样的基础设施，使得科研人员能够高效地聚合和验证散布在多篇文献中的特定发现。通过改进的信息搜索方式，助力科学家与 AI 代理更准确地定位相关信息，对其来源进行验证，并进行文献综合。这一方法在化学文献分析和知识发现等领域具有重要意义。
Qwen-UI-Agent Technical Report: Toward Next-Generation Real-World Centric Foundation GUI Agents
👍 277
GUI agents have the potential to become a general purpose executor over existing digital devices. To advance them toward real-world use, we envision agents that operate reliably on real devices, execute workflows across platforms, combine GUI interaction with CLI execution, complete long-horizon tas
中文介绍 论文探讨了下一代以现实世界为中心的 GUI 代理的构建，设计了一种可靠的基础 GUI 代理，能够跨平台执行工作流，并融合图形用户界面（GUI）与命令行接口（CLI）操作。通过这一创新，代理将具备更为广泛的应用潜力，对多种数字设备产生影响，推动人机交互领域的发展。
Metis: Memory Foundation Model
👍 252
Recent advances in AI agents have increasingly internalized native capabilities into their underlying foundation models, giving rise to multimodal foundation models and large reasoning models. However, agent memory is still primarily implemented through external modules, leaving the native memory ca
中文介绍 论文引入了 Metis，一个集成记忆功能的基础模型，通过提升 AI 代理内部的记忆能力，减少对外部模块的依赖。这种新型架构有望推动多模态和推理模型的进步，使得代理能具备更高效的学习和推理能力，影响未来智能系统中的记忆机制研究。
Frontis-MA1: Training an AI4AI Model towards Recursive Self-Improvement in Machine Learning Engineering
👍 161
Recursive self-improvement (RSI) requires AI systems that improve the process of building AI (i.e., AI4AI); machine learning engineering (MLE) offers a concrete, executable testbed for studying this capability. We introduce OpenMLE, an open full-stack system for RSI research in MLE, spanning verifia
中文介绍 论文介绍了 Frontis-MA1，一个用于研究机器学习工程中递归自我改进（RSI）的开放式完整系统。该系统涵盖了验证、优化和学习过程，极大地促进了 AI 系统在 AI 自我改进领域的普及和发展，为提升机器学习的自主优化能力奠定了基础。
PhiZero: A World Model Built Around Physical Language
👍 154
We introduce PhiZero, a physical world model built around physical language, a compact discrete representation of world-state transitions. Existing physical world models typically predict future videos directly in pixel space, leaving the underlying world dynamics implicit within high-dimensional vi
中文介绍 PhiZero 是一种围绕物理语言构建的世界模型，旨在通过紧凑的离散表示形式预测世界状态的转换。现有物理世界模型通常在像素空间直接预测未来视频，导致世界动态较为隐晦。通过引入物理语言，PhiZero 提供了一种新的视角和方法来理解和预测物理现象，对智能体行为建模和物理推理具有重要意义。
DistillAlign: Coordinating Mode Covering and Mode Seeking in Autoregressive Video Distillation
👍 88
Existing autoregressive video distillation methods commonly adopt a Distribution Matching Distillation (DMD)-based multi-stage pipeline. However, they typically decouple the initialization and DMD stages -- which then pursue different target distributions -- and judge the intermediate student mainly
中文介绍 DistillAlign 提出了协调模式覆盖与模式寻求的自回归视频蒸馏新方法，克服了现有方法在初始化和 DMD 阶段解耦的问题。研究表明，该方法在保留信息的同时改善了中间学生模型的表现，为视频处理和生成领域提供了新的思路与方法。
VideoCoCo: Code-as-CoT for Physically-Consistent Video Generation via an Agentic Dual-Engine System
👍 63
Text-to-video models have achieved remarkable visual quality, yet they still struggle to generate physically consistent dynamics because the temporal evolution of a scene must be inferred implicitly from a highly compressed text prompt. Existing chain-of-thought approaches introduce intermediate pla
中文介绍 VideoCoCo 通过代理双引擎系统提升了文本到视频生成的物理一致性，解决了模型在时间维度上难以推理的问题。此方法结合了链式思考与中间操作的理念，为生成高质量、动态一致的视频提供了新的探索方向，对多媒体生成技术领域有潜在影响。
Memory Decoder at Scale: A Pretrained, Parametric Long-Term Memory
👍 48
Decoder-only language models entangle long-term memory and reasoning in a single parameter set, making it difficult to scale memory capacity independently. Memory Decoder introduces a parametric long-term memory module but only studies it at a relatively small scale. In this work, we present Memory
中文介绍 Memory Decoder 提出了一个参数化的长时记忆模块，解决了解码器语言模型中长期记忆与推理交织的问题。通过引入该模块，研究实现了在相对更小规模下的深入研究，为记忆与推理的独立扩展提供了新路径，具有重要的工程应用价值。
Beacon: Knowing When and How to Perform Agentic Visual Reasoning
👍 46
The fundamental goal of agentic visual reasoning is to improve the success rate of multimodal large language models (MLLMs) on complex tasks, rather than merely equipping them with a sophisticated yet inefficient reasoning paradigm. In this work, we rethink agentic visual reasoning through two key d
中文介绍 论文重新思考了代理视觉推理的基本目标，旨在提高多模态大型语言模型（MLLMs）在复杂任务中的成功率。通过两个关键的策略，研究提出了一种更高效、更加有效的推理框架，推动了多模态学习和人机交互领域的发展，展示了其在实际应用中的重要性。
CLBench-V: Evaluating Multimodal Context Learning from Grounding to Knowledge Acquisition
👍 42
Real-world tasks often require models to learn from task-specific context rather than relying only on pre-trained knowledge. While recent work has highlighted this capability as context learning, existing evaluations mainly focus on textual contexts. In many practical settings, however, the context
中文介绍 CLBench-V 研究了多模态上下文学习的评估，强调在真实任务中，模型需要从任务特定的上下文中学习，而不仅仅依赖于预训练知识。该研究为未来模型在实际应用中的适应性与鲁棒性提供了新的评估视角，推动了跨模态学习的研究进展。
BM25 Wins at Scale: A Scaling Study of Retrieval-Augmented Generation Paradigms
👍 40
Retrieval-augmented generation (RAG) spans lexical and dense retrieval, graph-based indexing, and agentic search, but these paradigms are usually evaluated on different benchmarks at one corpus size, leaving their accuracy-cost scaling unclear. To bridge this gap, we present a controlled study that
中文介绍 BM25 Wins at Scale 通过控制实验研究了检索增强生成（RAG）范式的准确率与成本扩展性，填补了多种检索技术在不同基准上的评估差距。这一研究为 RAG 方法在大规模数据中的应用提供了实证基础，对信息检索与生成任务的效率提升具有指导意义。
ACE-Data-0: Human-Centric Ambient Capture as Embodied Data Engine
👍 34
Embodied intelligence faces a fundamental data bottleneck. Models must capture how first-person perception, whole-body motion, dexterous manipulation, object state, sound, and touch evolve together as humans pursue goals over time. Existing datasets fragment this experience across viewpoints, modali
中文介绍 ACE-Data-0 通过人本环境采集实现了对多模态数据的全面捕捉，解决了现有数据集在视角和模态上的碎片化问题。该方法在捕捉第一人称感知及整体运动状态等方面的进展，将推动身体智能体和数据模型的协同发展，具有重要的应用前景。
Beyond Borrowed Histories: Person-Aligned User Simulation for Interactive Role-Playing Evaluation
👍 30
Role-playing agents (RPAs) have become one of the most important consumer applications of large language models. Users engage in multi-turn conversations with RPAs for experiences such as emotional comfort, making reliable evaluation essential for measuring capability, comparing systems, and guiding
中文介绍 论文探讨了通过用户个性化角色模拟来评估交互式角色扮演代理的能力，强调在多轮对话中的可靠评估。研究为大语言模型在情感慰藉等消费者应用中的能力测量提供了重要标准，对代理的开发和测试具有现实意义。
RefCaptioner: Multi-Reference Image-Grounded Video Captioning
👍 25
Existing video captioning models generate natural descriptions of video content but cannot explicitly ground local visual elements to multiple reference images. We introduce multi-reference image-grounded video captioning, a new task requiring factual video descriptions with phrase-level reference g
中文介绍 RefCaptioner 提出了多引用图像支撑的视频字幕生成任务，强调在生成视频内容自然描述的同时，需显式地将视觉元素与多张参考图像进行对接。此方法为视频描述的真实性与准确性提供了新思路，推动了图像与视频理解的交叉研究。
See2Think: Do Multimodal Models Really Use Intermediate Visual States?
👍 22
Multimodal large language models increasingly use sketches, annotations, tools, and intermediate images during reasoning, but it remains unclear whether they truly rely on these visual states. Existing benchmarks are limited both by task collections with narrow coverage or partially text-solvable sa
中文介绍 See2Think 研究了多模态大型语言模型是否真正依赖于中间视觉状态，指出现有基准面临任务集覆盖狭窄的问题。通过深入分析，该研究为多模态推理模型在有效性和可靠性提升提供了新的视角，推动了未来的评估标准。
MindForge: Teaching Small Language Models Whole-Life-Cycle Software Engineering via Source-Free Program Synthesis
👍 22
Coding agents have made substantial progress on software engineering tasks that modify existing codebases, including bug fixing and feature implementation. However, constructing a complete program from scratch remains a major challenge: even the frontier models evaluated on ProgramBench fully resolv
中文介绍 MindForge 通过无源程序合成的方法培养小型语言模型的软件工程全生命周期能力，旨在解决从头构建程序的挑战。该研究为编码代理提供了创新训练方法，标志着在程序生成领域的进步，具有广泛的应用潜力。
SpatialCLI: Learning to Reason With Spatial Tools, Then Without Them
👍 21
Vision-language models (VLMs) are increasingly used in embodied agents to interpret visual inputs, reason about spatial relationships, and make task-level decisions based on that reasoning. However, a fundamental capability mismatch remains: general VLMs can reason about the overall task but often m
中文介绍 SpatialCLI 探讨了视觉语言模型在有空间工具条件下的推理能力，指出一般模型在整体任务推理中存在能力不匹配的问题。该研究为推动 embodied AI 领域中空间关系理解与决策支持提供了新思路，具有重要的应用意义。
SpecFirst: Behavioral Specification Elicitation as a First-Class Step in Agent-Based Program Synthesis from Scratch
👍 18
LLM-based agents excel at software engineering tasks where an existing codebase provides context, but constructing a program from scratch remains fundamentally harder. Recent benchmarks such as ProgramBench quantify this gap: given only natural-language documentation and an execute-only binary as a
中文介绍 SpecFirst 提出了将行为规范提取作为从头开始的程序合成中的首要步骤，旨在解决模式构建中的挑战。通过针对现有自然语言文档的研究，推动了基于 LLM 的代理在软件开发领域的能力，具有重要的实践应用前景。
β-OPSD: Deriving with Policy Optimization, Training with Self-Distillation
👍 17
On-policy self-distillation (OPSD) is a promising approach to improve reasoning language models, but it remains brittle in practice: making it work reliably often requires substantial engineering effort. We identify a structural source of this difficulty: vanilla OPSD is precisely the β=1 member of
中文介绍 论文探讨了自我蒸馏（OPSD）对推理语言模型的改进效果，指出现有方法的脆弱性，并提出了一种新方法（β-OPSD）来提高其稳定性。该研究对模型训练和推理技术的发展具有重要启示，促进了智能体在复杂任务中的应用。
ShadowDancer: Teaching Video World Models Any Action by Learning Unified Dynamics Representations from a Video and Its Shadow
👍 16
We present ShadowDancer, a novel approach to any-action, frame-level control of interactive video world models. The obstacle is representational: existing interfaces either encode an action loosely, leaving how it unfolds for the model to improvise, or encode it exactly through structured signals th
中文介绍 ShadowDancer 提出了一种新的方法，通过学习视频及其影像的统一动态表征，实现帧级别的任意动作控制。该方法为交互式视频世界模型提供了一种有效的控制方式，对未来智能体及其在图像处理中的应用有重要影响。
You already own the most valuable thing in shopping. You just can't reach it.
@fridayresearch_ · 1.9K 粉丝 · 1.6M 阅 · 597 赞 · 451 转
Your taste is scattered across forty companies that each own a piece and answer to advertisers. FRIDAY is building the version you own — a personal shopping intelligence that lives on your device and
中文介绍 博主指出，消费者的购物喜好分散在许多公司手中，而FRIDAY正在创建一种个人购物智能工具，让用户能够更好地掌控自己的购物体验。这一趋势表明，个性化购物助理在提升用户体验方面的重要性日益受到关注。
MiniMax H3: An Open Model Breaking the Boundaries Between Tasks and Modalities
@MiniMax_AI · 107.4K 粉丝 · 42.3K 阅 · 696 赞 · 103 转
Today, we're launching MiniMax H3, a general-purpose multimodal generation model. H3 understands unified context across text, images, video, and audio, generating video with native stereo sound, up to
中文介绍 MiniMax H3作为一款通用的多模态生成模型，具备理解文本、图像、视频和音频的统一上下文的能力，可以生成带有立体声的高质量视频。此类技术的发布标志着多模态AI技术的进一步发展。
AI News Roundup!
@AITECHio · 455.4K 粉丝 · 32.6K 阅 · 512 赞 · 110 转
ChatGPT Can Now Talk While Running Your Computer OpenAI has taken AI agents another step forward. You can now have a natural voice conversation with ChatGPT while it operates your computer in the
中文介绍 OpenAI新推出的功能使ChatGPT能够在远程操作计算机的同时与用户进行自然对话。这一进展显示出AI代理在个人计算和交互中的应用潜力正在迅速扩展。
Anthropic Are Buying Rare Books, Feeding Them Into AI, Then Destroying Them.
@ActionModelAI · 57.6K 粉丝 · 6.8K 阅 · 521 赞 · 390 转
It sounds like the plot of a dystopian film. But according to recently released court documents, internal company communications, and a reported $1.5 billion settlement, it's something that has
中文介绍 近期曝光的法律文件显示，Anthropic正在收购稀有书籍，将其信息输入AI系统后予以销毁，此举引发了对于知识产权和AI伦理的深刻讨论。行业信号表明，AI技术与文化资产之间的关系正引发广泛关注。
[AINews] not much happened today
apart from DeepSeek V4-Flash 0731, a quiet day.
中文介绍 除了DeepSeek V4-Flash 0731发布外，今日整体较为平静。
Ten advances in mathematics and theoretical computer science
OpenAI shares new results on long-standing open problems in mathematics and theoretical computer science, including advances in geometry, cryptography, and complexity.
中文介绍 OpenAI分享了数学和理论计算机科学中的十项重要进展，涵盖几何、密码学和复杂性等领域的长期未解问题。
Advancing responsible AI across Europe
OpenAI shares how its safety, security, transparency, and provenance practices support responsible AI governance in Europe. The work will continue as the EU AI Act advances.
中文介绍 OpenAI介绍其在欧洲支持负责任的人工智能治理的措施，包括安全、透明度和来源管理，随着欧盟人工智能法案的推进，该工作将继续进行。
Building abundant intelligence
A full-stack approach to making advanced AI more capable, more affordable, and more widely useful.
中文介绍 OpenAI采用全栈方法，使先进人工智能更具能力、成本更低且更广泛实用。
Univé builds an AI-ready workforce
See how Univé built an AI-ready workforce with ChatGPT Enterprise by combining leadership, responsible governance, and employee-led innovation to transform work at scale.
中文介绍 Univé通过结合领导力、负责任的治理和员工主导的创新，成功打造了能适应AI的工作团队，借助ChatGPT Enterprise实现了大规模转型。
not much happened today
**DeepSeek** launched the public-beta of **DeepSeek-V4-Flash API**, boasting a significant post-training performance leap without architecture or size changes, achieving a **Terminal-Bench score of 82.7** and nearing **GPT-5.6 Luna's 51** score at about **60% lower cost per task**. The model feature
中文介绍 DeepSeek发布DeepSeek-V4-Flash API公测版，称其经过训练后的表现显著提升，Terminal-Bench得分达82.7，接近GPT-5.6 Luna的51分，成本减少约60%。
[AINews] GPT 5.6 price cut by 20%-80%: Cost of GPT 5.4 Intelligence dropped 13x in 4 months due to GPT 5.6 recursive self-optimization
Distillation is all you need!
中文介绍 GPT 5.6的价格降低20%-80%，而GPT 5.4智力成本在4个月内减少13倍，归因于GPT 5.6的递归自优化。
Disrupting a Criminal Scam Operation
OpenAI disrupted a Cambodia-based scam operation using ChatGPT to support investment, romance, gambling, and impersonation schemes.
中文介绍 OpenAI利用ChatGPT干预了一项位于柬埔寨的诈骗行动，该行动涉及投资、恋爱、赌博及冒充等多种诈骗手段。
Inkling-Small 🧠, GPT-5.6 price cuts 💸, Gemini Robotics 2 🤖
GPU Management: Why Idle GPUs Are the New Grounded Aircraft
Gemini Robotics ER 2: powering robotics with video understanding, task orchestration, and multi-robot collaboration
Gemini Robotics ER 2 helps robots reason, collaborate, and solve real-world tasks. It represents a step change in video understanding, tool orchestration, and multi-robot collaboration for robotic applications.
中文介绍 Gemini Robotics ER 2助力机器人在视频理解、任务协调和多机器人协作领域取得重要突破，帮助机器人解决现实任务。
Ontologies Are So Back: Why AI Agents Are Reviving the Semantic Web
AI engineers are rediscovering ontologies as a way to keep probabilistic agents inside deterministic boundaries.
中文介绍 AI工程师重新发现本体论，以保持概率代理在确定性边界内运行。
A fundamental flaw leaves LLMs strikingly vulnerable to attack
It is impossible to make large language models fully secure against hacks because of a fundamental flaw in how they work, a team of researchers argue in a paper presented at the International Conference on Machine Learning, a top AI conference, this month. The claim has huge implications for the saf
中文介绍 研究团队在国际机器学习会议上指出，由于大型语言模型运作的基本缺陷，无法完全保障其安全，可能面临攻击风险。
Advancing the price-performance frontier with GPT-5.6
Explore lower GPT‑5.6 pricing for Luna and Terra—and how OpenAI’s more efficient models help enterprises deploy AI workflows at scale.
中文介绍 OpenAI发布了更低价的GPT-5.6 Luna和Terra，提升企业在大规模部署AI工作流程中的效率。
not much happened today
**OpenAI** aggressively cut prices for **GPT-5.6 Luna** by 80% and **Terra** by 20%, introducing a faster **Sol Fast** tier with up to 2.5× lower latency at double the price, improving agent workflow costs by roughly 10×. The **ARC-AGI-3** debate highlighted that the complete agent system, including
中文介绍 OpenAI大幅下调GPT-5.6 Luna价格80%，Terra下调20%，推出延迟更低的Sol Fast层，改进代理工作流成本约10倍。
当前市场环境面临波动，尤其在美股市场中，S&P 500 ETF (SPY) 1日上涨0.72%，显示出一定的抵抗力。加密货币领域，恐慌贪婪指数为27，反映出市场的恐慌情绪，主导的比特币(BTC)市值占比56.3%，已显示出一定的市场控盘。但整体加密市场在过去24小时的总市值仅微升0.42%，暗示缺乏强劲的买入信号。同时，美元指数DXY接近52周高位，显示出鹰派迹象可能影响其他市场的波动性。
今日关注
AAPL
Apple
偏下行
当前价格为308.91美元，RSI为43.2，低于50，显示出价格可能向下突破。同时，MACD已经形成死叉，表明下行动能增强。最近5天的变化为-7.24%，具备下行压力。
MSFT
Microsoft
偏下行
目前价格464.72美元，RSI为74.5，已超买。同时，5天涨幅为21.75%，但面临回调的风险。技术中MACD信号虽未形成金叉，但在超买区域，风险增加。
GOOGL
Alphabet
偏上行
当前价格为356.13美元，虽然RSI为54.3，但MACD已形成金叉，暗示动能转强。最近5天上升幅度达11.38%，显示出强劲的回升趋势。
全部资产
^VIX
VIX 恐慌指数
$15.99
-6.44%
- 5 日
- -13.94%
- 距 52w 高
- -54.7%
- RSI(14)
- 45.0
- 趋势
- 空头
- SMA 20 / 50 / 200
- 17.23 / 17.38 / 18.70
- MACD / 信号
- 0.150 / 0.150
空头排列
^TNX
10Y 美债收益率 (%)
$4.75
+1.76%
- 5 日
- +1.41%
- 距 52w 高
- -0.0%
- RSI(14)
- 66.6
- 趋势
- 多头
- SMA 20 / 50 / 200
- 4.60 / 4.53 / 4.28
- MACD / 信号
- 0.050 / 0.043
接近 52 周高多头排列
DX-Y.NYB
美元指数 DXY
$99.80
-0.21%
- 5 日
- -1.64%
- 距 52w 高
- -2.0%
- RSI(14)
- 35.1
- 趋势
- 中性
- SMA 20 / 50 / 200
- 100.94 / 100.42 / 99.14
- MACD / 信号
- 0.007 / 0.183
MACD 死叉 (3 天前)接近 52 周高
SPY
S&P 500 ETF
$747.03
+0.72%
- 5 日
- +1.10%
- 距 52w 高
- -1.8%
- RSI(14)
- 53.0
- 趋势
- 多头
- SMA 20 / 50 / 200
- 745.69 / 744.99 / 700.39
- MACD / 信号
- -0.822 / -0.051
接近 52 周高多头排列
QQQ
Nasdaq 100 ETF
$687.99
+0.65%
- 5 日
- +0.55%
- 距 52w 高
- -8.1%
- RSI(14)
- 45.1
- 趋势
- 中性
- SMA 20 / 50 / 200
- 701.02 / 715.09 / 645.04
- MACD / 信号
- -10.015 / -7.481
AAPL
Apple
$308.91
-7.35%
- 5 日
- -7.24%
- 距 52w 高
- -10.3%
- RSI(14)
- 43.2
- 趋势
- 中性
- SMA 20 / 50 / 200
- 324.37 / 309.50 / 277.96
- MACD / 信号
- 6.891 / 8.256
MACD 死叉 (今天)
MSFT
Microsoft
$464.72
+3.02%
- 5 日
- +21.75%
- 距 52w 高
- -16.1%
- RSI(14)
- 74.5
- 趋势
- 中性
- SMA 20 / 50 / 200
- 396.87 / 399.40 / 433.58
- MACD / 信号
- 9.300 / 1.831
RSI 超买
NVDA
Nvidia
$200.75
+2.93%
- 5 日
- -2.94%
- 距 52w 高
- -15.1%
- RSI(14)
- 48.3
- 趋势
- 中性
- SMA 20 / 50 / 200
- 203.30 / 206.17 / 193.11
- MACD / 信号
- -1.955 / -0.994
MACD 死叉 (4 天前)
GOOGL
Alphabet
$356.13
+6.73%
- 5 日
- +11.38%
- 距 52w 高
- -12.8%
- RSI(14)
- 54.3
- 趋势
- 中性
- SMA 20 / 50 / 200
- 348.05 / 358.74 / 326.27
- MACD / 信号
- -6.143 / -6.625
MACD 金叉 (今天)
TSLA
Tesla
$311.21
+0.76%
- 5 日
- -0.58%
- 距 52w 高
- -37.6%
- RSI(14)
- 31.7
- 趋势
- 空头
- SMA 20 / 50 / 200
- 363.93 / 390.76 / 411.60
- MACD / 信号
- -25.724 / -18.794
空头排列
META
Meta
$556.71
+3.28%
- 5 日
- -6.47%
- 距 52w 高
- -30.1%
- RSI(14)
- 37.8
- 趋势
- 空头
- SMA 20 / 50 / 200
- 620.80 / 602.18 / 634.49
- MACD / 信号
- -9.553 / 1.591
空头排列
BTC-USD
Bitcoin
$63,038.28
+0.36%
- 5 日
- -1.08%
- 距 52w 高
- -50.0%
- RSI(14)
- 45.1
- 趋势
- 空头
- SMA 20 / 50 / 200
- 64,389.84 / 63,375.49 / 71,293.79
- MACD / 信号
- -36.368 / 184.740
空头排列
ETH-USD
Ethereum
$1,870.67
+0.55%
- 5 日
- -1.05%
- 距 52w 高
- -62.2%
- RSI(14)
- 51.7
- 趋势
- 中性
- SMA 20 / 50 / 200
- 1,885.70 / 1,777.20 / 2,102.71
- MACD / 信号
- 28.989 / 36.414
MACD 死叉 (3 天前)
SOL-USD
Solana
$72.89
+0.14%
- 5 日
- -1.69%
- 距 52w 高
- -71.2%
- RSI(14)
- 42.3
- 趋势
- 空头
- SMA 20 / 50 / 200
- 75.41 / 74.90 / 86.08
- MACD / 信号
- -0.745 / -0.326
空头排列
BABA
阿里巴巴 (BABA)
$122.25
+5.10%
- 5 日
- +9.02%
- 距 52w 高
- -36.5%
- RSI(14)
- 64.2
- 趋势
- 中性
- SMA 20 / 50 / 200
- 113.41 / 114.08 / 141.06
- MACD / 信号
- 1.808 / 0.888
PDD
拼多多 (PDD)
$88.56
+1.28%
- 5 日
- +7.14%
- 距 52w 高
- -36.5%
- RSI(14)
- 60.1
- 趋势
- 中性
- SMA 20 / 50 / 200
- 85.12 / 84.14 / 103.82
- MACD / 信号
- 0.933 / 0.376
JD
京东 (JD)
$33.01
+2.17%
- 5 日
- +9.34%
- 距 52w 高
- -10.4%
- RSI(14)
- 79.1
- 趋势
- 中性
- SMA 20 / 50 / 200
- 29.74 / 28.83 / 29.44
- MACD / 信号
- 1.153 / 0.772
RSI 超买
0700.HK
腾讯控股 (0700.HK)
HK$475.20
+0.72%
- 5 日
- +9.34%
- 距 52w 高
- -30.4%
- RSI(14)
- 57.1
- 趋势
- 中性
- SMA 20 / 50 / 200
- 461.55 / 449.76 / 539.82
- MACD / 信号
- 3.361 / 2.135
MACD 金叉 (1 天前)
GC=F
黄金期货
$4,107.00
+0.17%
- 5 日
- +0.97%
- 距 52w 高
- -26.5%
- RSI(14)
- 45.6
- 趋势
- 空头
- SMA 20 / 50 / 200
- 4,067.18 / 4,200.92 / 4,479.49
- MACD / 信号
- -36.586 / -49.894
空头排列
CL=F
WTI 原油期货
$84.67
+1.29%
- 5 日
- -5.20%
- 距 52w 高
- -29.1%
- RSI(14)
- 54.6
- 趋势
- 多头
- SMA 20 / 50 / 200
- 80.28 / 82.21 / 75.79
- MACD / 信号
- 1.453 / 0.838
多头排列
USDCNY=X
美元 / 人民币
¥6.75
-0.07%
- 5 日
- -0.31%
- 距 52w 高
- -6.4%
- RSI(14)
- 32.6
- 趋势
- 空头
- SMA 20 / 50 / 200
- 6.77 / 6.78 / 6.91
- MACD / 信号
- -0.007 / -0.004
接近 52 周低空头排列
本报告仅供技术指标解读参考，过去走势不代表未来表现。投资决策需基于更全面的信息和考量。
Nirmal Purja, Renowned Mountaineer, Is Found Dead After Pakistan Avalanche
The British-Nepali mountaineer was one of 10 climbers who were struck by an avalanche on Thursday on Broad Peak, one of the world’s highest mountains.
中文摘要 著名的英国-尼泊尔登山者尼尔马尔·普尔贾在巴基斯坦博罗斯峰遭遇雪崩后被发现遇难。他是10名登山者中的一员，雪崩发生在周四，这座山是世界最高峰之一。
Spanish PM calls for urgent EU meeting after ‘selfish’ responses to Ceuta crisis
Pedro Sánchez hits out at ‘polarising and unlawful reaction’ over mass crossing into Spanish enclave in north Africa Europe live – latest updates Spain’s prime minister, Pedro Sánchez, has called for an urgent meeting of EU interior ministers after what he called some member states’ “selfish, polari
中文摘要 西班牙首相佩德罗·桑切斯针对一些欧盟国家对西属北非飞地塞乌塔危机反应不当，呼吁召开紧急会议以应对移民危机。
EU offered Spain support ‘from first moment’ of Ceuta crisis, says von der Leyen, after Sánchez hits out – Europe live
‘Not a single person reached mainland Spain or rest of EU,’ says European Commission president, after Spanish PM condemned ‘selfish, polarising’ reaction of some countries Back to Ukraine, where nine people were killed in Kyiv this morning from a Russian missile and drone attack. Damage was reported
中文摘要 欧盟委员会主席乌尔苏拉·冯德莱恩表示，欧盟在塞乌塔危机中从一开始就提供支持，此前桑切斯批评某些国家反应‘自私’和‘极端分化’。同时，乌克兰基辅发生俄方导弹和无人机袭击，造成9人遇难。
Migration Surge on the Spain-Morocco Border
Our Madrid bureau chief, Jason Horowitz, visits the coastal town of Ceuta, a Spanish territory in North Africa where there has been an unprecedented influx of migrants crossing the border from Morocco.
中文摘要 随着摩洛哥和西班牙之间的边境出现前所未有的移民潮，《纽约时报》的马德里分社记者探访了位于北非的西属飞地塞乌塔。
Justice Dept. Subpoenas Times Freelancer in Effort to Identify Sources
A previously undisclosed subpoena, involving reporting on a failed SEAL Team 6 mission in North Korea, reflects the government’s expanding campaign to investigate leaks of secrets.
中文摘要 美国司法部对《纽约时报》自由撰稿人发出传票，涉及对北朝鲜一次失败的海豹突击队6号任务的报道，此举反映出政府对泄露机密的调查力度加大。
India cracks down on protesters accused of ‘abusing’ Modi
Police cases, social media takedowns and online abuse have followed many who took part in India’s youth protests.
中文摘要 印度警方正在对参与抗议的年轻人采取措施，其中一些人被指控“侮辱”总理莫迪，包括警方立案、社交媒体内容删除等。
Mountaineer Nirmal Purja killed in Pakistan avalanche, his company says
The renowned mountaineer was known for scaling the world's 14 highest peaks in just over six months in 2019.
中文摘要 尼尔马尔·普尔贾的公司确认，这位著名登山者在巴基斯坦雪崩中遇难。他在2019年曾在半年内成功攀登世界14座最高峰。
Israeli strike destroys Gaza hospital warehouse
Israeli strike destroys medical supplies warehouse at Gaza's Al-Aqsa Martyrs Hospital, staff say.
中文摘要 以色列对加沙的阿克萨烈士医院进行袭击，摧毁了医疗供应仓库。
人形机器人：特朗普对华贸易战的最新战场？
美国已禁止外国制造的人形机器人进入本国市场，而这一市场目前由中国科技企业占据主导地位。尽管华盛顿方面称此举是出于国家安全考虑，但分析人士认为，美国总统特朗普实际上是在为本国机器人制造商争取追赶中国竞争对手的时间。
中文摘要 美国禁止外国制造的人形机器人进入市场，此举被认为是为了美国本土机器人的发展，同时也是对中国机器人的竞争应对。
Is Spain being punished?
In the months leading up to the crisis in Ceuta, pro-Israel voices conspicuously raised the issue of the enclave
中文摘要 在塞乌塔危机前的几个月里，支持以色列的声音在社交媒体上涌现，争议聚焦于该飞地问题。
Dozens injured as earthquake hits near Naples in seismic swarm
At least five taken to hospital and about 300 evacuated in strongest quake to hit southern Italian region in 40 years At least 26 people have been injured after a 4.7-magnitude earthquake rocked a region near Naples, in southern Italy. The quake, which occurred at 7.46pm on Friday night, caused powe
中文摘要 意大利南部近那不勒斯发生了4.7级地震，导致至少26人受伤，5人住院，约300人被疏散。这是该地区40年来最强的地震。
Russian Missiles Kill 9 in Kyiv After Trump Backs Off Air-Defense Pledge
Ukraine’s supply of Patriot missile interceptors is dwindling, and on Friday, President Trump backed away from his promise to let Ukraine make them itself.
中文摘要 在乌克兰，俄方导弹袭击基辅，导致9人死亡。与此同时，特朗普放弃了支持乌克兰自行制造爱国者导弹的承诺。
Children run marathon to spotlight Gaza’s growing orphan crises
Children run marathon to spotlight Gaza's growing orphan crisis
中文摘要 儿童们举行马拉松活动，以引起人们对加沙日益增长的孤儿危机的关注。
Nine killed in strikes on Kyiv, as Ukraine sinks Russian container ship
The attack comes as Moscow escalates deadly ballistic missile attacks on the Ukrainian capital.
中文摘要 在基辅的空袭造成9人遇难，俄方同时在海上打击乌克兰，沉没了一艘俄罗斯集装箱船。
长平观察：为什么中国培养不出菲尔兹奖学者？
中国官方舆论仍在为两位中国籍学者获得数学菲尔兹奖狂欢，同时一些网民开始吐槽“受挫”的求学经历。时事评论作家长平指出，同样的新闻题材，在二十年前的中国舆论中可能会有不一样的反应。
中文摘要 中国关于两位学者获得菲尔兹奖的庆祝声中，一些网民对自身求学经历感到挫败。时事评论员长平指出，二十年前舆论会有不同反应。
The Father of 401(k) Has A New Retirement Plan
Known as the "Father of 401(k)" Ted Benna has co-created Radish: a new employer-funded retirement plan that allows employees to access funds for immediate financial needs such as car repairs or medical expenses without relying on paycheck deductions. Joining David Gura and Christina Ruffini on Bloom
中文摘要 被誉为“401(k)之父”的Ted Benna与他人共同创建了Radish，这是一个新的雇主资助退休计划，让员工能够在不依赖工资扣除的情况下，提前使用资金应付汽车维修或医疗费用等紧急财务需求。
Ukraine Gains Ground as Putin Faces Pressure
Former US Ambassador Bill Taylor joins Bloomberg This Weekend and says Ukraine maintains momentum on the battlefield despite intensified Russian missile attacks, while urging the US and its allies to accelerate deliveries of Patriot interceptors. Taylor explains to hosts David Gura and Christina Ruf
中文摘要 美国前大使比尔·泰勒表示，尽管俄罗斯的导弹攻击加强，乌克兰在战场上仍保持势头。他呼吁美国及其盟友加快派遣爱国者防空系统的步伐。
Can Mergers Close Europe’s Tech Gap?
The European Union is rewriting its merger rules for the first time in two decades, hoping to create business “champions” capable of competing with the US. Ingrid Vandenborre of Skadden and Yale’s Fiona Scott Morton break down what's changing, why it matters for investors, and whether competition po
中文摘要 欧盟首次在20年内修订合并规则，旨在打造具有竞争力的商业“冠军”，以应对美国公司。Skadden律师事务所的Ingrid Vandenborre和耶鲁大学的Fiona Scott Morton解析了此次变化及其对投资者的重要性。
Wildfires Spread Across Southern Europe
Bloomberg Athens Deputy Bureau Chief Paul Tugwell says Greece's wildfire response is under severe strain as gale-force winds, dry conditions and difficult terrain hamper firefighting efforts, with major blazes continuing to spread in central and southern parts of the country. He says European Union
中文摘要 希腊在面对强风、干燥天气及困难地形的挑战下，野火响应受到严重压力，中央及南部地区主要火灾仍在蔓延。
FIFA Faces Backlash Over Investment Plan
Bloomberg Senior Sports Reporter Randall Williams says FIFA President Gianni Infantino is under mounting pressure after withdrawing plans to sell a stake in the World Cup following opposition from UEFA and other soccer leaders. He says critics argue FIFA's record revenues and multibillion-dollar res
中文摘要 国际足联主席贾尼·因凡蒂诺因撤回出售世界杯股份计划而面临来自欧洲足球协会等团体的压力，批评者认为国际足联的收入创纪录而管理层决策不力。
Exxon, Chevron Warn Fuel Prices to Endure as War Knocks Refining
High fuel prices are likely to stick around even if oil prices drop in the coming months as the wars in Russia and Middle East leave global refining capacity critically short, ExxonMobil Holdings Corp. and Chevron Corp. warned.
中文摘要 埃克森美孚和雪佛龙警告称，即使未来几个月油价下跌，高油价仍可能持续，因俄乌战争和中东局势导致全球炼油能力严重短缺。
Bipartisan Governors Launch Task Force to Tackle Immigration Reform
From Oklahoma City, the site of the annual meeting of the National Governor’s Association, Oklahoma Governor Kevin Stitt, Marylands Governor Wes Moore, Utah Governor Spencer Coz, and Delaware Governor Matt Meyer joined Bloomberg This Weekend host Christina Ruffini to announce the launching of a bi-p
中文摘要 在全国州长协会年会上，来自各州的两党州长宣布成立任务小组，旨在解决移民改革问题。
Warsh’s Silent Treatment Has S&P Traders Bracing for Wild Swings
A relentless run of uncertainty, from war to tariff fights to inflation that refuses to die, has challenged investors this year. Now they have to balance another risk: A Federal Reserve that won’t tell them what it’s thinking.
中文摘要 疫情、关税争端和持续高通胀带来的不确定性对投资者构成挑战，投资者还需应对美联储缺乏沟通所带来的风险。
Faisal Islam: Four reasons why Fifa's World Cup plan never stacked up
The actual presentation of FIFA's plan to part-privatise the World Cup raises many more questions than it answers.
中文摘要 关于国际足联部分私有化世界杯计划的实际呈现，提出了更多问题，远未找到答案。
Phoenix Extends Record-Breaking Heat as West Faces Fire Threat
Phoenix has had a record-breaking run of high temperatures and that streak will continue through the coming week, as heat and fire risks spread across the wider US West.
中文摘要 凤凰城经历破纪录的高温天气，并预计这一情况将在未来一周继续，致使美国西部地区面临火灾风险。
By The Way: Headlines You May Have Missed the Week of 7/26
The secret to a billion dollar movie, Willie Nelson speaks out against Data Centers, mystery packages, and Marmots on OnlyFans. Bloomberg News' Tim Stenovec joins David Gura and Christina Ruffini on Bloomberg This Weekend to share headlines you may have missed. (Source: Bloomberg)
中文摘要 Bloomberg新闻的Tim Stenovec分享了最近一些你可能错过的头条，包括一部十亿美元电影的秘密、Willie Nelson对数据中心的反对以及神秘包裹等话题。
India Youth Protests Shake Modi Government
India's youth-led Cockroach Movement gains momentum as protests over exam paper leaks and rising unemployment force the resignation of the country's education minister. Bloomberg News Senior Reporter Dan Strumpf joins David Gura and Christina Ruffini on Bloomberg This Weekend and explains how the mo
中文摘要 印度青年主导的“蟑螂运动”因考试泄露和失业问题的抗议活动迫使教育部长辞职，运动逐渐获得动力。
现在做 API 中转站怎么样？
16 回复 · 程序员 节点
请职场大佬请教，准备跳槽前应该是一个什么状态
9 回复 · 程序员 节点
7 月份独立开发复盘
23 回复 · 程序员 节点
DeepSeek 官方 Harness 不一定会开源
8 回复 · 程序员 节点
在 DeepSeek 官方 Harness 没有出来之前，用什么 Agent 更好？
50 回复 · 程序员 节点
如果下个月 AI 按 Apple ID 来区分的话，是不是还得注册个港区的 iD？
7 回复 · Apple 节点
App Store 中国区充值返 10%各位充了多少？
46 回复 · Apple 节点
M1 Max 能装 macOS 27 Public Beta 么？
16 回复 · Apple 节点
微信在 Mac 上频繁闪退（等）
7 回复 · Apple 节点
Apple Store App 登录美区账号却强制显示国区，有人遇到过吗？
8 回复 · Apple 节点
该源今日无内容。