每日简报

2026-10-29

← 历史归档

JetBrains/thinkrail

TypeScript · ★ 5,012 · 🍴 259 · 📈 1,075 stars today

Vibe code with pi in a lightweight, real IDE that customises itself around the way you work — The Vibe You Need

中文介绍 JetBrains 推出的轻量级实时 IDE,可根据你的工作方式自动定制界面与流程,内置 AI 辅助编程能力,旨在提供贴合个人习惯的开发体验。

elayadesign/ai-design-skills

★ 11,082 · 🍴 795 · 📈 1,938 stars today

中文介绍 面向 AI 辅助设计的一套技能集,为 AI 编程工具提供设计相关的指令与工作流,帮助开发者通过 AI 生成更符合设计规范与审美的界面。

obra/superpowers

Shell · ★ 307,420 · 🍴 27,348 · 📈 1,096 stars today

An agentic skills framework & software development methodology that works.

中文介绍 一套代理式技能框架与软件开发方法论,定义 AI Agent 在执行任务时的能力模块与流程,旨在提升 AI 驱动开发的可靠性与效率。

HKUDS/DeepTutor

Python · ★ 47,926 · 🍴 5,947 · 📈 2,376 stars today

DeepTutor: Lifelong Personalized Tutoring. https://deeptutor.info/.

中文介绍 香港大学数据科学实验室开发的终身个性化辅导系统,基于大模型为学习者提供自适应学习路径与个性化答疑,可长期跟踪学习状态并动态调整教学策略。

macro-inc/macro

Rust · ★ 5,406 · 🍴 517 · 📈 106 stars today

Macro is a unified workspace for teams: email, chat, docs, tasks, agents, calls, and CRM — @-linked together with shared AI memory.

中文介绍 面向团队的统一工作空间,整合邮件、聊天、文档、任务、AI 代理、通话与 CRM,通过 @ 链接实现跨工具协作,并共享统一 AI 记忆以提升信息流转效率。

Nutlope/hallmark

CSS · ★ 40,666 · 🍴 2,111 · 📈 2,109 stars today

Anti-AI-slop design skill for Claude Code, Cursor, and Codex.

中文介绍 为 Claude Code、Cursor 和 Codex 等 AI 编程工具提供反 AI 味的设计技能包,通过明确的设计准则与审查流程,减少 AI 生成的粗糙界面,提升输出品质。

pbakaus/impeccable

Rust · ★ 92,590 · 🍴 5,477 · 📈 1,471 stars today

The design vocabulary for your coding agent. Steer with polish and bolder, review with critique and audit, catch AI slop before it ships.

中文介绍 为编码代理提供设计词汇表与审查技能,涵盖打磨、批评、审计等环节,帮助开发者在 AI 生成的代码中及早发现并修正设计问题,防止低质输出上线。

makecindy/cindy

TypeScript · ★ 8,076 · 🍴 943 · 📈 740 stars today

Consider it done. The open-source AI agent that works out of the box · 想到,就能做到。开源、开箱即用的 AI Agent。

中文介绍 开源、开箱即用的 AI Agent,能理解自然语言指令并自动执行多步骤任务,适用于日常办公自动化、信息整理与流程编排等场景,降低 AI 使用门槛。

nightly-labs/openbot

TypeScript · ★ 3,228 · 🍴 315 · 📈 1,053 stars today

A local-first desktop workspace for persistent AI teammates. Run Codex, Claude, and Grok with dedicated workspaces, task queues, file sharing, browser control, and agent-to-agent collaboration.

中文介绍 本地优先的桌面工作区,用于运行持久化 AI 队友。支持 Codex、Claude、Grok 等模型,提供独立工作区、任务队列、文件共享、浏览器控制及代理间协作功能。

togg53192-cmd/jailbreaks

★ 7,060 · 🍴 1,013 · 📈 1,478 stars today

LIST OF ALL MY JAILBREAKS

中文介绍 作者收集整理的各类 AI 越狱提示词列表,用于绕过模型安全限制,探索模型边界行为,仅供安全研究与教育参考。

hmjz100/LinkSwift

JavaScript · ★ 34,547 · 🍴 2,443 · 📈 1,380 stars today

一个基于 JavaScript 的网盘文件下载地址获取工具。基于【网盘直链下载助手】修改 ,支持 百度网盘 / 阿里云盘 / 中国移动云盘 / 天翼云盘 / 迅雷云盘 / 夸克网盘 / UC网盘 / 123云盘 八大网盘

中文介绍 基于 JavaScript 的网盘文件下载地址获取工具,由网盘直链下载助手修改而来,支持百度、阿里、移动、天翼、迅雷、夸克、UC 及 123 云盘等八大网盘。

expo/skills

Shell · ★ 3,558 · 🍴 226 · 📈 216 stars today

A collection of AI agent skills for working with Expo projects and Expo Application Services

中文介绍 专为 Expo 项目与 Expo Application Services 设计的 AI Agent 技能合集,帮助开发者通过 AI 工具更高效地处理 Expo 项目配置、构建与部署等常见任务。

ruvnet/RuView

Rust · ★ 101,019 · 🍴 13,160 · 📈 596 stars today

π RuView turns commodity WiFi signals into real-time spatial intelligence, vital sign monitoring, and presence detection — all without a single pixel of video.

中文介绍 利用普通 WiFi 信号实现实时空间感知、生命体征监测与存在检测的技术,无需摄像头即可获取环境信息,适用于隐私敏感场景与智能家居。

xiaolai/the-craft-of-selfteaching

Jupyter Notebook · ★ 18,484 · 🍴 17,886 · 📈 295 stars today

One has no future if one couldn't teach themself.

中文介绍 李笑来关于自学方法的作品,探讨如何通过自主教学实现个人成长,强调自学能力对未来发展的关键作用,并提供实操路径与思考框架。

Multiscale Policy Optimization for Long-Horizon LLM Agents

👍 372

Long-horizon LLM agents interact with their environments through complete assistant turns while generating each action token by token. We observe that critic values vary primarily across turns, motivating optimization anchored at turns with finer credit allocation among tokens. We propose Multiscale

中文介绍 该论文针对长视程LLM智能体提出多尺度策略优化(Multiscale Policy Optimization)。由于批评值主要在回合间变化,该方法以回合为锚点,在回合内进行更细粒度的信用分配,优化令牌生成。此策略有望提升长程决策性能,但摘要未提供量化结果。

The Trinity:Capability-Driven Recursive Self-Improvement for Multimodal Agents

👍 311

Multimodal intelligence can be viewed as human-guided co-evolution between worlds and agents through interaction. However, scaling this process faces two bottlenecks: constructing heterogeneous, verifiable environments is costly, and manual Harness iteration cannot keep pace with growing task divers

中文介绍 该论文提出「Trinity」框架,通过能力驱动的递归自改进实现多模态智能体与环境的协同进化。为解决环境构建成本高和人工Harness迭代慢两大瓶颈,Trinity采用能力驱动的自改进机制,使智能体能够自主构建可验证环境并递归提升自身能力,以适应任务多样性增长。

VibeVideoBench: Evaluating Multimodal Agents on Code-Driven Video Generation

👍 171

Diffusion video models produce striking clips, but they cannot reliably render an exact formula, a correct chart, a rule-consistent game or narration synchronized to on-screen events. A different route is emerging: multimodal agents that write code to render every frame and synthesize every sound. W

中文介绍 论文提出新基准VibeVideoBench,用于评估通过编写代码来生成视频的多模态智能体。相较于扩散模型,该方法能精确渲染公式、图表及规则一致的游戏,并同步音视频。基准涵盖多种代码驱动视频生成任务,以评估智能体理解和执行复杂指令的能力。

Memory Tax: Dissecting the Net Value of Agent Memory

👍 111

Memory is increasingly used in long-horizon agents, yet the net value of building memory over direct retrieval from raw history remains largely unclear. Through a large-scale evaluation across multiple benchmarks and model backbones, we find many dedicated memory systems underperform naive RAG in ma

中文介绍 论文通过大规模多基准评估,剖析了长视程智能体中构建记忆网络的净价值。研究发现,许多专用记忆系统在多数任务上表现不如直接对原始历史进行RAG检索,揭示了「记忆税」问题。该结果挑战了当前广泛使用的复杂记忆架构,并为设计高效记忆机制提供了实证依据。

Can Test-Time Memory Evolution Outperform Test-Time Training for Scientific Discovery?

👍 109

Recent test-time training (TTT) frameworks have enabled open-weight large language models (LLMs) to outperform inference-only frameworks powered by frontier models on scientific discovery tasks. However, TTT introduces additional computational cost for online weight updates during the discovery proc

中文介绍 论文比较了测试时记忆演化与测试时训练(TTT)在科学发现任务中的性能。TTT虽能提升开源LLM性能,但需在线更新权重,计算成本高。研究者提出测试时记忆演化方法,在不更新权重的情况下动态调整记忆,旨在以更低成本达到或超越TTT效果,但需验证具体性能差异。

Can Discrete Tokens Serve as Semantically Lossless Representations of Visual Signals?

👍 88

Discrete visual tokens represent vision in a discrete form analogous to language, offering a native interface for multimodal modeling under a shared next-token prediction objective. However, quantization error is widely assumed to cause unavoidable semantic loss, fundamentally limiting the semantic

中文介绍 论文探讨离散视觉令牌是否可作为视觉信号的无损语义表示。传统观点认为量化误差会导致不可避免的语义损失。该工作挑战此假设,提出新的离散化方法或训练策略,可能通过改善量化误差或利用其结构信息,使离散令牌在语义上接近连续表示,从而提升多模态模型的视觉理解能力。

HABIT-Bench: Evaluating Habit Induction from Longitudinal Weak Evidence in Agent Memory

👍 78

In long-term interactions with large language model (LLM) agents, users reveal habits through their choices and actions. These habits capture implicit user information that agents must infer across conversations. However, individual observations may provide only partial evidence about a habit or whe

中文介绍 论文提出HABIT-Bench基准,评估LLM智能体从长时交互中推断用户习惯的能力。由于个体观察仅提供部分证据,习惯推断颇具挑战。该基准提供纵向弱证据环境,测试智能体能否整合跨对话信息,识别隐性用户模式,并可能评估其预测和适应能力。

Rollback: Experience Backtracking in World Models for Policy Improvement

👍 50

Action-conditioned world models offer a scalable alternative to costly real-world interaction for improving robot policies. However, existing world-model-based methods largely treat world models as forward-only surrogates of the physical environment, leaving untapped a capability no physical environ

中文介绍 论文提出Rollback方法,利用世界模型中的经验回溯来改进机器人策略。传统世界模型仅作前向代理,忽略回溯能力。Rollback通过回溯过往经验,使智能体能够从错误中学习并优化决策。在机器人控制任务中,该方法有望减少真实交互成本,提升策略收敛速度。

CodeCap-Bench: Can Code Serve as a Good Video Caption?

👍 49

Video captioning aims to preserve visual information through compact descriptions, predominantly using natural language. However, precisely expressing fine-grained spatial structures and temporal dynamics in natural language often leads to lengthy and potentially ambiguous descriptions. In this work

中文介绍 论文提出CodeCap-Bench基准,探索代码是否可作为视频描述的有效替代方案。自然语言描述在表达细粒度空间结构和时间动态时往往冗长且模糊。该研究评估代码描述在视频理解中的表现,发现其可能在保持精确性的同时更紧凑,为视频理解提供新范式。

ThinkWorld: World Imagination Reinforces Chain-of-Dynamics Reasoning

👍 47

Vision-Language-Action (VLA) models have made substantial progress in robotic manipulation, but their generalization remains constrained by the diversity and coverage of action-labeled robot demonstrations. Human videos provide much broader behavioral and environmental experience, while learning fro

中文介绍 论文提出ThinkWorld框架,通过世界想象强化链式动力学推理。为缓解VLA模型因动作标注数据不足导致的泛化瓶颈,该方法利用人类视频丰富经验,构建世界模型以模拟动力学,从而增强机器人操作策略的泛化能力,无需额外动作标签。

Zooming without Zooming v2: From Seeing Details to Connecting Evidence for Generalizable Multimodal Perception

👍 24

Multimodal perception forms the foundation of multimodal intelligence. In particular, spatial perception and multimodal long-context understanding are two pivotal tasks that underpin embodied AI and multimodal agents. However, synthesizing targeted, high-quality training data for these tasks remains

中文介绍 该论文介绍Zooming without Zooming v2,用于提升多模态感知的泛化能力。该方法专注于空间感知和多模态长上下文理解,通过合成高质量训练数据来连接证据,使模型能从细节扩展到全局证据关联。此框架有望在无需大规模人工标注下增强多模态智能体的感知能力。

On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction

👍 22

Diffusion language models (DLMs) promise parallel text generation, but in practice face an unresolved speed-quality trade-off. Exact decoding fills in only one masked position per model pass, while faster multi-token decoding schemes are lossy. Speculative decoding is the standard lossless accelerat

中文介绍 论文提出SPOT方法,增强扩散语言模型(DLM)的投机解码。DLM存在速度与质量权衡困境,而SPOT通过预测提交顺序实现无损加速。该方法能并行解码多个位置,同时保持与精确解码相同的质量,有效提升DLM生成速度,解决其速度瓶颈。

RSI Arena: Towards Recursive Self-Improvement with Human Feedback

👍 19

Can AI agents autonomously train useful language models and improve them using feedback from users? We introduce RSI Arena, a two-stage study of autonomous post-training as a step toward recursive self-improvement. Ten frontier models act as training agents, each starting from NVIDIA Nemotron 3.5 Li

中文介绍 论文介绍RSI Arena,研究基于人类反馈的递归自改进。十个前沿模型作为训练代理,从NVIDIA Nemotron 3.5开始,通过用户反馈自主训练和提升语言模型。该两阶段研究探索AI能否自主完成后训练并利用人类反馈实现递归改进,为AGI研究提供实证基础。

Channel Selection for Mixture-of-Experts Decoding Needs No Calibration Data

👍 18

Mixture-of-Experts (MoE) models are bottlenecked by loading routed-expert weights from memory during interactive decoding at small batch sizes. Because most intermediate channels in a routed expert contribute little to any single token, channel selectors predict which channels matter and read only t

中文介绍 论文提出无需校准数据的通道选择方法,用于加速MoE模型解码。由于路由专家权重加载是瓶颈,且多数中间通道对单令牌贡献小,该方法预测关键通道并仅加载它们,从而减少内存访问。此方法无需校准数据集,直接提升小批量解码效率,降低推理延迟。

What Effort Level Should I Choose? On the Economics of Reasoning Effort in LLM Agents

👍 15

Generative models increasingly expose a reasoning effort setting that lets users scale inference-time compute, trading higher cost for potentially improved performance. Users must choose an effort level before knowing whether extra reasoning will help, and even a successful run does not reveal wheth

中文介绍 论文从经济学角度研究LLM智能体中推理努力级别的选择问题。用户需在不知额外推理是否有益的情况下预选努力级别,且成功运行不揭示其性价比。该研究建立决策框架,分析推理成本与性能的权衡,帮助用户选择最优努力级别,最大化效用。

h1: A Taste of the Bitter Lesson in Automated Theorem Proving

👍 13

While progress in informal mathematics has come largely from scaling the chain-of-thought and agentic reasoning of frontier language models, formal theorem proving has increasingly relied on human-designed workflows around these models, such as planning, decomposition, and orchestrating specialized

中文介绍 论文借鉴「苦涩的教训」,提出h1方法,用于自动定理证明。与依赖人工设计的规划和分解流程不同,h1将规模扩展与简单搜索结合,利用学习到的启发式引导证明搜索。该方法有望减少人工干预,提升形式化定理证明的自动化水平,但需验证实际性能。

Omni-InteractScope: Benchmarking Proactive and Agentic Interaction in Continuous Omnimodal Streams

👍 11

Omnimodal large language models (Omni-LLMs) are increasingly expected to assist users while observing continuous audiovisual streams. However, existing evaluations provide limited insight into how reliably these models connect stream evidence and user needs to timely, sustained interventions. We int

中文介绍 论文提出Omni-InteractScope基准,评估全模态LLM在连续音视频流中的主动交互能力。该基准关注模型如何连接流证据与用户需求,并及时、持续地进行干预。通过测试主动性和代理性行为,揭示现有模型在动态多模态环境下的可靠性不足。

Beyond the Estimate: Wiener-Informed Decoding for Easier Training of Pixel-Space Flow Models

👍 9

Learning velocity fields directly in pixel space is difficult and slow. We show that coarse second-order image statistics can ease this task: frozen Wiener estimates improve velocity learning, with gains extending beyond the linear predictions they explicitly provide. Moreover, we observe similar bu

中文介绍 论文提出Wiener-Informed Decoding方法,简化像素空间流模型训练。利用冻结的Wiener估计(二阶图像统计量)辅助速度场学习,能改善训练效率,且增益超越线性预测。此方法降低像素空间流模型训练难度,加速收敛,且可能推广到其他生成任务。

Your Policy Is Secretly a Harness: The Decision Geometry of Agent Interfaces

👍 6

Language model agents are post-trained and deployed through a harness, the prompts, tools, and control logic that present a task to the model and execute its outputs. Harnesses are often revised and judged by how an already trained model performs under them, which assumes that what the model learned

中文介绍 论文揭示语言模型智能体的策略本质上是一种「Harness」,即提示、工具和控制逻辑的接口。当前Harness的修改和评估基于已训练模型的性能,忽略了策略对接口的依赖。该研究通过决策几何分析,说明策略与Harness的不可分性,为智能体部署设计提供新视角。

SearchHorizonBench: A Benchmark for Search Value in AI Agents

👍 6

AI agents must decide when another search is worth its cost. Answer-success probabilities capture question difficulty, while this decision requires estimating the additional benefit of search. We introduce SearchHorizonBench, a benchmark for learning and evaluating search value from execution experi

中文介绍 论文引入SearchHorizonBench基准,用于评估AI智能体的搜索价值决策能力。该基准要求智能体根据执行经验学习何时值得进行额外搜索,以平衡成本与收益。通过此基准,研究者可系统衡量智能体在不确定环境中优化搜索策略的能力。

Databricks & the Future of the Firm

@bhorowitz · 768.2K 粉丝 · 187.3K 阅 · 518 赞 · 71 转

So many aspects of our lives are going to change over the next twenty years. And one that’s easily in the top five is the question: “What do you do at work all day?” Two massive breakthroughs are on

中文介绍 探讨未来20年工作方式的巨变,认为两大突破将重塑日常职业,Databricks 与未来企业形态成为核心议题,思考AI如何改变人们在工作中的时间分配。

What I brought home from AGNTCon + MCPCon

@BradGroux · 7.1K 粉丝 · 145.7K 阅 · 509 赞 · 3 转

During Andrew Nolan’s session at AGNTCon + MCPCon last week, I asked where field workers fit into the agentic web. I work with businesses where people wear coveralls, operate equipment, and spend much

中文介绍 作者参加 AGNTCon + MCPCon 后,聚焦现场工人如何在智能体网络中发挥作用。结合自身服务的企业,反思AI代理如何适配一线操作员的需求,强调非办公室岗位的AI接入价值。

Replit and the dawn of the self-driving company

@pmarca · 6.7M 粉丝 · 92.9K 阅 · 587 赞 · 78 转

One thing I love about Amjad is, sometimes he really says the thing and articulates something that other people weren’t brave enough or clear enough in their thinking to express. He has one of those

中文介绍 Marc Andreessen 评价 Amjad 关于 Replit 的构想,称其点出了他人不敢明说的趋势:AI 正催生「自动驾驶公司」,即组织在无人类持续干预下自我运转的新形态。

How the Claude Code team does code review

@addyosmani · 472.8K 粉丝 · 70.4K 阅 · 1.2K 赞 · 85 转

We previously mentioned that Claude writes about 90 percent of the code for Claude Code. So how do we do code review at this scale? Who reads code anyway? What happens if code goes wrong? This post

中文介绍 Claude Code 团队 90% 代码由 Claude 编写,本文分享其规模化代码审查实践:谁读代码、出错如何处理,为AI主导开发流程提供一线经验。

Getting started with Claude Code Mods

@addyosmani · 472.8K 粉丝 · 46.3K 阅 · 553 赞 · 52 转

Mods were released in Claude Code a few weeks ago, and I'm sharing this guide to make it easier to get started with them. Mods are small JavaScript or TypeScript files that you run within your Claude

中文介绍 介绍 Claude Code 新推出的 Mods 功能:小型 JS/TS 文件在会话内运行,可定制化扩展工作流。提供入门指南,帮助开发者快速上手自定义行为。

How I Design with AI. (pt 2)

@reactiverobot · 1.9K 粉丝 · 40.3K 阅 · 508 赞 · 36 转

I hate slop. You hate slop. Agents love sneaking slop into our designs. In Part 1, I explained my AI design process for individual surfaces. The reality is that today we build complex features

中文介绍 AI 设计流程续篇:针对复杂功能界面,揭示智能体如何偷偷引入「垃圾设计」。作者分享多层级设计策略,强调人工审美在AI工作流中的不可替代性。

Biology Beyond Human Intuition

@DrMorganLevine · 22.4K 粉丝 · 33.6K 阅 · 500 赞 · 76 转

Why understanding life may require learning to see it differently Living systems are extraordinarily complex. They operate across molecular, cellular, tissue, organismal, and ecological scales, with

中文介绍 反思生物学研究:生命系统跨分子、细胞、组织等多尺度运作,复杂性远超直觉。呼吁通过新视角观察,才能理解生命的真实运作机制。

Grok Bot is far more capable than you think. Fix these mistakes holding it back

@XFreeze · 291.5K 粉丝 · 33.4K 阅 · 509 赞 · 62 转

Most people are barely scratching the surface of what Grok Bot is capable of Not because the Bot lacks capabilities, but because a handful of habits are quietly holding it back Here are some common

中文介绍 指出 Grok Bot 能力被低估,多数人因几个坏习惯限制其表现。列举常见误区并给出修正方案,帮助用户解锁更高效的智能体应用。

Inside Episode 1: The Oath

@imagine · 105.2K 粉丝 · 25.5K 阅 · 599 赞 · 70 转

We're sharing our workflows, from script to final edit, so you can learn alongside us and use these techniques in your own creative projects. Recently, we’ve partnered with Wonder Studios on a series

中文介绍 与 Wonder Studios 合作推出系列,公开从剧本到成片的完整工作流。旨在让创作者边看边学,将AI辅助技术应用于个人影视项目。

Creating AI value with enterprise context

Across a business, any number of AI platforms, systems, and tools may be at work simultaneously—for example, AI-powered customer platforms, conversational assistants, and workflow and process automation tools. Each one is sophisticated, advanced, and quick to respond. But, when set to work, the outc

中文介绍 MIT科技评论指出,企业内多个AI平台和工具并行运作,如客户平台、对话助手和流程自动化。每个系统独立高效,但缺乏统一协调,限制了整体价值。文章探讨如何通过企业上下文整合AI,释放更大潜力。

Why Agents Love Supabase

Paul Copplestone and Ant Wilson on building the backend for AI agents

中文介绍 Latent Space访谈Supabase创始人Paul Copplestone和Ant Wilson,讨论为何AI智能体偏好使用Supabase作为后端。他们分享如何利用Supabase构建可靠的AI代理基础设施,包括数据库、认证和实时功能。

Gemini 4 Carbon: Our next frontier model

中文介绍 DeepMind发布新前沿模型Gemini 4 Carbon。该模型旨在处理碳相关任务,可能应用于气候建模、碳排放追踪等领域。详情尚未披露,但预示谷歌在环保AI方向的新布局。

OpenAI and ServiceNow expand partnership across enterprise workflows

OpenAI and ServiceNow expand their partnership to bring OpenAI models and agents into ServiceNow and ServiceNow workflows into ChatGPT and Codex.

中文介绍 OpenAI与ServiceNow扩大合作,将OpenAI模型和代理引入ServiceNow平台,同时将ServiceNow工作流整合到ChatGPT和Codex中。此举旨在打通企业级工作流,提升自动化效率。

Pairing industrial AI autonomy and human control

中文介绍 MIT科技评论探讨工业AI自主性与人类控制的平衡。文章分析在自动化程度提升的同时,如何保持人类监督和决策权,确保安全与效率,并讨论人机协作的最佳实践。

The future of the firm

How AI might enter the org chart—beneath, beside, and between us.

中文介绍 OpenAI发文探讨AI如何进入企业组织结构——“之下、旁边、之间”。文章分析AI代理可能重塑公司层级、协作方式和决策流程,预测组织形态的演变。

Legora is building the agentic operating system for legal work

See how Legora uses OpenAI models, ChatGPT, and Codex to accelerate legal workflows, expand contract analysis, and free lawyers to focus on higher-value work.

中文介绍 Legora利用OpenAI模型、ChatGPT和Codex构建法律工作的代理操作系统,加速法律流程、扩大合同分析范围,并让律师专注于高价值工作。展示AI在法律行业的具体应用场景。

[AINews] Anthropic ends Coding Interviews: "Hunger beats tenure"

Since coding is solved, why code for the job?

中文介绍 Anthropic宣布结束编程面试,引发关注。该决策基于“编码已被解决,为何还要编码应聘”的逻辑,强调更看重候选人的动机和潜力(“饥饿感”),而非资历或学历。

How Anthropic ships Claude with Claude: Separating the Brain from the Hands

Katelyn Lesse and Angela Jiang on Opus 5.5, the Claude Platform, and the engineering behind AI agents.

中文介绍 Latent Space访谈Anthropic的Katelyn Lesse和Angela Jiang,讨论Opus 5.5、Claude平台及AI代理工程实践。核心观点是分离模型“大脑”与代理“手”,以提升安全性和可靠性。

How Anthropic ships the #1 LLM in the World safely: Separating the Brain from the Hands

Katelyn Lesse and Angela Jiang join Latent Space to explain the engineering behind the Claude Platform: launching Opus 5.5, separating the model’s “brain” from the agent’s “hands,” scaling APIs, and building reliable managed agents. A new introduction ...

中文介绍 Anthropic的Katelyn Lesse和Angela Jiang在Latent Space解释Claude平台背后的工程:发布Opus 5.5,将模型“大脑”与代理“手”分离,扩展API,构建可靠托管代理。介绍如何安全交付全球第一LLM。

Build specialized multimodal models with LFM3

中文介绍 LiquidAI发布LFM3,用于构建专用多模态模型。该模型支持文本、图像等模态融合,旨在为特定任务提供定制化解决方案,提升多模态应用的灵活性和性能。

Can we trust LLMs to judge whether a research idea is novel?

中文介绍 艾伦AI研究所提出Novelty Bench基准,测试LLM能否判断研究想法的新颖性。该研究评估LLM的评判能力,为科研创新提供辅助工具,但也提出对可靠性的质疑。

Genie 4: A new generation of world simulation

Genie 4: our new world model unlocking unprecedented realism in real-time simulation.

中文介绍 DeepMind发布Genie 4世界模型,实现实时模拟中前所未有的真实感。该模型可生成交互式虚拟环境,应用于机器人训练、游戏开发和具身AI等领域。

市场总览

美股整体偏中性:SPY现价772.31略高于SMA20(770.90),RSI 53.3正常,但MACD微负;QQQ价格725.61低于SMA20(732.48),RSI 49.6中性,MACD负值扩大;大型科技股分化,TSLA RSI 63.6偏强且MACD金叉,META RSI 30.7接近超卖、5日跌12.19%。加密市场中性偏弱:加密恐慌贪婪指数51中性,总市值2.74万亿(24小时-2.16%),BTC现价81259.67低于SMA20(81833.95),RSI 47.6正常,MACD负值;ETH现价2357.22低于SMA20(2417.65)及SMA50(2534.04),RSI 39.1偏弱。中概股整体偏下行:BABA、PDD、JD、腾讯均处于空头排列,RSI在32-41区间,且腾讯接近52周低点。商品外汇分化:黄金期货RSI 38.7、价格低于各均线呈空头排列;WTI原油现价101.85明显高于SMA20(96.28)、RSI 59.0、MACD正柱,偏强;美元兑人民币RSI 9.9极度超卖;美元指数RSI 57.1接近52周高;VIX 16.91中性。整体风险偏好温和,技术面分化明显。

今日关注

USDCNY=X 美元 / 人民币
偏下行

美元兑人民币现价6.65,RSI 14仅9.9处于极端超卖区,价格低于SMA20(6.69)、SMA50(6.70)、SMA200(6.80)形成空头排列,MACD柱状线-0.0111且信号线-0.0086维持负值,近1日跌0.23%、5日跌0.34%,接近52周低点,技术面明确偏弱。

CL=F WTI 原油期货
偏上行

WTI原油现价101.85,高于SMA20(96.28)、SMA50(93.96)、SMA200(85.92)呈多头排列,RSI 14为59.0处于中性偏强区,MACD柱状线2.4771为正且大于信号线2.0582,近1日涨1.95%、5日涨4.84%,动能偏多。

META Meta
偏下行

Meta现价612.49,RSI 14仅30.7接近超卖区,价格远低于SMA20(701.83)和SMA50(667.03),MACD柱状线-16.1162大幅为负、信号线0.257转负,近5日跌12.19%,接近52周低点,空头动能显著。

SPY S&P 500 ETF
中性

SPY现价772.31,RSI 14为53.3处于中性区,价格略高于SMA20(770.90)和SMA50(767.63)但MACD柱状线-0.0063微负、信号线0.4624为正,近1日涨0.34%、5日跌0.10%,多空信号交织。

0700.HK 腾讯控股 (0700.HK)
偏下行

腾讯现价387.6,RSI 14为32.1接近超卖,价格低于SMA20(411.70)、SMA50(428.36)、SMA200(482.09)呈空头排列,MACD柱状线-12.9448且信号线-10.0293均为负,近5日跌1.57%,离52周低点仅2.54%,弱势明显。

全部资产

^VIX

VIX 恐慌指数

$16.91 -3.48%
5 日
+1.38%
距 52w 高
-52.1%
RSI(14)
50.2
趋势
中性
SMA 20 / 50 / 200
16.75 / 16.07 / 18.23
MACD / 信号
0.608 / 0.552

^TNX

10Y 美债收益率 (%)

$5.04 +1.04%
5 日
-1.96%
距 52w 高
-5.9%
RSI(14)
46.7
趋势
多头
SMA 20 / 50 / 200
5.15 / 5.00 / 4.54
MACD / 信号
-0.012 / 0.021
多头排列

DX-Y.NYB

美元指数 DXY

$102.08 -0.23%
5 日
+0.30%
距 52w 高
-0.8%
RSI(14)
57.1
趋势
多头
SMA 20 / 50 / 200
102.05 / 100.69 / 99.60
MACD / 信号
0.309 / 0.386
接近 52 周高多头排列

SPY

S&P 500 ETF

$772.31 +0.34%
5 日
-0.10%
距 52w 高
-1.2%
RSI(14)
53.3
趋势
多头
SMA 20 / 50 / 200
770.90 / 767.63 / 728.15
MACD / 信号
-0.006 / 0.462
接近 52 周高多头排列

QQQ

Nasdaq 100 ETF

$725.61 +0.60%
5 日
+0.06%
距 52w 高
-4.9%
RSI(14)
49.6
趋势
多头
SMA 20 / 50 / 200
732.48 / 725.42 / 678.72
MACD / 信号
-4.373 / -2.346
多头排列

AAPL

Apple

$343.83 -0.94%
5 日
-0.05%
距 52w 高
-2.2%
RSI(14)
56.1
趋势
多头
SMA 20 / 50 / 200
340.61 / 331.73 / 295.69
MACD / 信号
3.916 / 4.062
MACD 死叉 (今天)接近 52 周高多头排列

MSFT

Microsoft

$531.01 -0.06%
5 日
-0.36%
距 52w 高
-3.7%
RSI(14)
56.1
趋势
多头
SMA 20 / 50 / 200
529.90 / 511.18 / 437.32
MACD / 信号
6.264 / 8.043
多头排列

NVDA

Nvidia

$215.38 -1.17%
5 日
-2.97%
距 52w 高
-11.5%
RSI(14)
43.9
趋势
中性
SMA 20 / 50 / 200
223.79 / 222.07 / 204.14
MACD / 信号
-2.674 / -1.554

GOOGL

Alphabet

$324.89 +1.34%
5 日
-2.95%
距 52w 高
-20.5%
RSI(14)
38.7
趋势
中性
SMA 20 / 50 / 200
339.81 / 341.51 / 341.20
MACD / 信号
-6.085 / -3.225

TSLA

Tesla

$404.22 -0.80%
5 日
+4.59%
距 52w 高
-19.0%
RSI(14)
63.6
趋势
中性
SMA 20 / 50 / 200
376.78 / 367.47 / 385.72
MACD / 信号
7.705 / 4.467
MACD 金叉 (2 天前)

META

Meta

$612.49 -1.24%
5 日
-12.19%
距 52w 高
-21.5%
RSI(14)
30.7
趋势
中性
SMA 20 / 50 / 200
701.83 / 667.03 / 632.14
MACD / 信号
-16.116 / 0.257
加密恐慌贪婪
51
中性
加密总市值
$2.74 T
-2.16% / 24h
BTC 主导率
59.5%
ETH 10.5%
24h 成交量
$69.1 B
活跃币 22,565

BTC-USD

Bitcoin

$81,259.67 +0.56%
5 日
+0.23%
距 52w 高
-28.5%
RSI(14)
47.6
趋势
中性
SMA 20 / 50 / 200
81,833.95 / 81,888.06 / 73,066.69
MACD / 信号
-430.713 / -204.681

ETH-USD

Ethereum

$2,357.22 +0.40%
5 日
-1.13%
距 52w 高
-41.6%
RSI(14)
39.1
趋势
中性
SMA 20 / 50 / 200
2,417.65 / 2,534.04 / 2,158.83
MACD / 信号
-56.794 / -50.784

SOL-USD

Solana

$102.33 +1.06%
5 日
-0.44%
距 52w 高
-49.3%
RSI(14)
40.0
趋势
中性
SMA 20 / 50 / 200
104.80 / 109.40 / 88.62
MACD / 信号
-2.548 / -2.209

BABA

阿里巴巴 (BABA)

$100.96 +0.72%
5 日
-3.30%
距 52w 高
-44.7%
RSI(14)
34.1
趋势
空头
SMA 20 / 50 / 200
106.75 / 110.90 / 126.93
MACD / 信号
-2.841 / -2.077
空头排列

PDD

拼多多 (PDD)

$77.47 -1.19%
5 日
-1.59%
距 52w 高
-44.4%
RSI(14)
40.8
趋势
空头
SMA 20 / 50 / 200
79.40 / 80.73 / 91.37
MACD / 信号
-0.464 / -0.258
MACD 死叉 (2 天前)空头排列

JD

京东 (JD)

$25.45 -0.47%
5 日
-1.32%
距 52w 高
-27.4%
RSI(14)
36.3
趋势
空头
SMA 20 / 50 / 200
26.36 / 27.16 / 28.63
MACD / 信号
-0.482 / -0.392
MACD 死叉 (4 天前)空头排列

0700.HK

腾讯控股 (0700.HK)

HK$387.60 +0.62%
5 日
-1.57%
距 52w 高
-41.6%
RSI(14)
32.1
趋势
空头
SMA 20 / 50 / 200
411.70 / 428.36 / 482.09
MACD / 信号
-12.945 / -10.029
接近 52 周低空头排列

GC=F

黄金期货

$4,068.10 -0.16%
5 日
-1.97%
距 52w 高
-27.2%
RSI(14)
38.7
趋势
空头
SMA 20 / 50 / 200
4,137.58 / 4,318.86 / 4,525.92
MACD / 信号
-63.541 / -60.936
MACD 死叉 (4 天前)空头排列

CL=F

WTI 原油期货

$101.85 +1.95%
5 日
+4.84%
距 52w 高
-14.8%
RSI(14)
59.0
趋势
多头
SMA 20 / 50 / 200
96.28 / 93.96 / 85.92
MACD / 信号
2.477 / 2.058
多头排列

USDCNY=X

美元 / 人民币

¥6.65 -0.23%
5 日
-0.34%
距 52w 高
-7.5%
RSI(14)
9.9
趋势
空头
SMA 20 / 50 / 200
6.69 / 6.70 / 6.80
MACD / 信号
-0.011 / -0.009
RSI 超卖接近 52 周低空头排列
风险提示

本报告基于公开行情数据计算的技术指标,仅供技术指标解读参考,不构成任何投资建议。所有指标如RSI、MACD、均线等均存在滞后性,可能因市场环境变化而失效。过去走势不代表未来表现,任何买卖决策请结合自身风险承受能力独立判断。

Australian politics live: Labor says ATO could have handled credit card ban better; man cautioned for allegedly stalking MPs

Follow the day’s news live Get our breaking news email, free app or daily news podcast Gas reservation policy set for lower house vote today Australia’s domestic gas reservation is set to be debated and voted on in the lower house today. Rather than international base companies, determining through

中文摘要 澳大利亚工党称,税务局本可更妥善处理信用卡禁令问题;一名男子因涉嫌跟踪议员受到警告。国内天然气储备政策今日提交众议院投票表决。

Trump Officials Discussing Sanctions Relief for Russian Fund, Russia Acknowledges

A Russian official said potential relief for a state investment fund was part of talks. Russia has sought to wring economic benefits from the negotiations on ending the war with Ukraine.

中文摘要 特朗普政府官员与俄方讨论放松对一家俄罗斯国家投资基金的制裁,俄方官员证实该话题是谈判内容之一。俄方试图从乌克兰停战谈判中获取经济利益。

Surprise Showing by Arab Parties in Israeli Vote Puts Them in Spotlight

The Arab parties have long been kept on the political sidelines. But stunning turnout by their supporters makes the Arab minority an Israeli political reality that cannot be ignored.

中文摘要 以色列阿拉伯政党在选举中意外获得大量支持,支持者投票率惊人,使阿拉伯少数群体成为不可忽视的政治力量,这些政党长期以来被排除在政治边缘。

Flávio Bolsonaro denies fleeing Brazil after landing in Miami

The defeated Brazilian candidate claims to be the victim of a coup d’état and political persecution, recycling unsubstantiated allegations of “electoral fraud”.

中文摘要 巴西败选候选人弗拉维奥·博索纳罗否认逃离巴西,称自己是政变和政治迫害的受害者,并重复未经证实的“选举舞弊”指控。他目前已抵达迈阿密。

Menindee Lakes water levels drop to lowest in five years as experts warn of mass fish deaths

NSW government imposes water restrictions after Menindee water volumes fall below 2021 levels Sign up for climate and environment editor Adam Morton’s free Clear Air newsletter here Water levels at the Menindee Lakes in western New South Wales have fallen to their lowest point in five years, prompti

中文摘要 新南威尔士州西部梅宁迪湖水位降至五年来最低,低于2021年水平,引发大规模鱼类死亡担忧。新州政府已实施限水措施。

Flávio Bolsonaro Lands in Miami After Fleeing Brazil

Mr. Bolsonaro, who lost Sunday’s presidential election, fled Brazil for the United States via Argentina after the police canceled his passport in relation to an investigation.

中文摘要 巴西败选总统候选人弗拉维奥·博索纳罗在警方因调查取消其护照后,经阿根廷逃往美国,已抵达迈阿密。

Girl seriously injured in wolf attack at German wildlife park

The eight-year-old was in an enclosure with other children and a staff member when she was attacked.

中文摘要 德国一野生动物园内,一名8岁女孩与其他儿童及一名工作人员同在围栏内时遭狼攻击,受重伤。

Flávio Bolsonaro calls himself the ‘leader of Brazil’ after fleeing to Miami

Far-right senator and family land in US after crossing from Brazil into Argentina and then flying via Ecuador Flávio Bolsonaro declared himself the “true president” of Brazil after losing last Sunday’s presidential election and fleeing to the United States. After landing on a private jet with his fa

中文摘要 巴西极右翼参议员弗拉维奥·博索纳罗在周日选举失利后,经阿根廷和厄瓜多尔逃往美国,自称巴西“真正总统”,已抵达迈阿密。

Australian SAS officers knew about unlawful ‘throwdowns’ on dead Afghans, report says

Nothing done to stop planting weapons on dead Afghans, ADF inspector says, as ‘breakdown of command’ allowed ‘egregious behaviour’ Get our daily political email, free app or daily news podcast Some Australian SAS officers knew that soldiers under their command were using “throwdowns” – planting weap

中文摘要 澳大利亚国防军督察报告称,部分SAS军官知晓部下对阿富汗死者非法放置武器(“扔下”行为),但未阻止。报告指出“指挥系统崩溃”纵容了“恶劣行为”。

New book makes fiery claims about Jared Kushner’s finances in Trump’s first term

Kushner threatens to sue, denying allegations that two slush funds were set up to influence his Middle East role A new book containing incendiary claims about Jared Kushner’s financial dealings during Donald Trump’s first term has ignited a political and media firestorm. The book by author Matthew C

中文摘要 一本新书指控特朗普首任期内,贾里德·库什纳通过两个秘密基金获取财务利益,以影响其中东事务角色。库什纳否认并威胁起诉。

Brazilian political drama unfolds as defeated Flavio Bolsonaro arrives in Miami

Defeated presidential candidate Flavio Bolsonaro and his family have arrived in the U.S. after Brazilian authorities ordered him to surrender his passport. He is facing multiple investigations in Brazil.

中文摘要 巴西败选总统候选人弗拉维奥·博索纳罗及其家人抵达美国,此前巴西当局要求其交出护照。他在巴西面临多项调查。

High-speed train crash kills at least 68 people in Spain

The crash, between a moving train and another that was stopped further up the track, was caused by an "inexplicable" error, the transport minister said.

中文摘要 西班牙发生高速列车相撞事故,一列行驶中列车撞上前方停驶列车,造成至少68人死亡。交通部长称事故由“无法解释的错误”导致。

Costa Rica’s mega-prison receives first inmates

President Laura Fernandez inaugurated the $35m mega-prison built to house up to 5,100 inmates near San Jose.

中文摘要 哥斯达黎加总统劳拉·费尔南德斯启用位于圣何塞附近、耗资3500万美元的巨型监狱,可容纳5100名囚犯,已接收首批囚犯。

EU backs away from offer to Canada to start ‘associate member’ talks

Watered-down version of statement to be issued at end of Montreal summit instead offers ‘ever closer association’ The EU has backed away from its offer to Canada to start talks on becoming an “associate member” after disagreements between members states just as an EU-Canada summit was set to begin i

中文摘要 欧盟因成员国意见分歧,放弃向加拿大提议开始“准成员”谈判,改为在蒙特利尔峰会结束时发表措辞较弱的声明,仅承诺“更紧密联系”。

BYD Short Sellers Retreat Ahead of Expected Profit Rebound

Short sellers are paring back bets against BYD Co. ahead of earnings, with the Chinese electric vehicle giant poised to post a rebound in profit helped by accelerating overseas sales and a domestic market recovery.

中文摘要 空头在比亚迪公布财报前削减看跌押注。比亚迪预计将公布利润反弹,受益于海外销售加速和国内市场复苏。

Google shares tumble 10% after $51bn AI outlay deepens cash burn

Search giant admits its ‘AI mode’ has hurt traditional search as sales and profits undershoot expectations

中文摘要 谷歌母公司Alphabet股价下跌10%,因510亿美元AI投入加剧现金消耗,且AI模式对传统搜索业务造成冲击,销售和利润未达预期。

Microsoft blames data centre delays for lower capital expenditure

Software group reports overall revenue rose 19% to $92.3bn in quarter to the end of September

中文摘要 微软将资本支出下降归咎于数据中心延误。截至9月底的季度,公司总营收增长19%至923亿美元。

Petrobras Pumps at Record as Iran War Upends Oil Markets

Brazil’s state-controlled oil producer Petrobras lifted production to a record and ran its refineries near full capacity in the third quarter, helping the country weather disruptions to global energy markets from the Iran war.

中文摘要 巴西国家石油公司Petrobras三季度产量创纪录,炼厂接近满负荷运转,帮助该国应对伊朗战争引发的全球能源市场动荡。

Bankers in Failed Firmus IPO Probed by Australian Regulator ASIC

Australia’s corporate regulator is seeking information from investment banks on the failed Firmus Grid Ltd. initial public offering earlier this month, stopping short of opening a formal investigation into the matter.

中文摘要 澳大利亚监管机构ASIC正寻求投资银行关于本月早些时候Firmus Grid Ltd.首次公开募股失败的信息,但未正式立案调查。

US Stock Futures Fall After Alphabet, Fed Holds: Markets Wrap

US stock-index futures retreated in early Asian trading after Alphabet Inc.’s earnings disappointed, reviving concerns about the billions of dollars companies are pouring into artificial intelligence.

中文摘要 亚洲早盘美国股指期货下跌,因Alphabet财报不及预期,引发市场对企业巨额AI投资的担忧。美联储维持利率不变。

Oil Declines as Traders Parse Signs of Fresh Diplomacy on Iran

Oil dropped as traders assessed fresh diplomatic efforts in the Middle East to avert a potential restart of US military strikes on Iran.

中文摘要 油价下跌,交易商评估中东新一轮外交努力,以期避免美国对伊朗恢复军事打击。

FirstFT: Weil leadership tells partners firm should wind down

Also in today’s newsletter: Google shares tumble as cash burn deepens and dozens killed in Spain high-speed train crash

中文摘要 FirstFT简报:Weil律师事务所领导层告知合伙人应逐步关闭公司。另报道谷歌股价因现金消耗加剧下跌,西班牙高铁事故致数十人死亡。

Fed mantiene tasas sin dar muchas pistas sobre diciembre

La Reserva Federal mantuvo las tasas de interés sin cambios y reafirmó su compromiso de contener la inflación, pero evitó dar señales claras sobre un posible aumento en diciembre. La decisión llega a menos de una semana de las elecciones de medio término en EE.UU., con los votantes preocupados por e

中文摘要 美联储维持利率不变,重申遏制通胀承诺,但未就12月可能的加息给出明确信号。决策发布距美国中期选举不到一周。

Sprouts Sinks After Outlook Cut on Cyclospora, Consumer Woes

Sprouts Farmers Market Inc. shares tumbled after slashing its outlook, as the grocer’s sales suffered from the cyclospora lettuce outbreak and a gloomy consumer.

中文摘要 Sprouts Farmers Market股价大跌,因公司下调展望。生菜环孢子虫疫情和消费者悲观情绪拖累销售。

Fed Officials Left Interest Rates Unchanged

Jon Ferro, Lisa Abramowicz and Tom Keene host a special edition of "Bloomberg Surveillance" as the Federal Open Market Committee voted 10-2 to hold its benchmark federal funds rate in a range of 3.75% to 4% and leave interest rates unchanged. (Source: Bloomberg)

中文摘要 美联储公开市场委员会以10比2投票维持基准联邦基金利率在3.75%至4%区间不变。

该源今日无内容。