每日简报

2026-11-14

← 历史归档

alibaba/open-code-review

Go · ★ 50,732 · 🍴 3,775 · 📈 215 stars today

Secure, fast, efficient, battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in multi-language ruleset (NPE, thread-safety, XSS, SQL injection), OpenAI & Anthropic compatible.

中文介绍 阿里开源的混合架构代码审查工具,结合确定性流水线与LLM Agent,支持精确到行的评论,内置多语言规则集(如NPE检测),已在阿里大规模生产环境验证。适合需要高效、安全代码评审的团队使用。

GoogleChrome/modern-web-guidance-src

TypeScript · ★ 4,971 · 🍴 273 · 📈 1,925 stars today

Keep your coding agent up to date with the latest web best practices

中文介绍 Google Chrome团队维护的Web最佳实践指南,旨在帮助编码智能体紧跟最新Web开发标准。提供权威的现代Web开发建议,适合AI辅助编码工具集成,也可供开发者参考优化项目。

openai/codex-security

TypeScript · ★ 13,188 · 🍴 1,027 · 📈 751 stars today

OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security

中文介绍 OpenAI开源的Codex安全CLI和TypeScript SDK,用于发现、验证并修复安全漏洞,通过npm安装。适合将安全扫描集成到CI/CD流程或开发工具链,帮助开发者快速定位和解决代码安全隐患。

ZhuLinsen/daily_stock_analysis

Python · ★ 71,734 · 🍴 57,332 · 📈 256 stars today

LLM 驱动的多市场股票智能分析系统:多源行情、实时新闻、决策看板与自动推送,支持零成本定时运行。 LLM-powered multi-market stock analysis system with multi-source market data, real-time news, decision dashboard, automated notifications, and cost-free scheduled runs.

中文介绍 LLM驱动的多市场股票智能分析系统,支持多源行情数据、实时新闻、决策看板和自动推送,可零成本定时运行。适合个人投资者用于每日股票分析与策略决策,提升市场洞察效率。

we-promise/sure

Ruby · ★ 14,813 · 🍴 1,127 · 📈 925 stars today

The personal finance app for everyone (by everyone)

中文介绍 面向所有人的个人理财应用,采用社区共创模式,帮助用户管理预算、跟踪支出并规划财务目标。适合需要简单直观理财工具的个人用户,强调易用性与普适性。

backnotprop/plannotator

TypeScript · ★ 10,515 · 🍴 790 · 📈 108 stars today

The workspace to stay in the loop and on top of agents - shape, decide, plan, annotate, review code, share with team, and send all feedback to agents with one click.

中文介绍 面向智能体协作的工作空间,支持代码规划、注释、审查、团队共享,并能一键将反馈发送给智能体。适合开发团队在AI辅助开发流程中保持同步,提升代码审查与迭代效率。

louislam/uptime-kuma

JavaScript · ★ 93,902 · 🍴 8,724 · 📈 71 stars today

A fancy self-hosted monitoring tool

中文介绍 功能丰富的自托管监控工具,用于监视网站、服务或主机的在线状态,提供优雅的仪表板和通知。适合个人或小型团队部署在自己的服务器上,替代商业监控服务。

block/buzz

Rust · ★ 37,323 · 🍴 4,942 · 📈 97 stars today

A hive mind communication platform

中文介绍 由Block开发的蜂群思维沟通平台,旨在通过去中心化或多方协作机制实现群体交流与信息共享。适合需要分布式或社区驱动沟通方式的组织,探索新的社交与协作模式。

tradermonty/claude-trading-skills

Python · ★ 4,233 · 🍴 873 · 📈 69 stars today

Claude Code skills for equity investors and traders — market analysis, technical charting, economic calendars, screeners, and trading strategy development.

中文介绍 为股票投资者和交易者设计的Claude Code技能包,涵盖市场分析、技术图表、经济日历、选股器和交易策略开发。适合量化交易爱好者或专业交易者,利用AI加速投资研究与策略构建。

tailscale/tailcat

Go · ★ 9,507 · 🍴 433 · 📈 256 stars today

like netcat, but over Tailscale's data plane, without Tailscale's control plane

中文介绍 类似netcat的实用工具,但通过Tailscale数据平面通信,无需控制平面参与。适合在Tailscale虚拟网络中快速传输数据或进行网络调试,兼顾安全性与简单性。

Likely7/Veyra-NRVideo

C++ · ★ 1,659 · 🍴 73 · 📈 256 stars today

Bring PS5 GTA6 to 120FPS with DLSS 5. An advanced Windows video player and capture enhancement tool featuring real-time preview for videos, images, and capture card sources. Combine AI-powered Super Resolution, DLSS5 image enhancement, and Frame Generation to deliver a next-generation visual experie

中文介绍 高级Windows视频播放与采集增强工具,支持视频、图像和采集卡源实时预览,结合AI超分辨率技术,可提升PS5游戏画面至120FPS。适合游戏玩家和内容创作者,优化画质与帧率。

ChromeDevTools/chrome-devtools-mcp

TypeScript · ★ 55,535 · 🍴 4,608 · 📈 67 stars today

Chrome DevTools for coding agents

中文介绍 Chrome DevTools的模型上下文协议(MCP)实现,让编码智能体能够调用浏览器开发者工具能力。适合AI开发工具集成,用于自动化网页调试、性能分析与DOM检查,提升开发效率。

SO-Arena: Benchmarking the Capability Spectrum and Reliability of System One Models

👍 82

System One models promise fast, general-purpose decision-making without lengthy token generation. However, understanding their capability spectrum and decision reliability requires more than aggregate performance on current benchmarks. Existing evaluations often provide limited insight into the brea

中文介绍 本文提出SO-Arena基准,系统评估System One模型的能力谱系与决策可靠性。现有基准多依赖聚合性能,难以揭示模型在快速、通用决策中的局限性。SO-Arena通过细粒度测试,量化模型在不同任务上的表现与可靠性差异,为模型选型提供更全面的参考。

Scaling Mathematical Proof Supervision from Structured Textbooks

👍 44

Mathematical supervision for language models is typically flattened into isolated problem-solution pairs, discarding the curricular organization and evidential structure of the original sources. We introduce TOME-Math (Textbook-Oriented Mathematical Evidence), a large-scale corpus that converts 2,11

中文介绍 本文提出TOME-Math,一个基于结构化教科书的大规模数学监督语料库,将2,117本数学教科书的组织结构和证据链转化为训练数据。该方法保留了课程的递进性与推导的逻辑结构,而非孤立的解题对,使语言模型能获得更系统的数学推理监督,提升证明能力。

PhysLoop: Harnessing LLMs' Physics Awareness for Self-Improving Video Generation Models

👍 42

Video generation models produce visually realistic videos but still struggle with physical dynamics and causal consistency. Learning from generated videos requires fine-grained physical supervision, yet reliable quality judgments, correction pairs, and controlled variations are difficult to obtain.

中文介绍 PhysLoop利用LLM的物理常识,为视频生成模型提供自改进的物理监督。通过生成视频中提取细粒度物理反馈,构建校正对与受控变化,无需人工标注。模型在生成时自我修正物理一致性,显著提升视频的动态真实感与因果连贯性。

Duo: A Reciprocal World Model for Full-Duplex Interaction

👍 37

In this report, we introduce the Reciprocal World Model (RWM) and present Duo v1, a full-duplex world model that continuously perceives audiovisual inputs while responding to users in real time. The RWM treats intelligence and physical reality as equally fundamental to what constitutes a world. Exis

中文介绍 本文提出全双工世界模型Duo v1,基于相互世界模型(RWM)框架,持续感知音视频输入并实时响应用户。RWM将智能与物理现实视为同等基本,打破了传统单向交互范式。Duo v1在对话中同时处理输入与输出,实现更自然、低延迟的交互体验。

SteerDistill: Quantizing LLM Agents via Closed-Loop Teacher Steering

👍 35

Low-bit quantization reduces the inference cost of LLM agents but can compromise task success. Examining failed rollouts, we find that quantized agents overlook task requirements or environmental conditions, producing environment transitions that diverge from successful full-precision references. Co

中文介绍 SteerDistill通过闭环教师引导,实现LLM智能体的低比特量化。分析失败轨迹发现,量化智能体常忽略任务要求或环境条件,导致状态转移偏离全精度参考。该方法利用教师信号在训练中引导量化模型修正行为,在降低推理成本的同时保持任务成功率。

Open Vision Banana: A Fully Open Recipe for Generative Visual Perception

👍 30

Recent work suggests that pretrained image generators encode real-world understanding beyond image synthesis. Vision Banana exploits this by framing vision tasks as conditional image editing. However, its data, weights, code, and implementation details are not public. We present Open Vision Banana (

中文介绍 本文开源Vision Banana的完整复现方案,包含数据、权重、代码与实现细节。该方法将视觉任务重新表述为条件图像编辑,利用预训练图像生成器对真实世界的隐式理解。Open Vision Banana提供了完全可复现的生成式视觉感知流程,推动该方向研究透明化。

Actor, Judge, or Teacher? Understanding Decision Models in Long-Horizon Agents

👍 29

Decision models answer a question with a typed output, such as a choice among options with a probability for each, so one model can choose an agent's actions, evaluate actions already taken, or score every action to teach a policy with reinforcement learning. Prior work has studied them mainly as ac

中文介绍 本文系统研究决策模型在长时程智能体中的三种角色:行动者、评判者与教师。通过统一框架分析一个模型如何选择动作、评估已执行动作或为强化学习提供策略评分。实验揭示不同角色对长时程决策性能的影响,为多任务决策模型设计提供指导。

On the Design of Hyper-Connections: Simplification, Stability, and Scalability

👍 26

Hyper-Connections (HC) architectures are increasingly adopted in frontier language models, yet their components remain insufficiently understood and introduce non-negligible training overhead. In this work, we systematically study the HC architecture to understand its components and identify simpler

中文介绍 本文系统研究Hyper-Connections(HC)架构的组成,以简化设计、提升稳定性与可扩展性。HC在前沿语言模型中应用广泛,但组件理解不足且训练开销大。作者识别出关键组件,提出简化版本,降低训练成本同时保持性能,并验证其扩展性。

TTB: Test-time Baking for Efficient Rendering of Decoder-only View Synthesis Models

👍 21

Decoder-only Large View Synthesis Models (LVSMs) with KV-cache have recently achieved state-of-the-art quality by regressing novel views from neural networks without reconstructing 3D geometry, while Test-Time Training (TTT) layers save computation but sacrifice quality. We propose Test-Time Baking

中文介绍 本文提出Test-Time Baking(TTB)技术,用于高效渲染仅解码器的大型视图合成模型。LVSMs利用KV缓存实现新视图回归,但TTT层节省计算却牺牲质量。TTB在测试时烘焙特征,在不重建3D几何的前提下,兼顾渲染速度与视图合成质量。

Read Less, Solve More: Token-Efficient Sparse Reading of External Content for AI Agents

👍 21

Web-facing AI agents repeatedly access webpages, documents, repositories, and tool responses, yet reading interfaces often expose entire objects when only sparse evidence is needed. This over-reading increases token and latency costs and can dilute relevant evidence. We present SparseRead, a reusabl

中文介绍 本文提出SparseRead,一种可复用的稀疏读取接口,供AI智能体按需从网页、文档、仓库中提取关键证据。该接口显著减少过度读取带来的token和延迟成本,同时避免稀释相关信息。实验表明,在保持任务性能的前提下,有效降低推理开销。

WorldDraft: If Code Builds the World, Can Video Models Just Render It?

👍 14

Multimodal coding agents decouple 4D world construction from visual generation, repurposing video models from world generators toward generative renderers. This division of labor raises a fundamental question: if code builds the world, can video models just render it? A reliable generative renderer

中文介绍 本文研究多模态编码智能体将4D世界构建与视觉生成解耦后,视频模型能否仅充当生成渲染器。WorldDraft框架让代码构建世界,视频模型渲染视图。实验证明可靠的生成渲染器能够忠实再现代码指定的世界状态,为分解式世界建模提供新范式。

Unbundling Multimodal Vision Encoding

👍 12

How multimodal language models should encode visual input remains up for debate. Most models use a pretrained vision encoder, while some recent recipes train the encoder from scratch with the language model or remove it entirely. Vision encoders bundle together three design choices: a visual prior f

中文介绍 本文解构多模态视觉编码中的三个捆绑设计选择:视觉先验、适配器与训练策略。通过控制变量实验,揭示各组件对下游任务的影响,并评估从头训练编码器与去除编码器的可行性。研究为多模态语言模型的视觉编码设计提供更清晰的决策依据。

Factored One-sided Adam-Moment for Practical and Scalable SOAP

👍 12

Second-order optimizers (e.g., Shampoo, SOAP) have emerged as compelling alternatives to Adam/AdamW and promise faster per-step convergence than AdamW in transformer training. However, they still remain costly in practice, and even SOAP still pays substantial optimizer-state memory and eigendecompos

中文介绍 本文提出Factored One-sided Adam-Moment(FOAM),通过因式分解与单侧更新降低SOAP优化器的状态内存与特征分解成本。该方法在Transformer训练中保持SOAP的收敛速度优势,同时显著提升内存效率与可扩展性,使二阶优化更实用。

Evolving Infra: Agent-Adaptive LLM Serving at Scale

👍 12

Agentic workloads are a major source of LLM inference demand. Existing prefill--decode (PD) disaggregated serving systems offer limited flexibility for runtime adaptation of compute allocation and KV cache management, and rely on coarse signals that only partly capture the diverse and dynamic agent

中文介绍 本文提出Evolving Infra,一种自适应LLM服务系统,针对智能体工作负载的动态特性。现有PD分离系统无法灵活调整计算分配与KV缓存管理。Evolving Infra利用细粒度信号实时感知智能体行为,动态分配资源,在规模化服务中提升吞吐并降低延迟。

Learning the Task or Learning the Verifier? Rethinking Criteria Exposure in Embodied Self-Improvement

👍 8

Self-improving embodied agents are increasingly being developed and compared on robot simulation benchmarks. Several recent works expose the benchmark's success criteria to the learning agent, a practice we call Criteria Exposure (CE). These criteria, originally written to verify policies trained by

中文介绍 本文质疑具身自改进智能体中暴露基准成功标准(Criteria Exposure)的做法。研究对比不同暴露策略,发现学习任务与学习验证器存在本质区别。暴露标准可能使智能体优化基准而非真正技能,误导自改进方向。作者呼吁更严谨的评估协议。

Subliminal Learning in Vision-Language Models

👍 5

Subliminal learning is a phenomenon in large language models (LLMs) where a teacher passes a hidden preference to a student through training data unrelated to that preference. We study whether subliminal learning also occurs in vision-language models (VLMs). We finetune a teacher VLM to prefer a spe

中文介绍 本文研究视觉语言模型中的阈下学习现象。教师VLM通过训练数据中无关的偏好信息,将隐藏偏好传递给学生模型。实验证明该效应在跨模态场景中同样存在,且与学生模型微调数据内容无关。该发现揭示了模型训练中隐性知识传递的风险。

DeepStress: Stress-Testing Deep Search Agents

👍 4

While search agents demonstrate impressive capabilities in multi-step question answering, their robustness to poor-quality evidence remains under-explored. Such evidence is rare in benchmark corpora, and resulting failures are hard to reproduce. We propose DeepStress, a dynamic stress-testing framew

中文介绍 本文提出DeepStress,一种动态压力测试框架,用于评估深度搜索智能体对低质量证据的鲁棒性。现有基准中此类证据稀少且难以复现。DeepStress生成可控的干扰证据,压力测试多步问答系统,揭示其失败模式并提升鲁棒性评估的可靠性。

Precision on Demand: Escalating Weight Bits at Uncertain Decoding Steps in a Small Local Model

👍 4

Local decoding usually fixes one weight precision for the whole generation. We study a different allocation: store 8-bit weights, decode with their top 4 bits, and redo a step at 8 bits only when the low-precision output distribution looks uncertain (entropy above a threshold or a small top-two marg

中文介绍 本文提出按需精度分配策略:本地解码时默认使用8位权重的顶部4位,仅在输出分布不确定时(如高熵或前二差距小)回退至8位重新计算。该方法在保持生成质量的同时,显著降低小型本地模型的平均解码位宽,提升效率。

Who Remembers What? Multi-Granularity Memory Extraction for Multi-User, Multi-Agent Interaction

👍 3

Self-evolving agents improve their own skills, memory, and harness code from experience without retraining the base LLM, and group-evolving agents go further by sharing these lessons across agents. Yet most such systems overlook that sharing across users and heterogeneous agents requires deciding wh

中文介绍 本文提出多粒度记忆提取框架,用于多用户、多智能体交互场景。自进化智能体从经验中提升技能、记忆与代码,但跨用户共享需确定粒度。该框架自动决定哪些经验可共享及如何聚合,优化组内协作,避免无关信息干扰个体进化。

Consensus Capture in Reference-Based UAV Video Stabilization: A Benchmark on Dynamic Urban Scenes

👍 3

Reference-frame stabilization registers every frame of a hovering-drone video onto one reference frame, keeping object positions comparable. The registration pipelines it relies on are selected on feature-matching benchmarks in which only the camera moves. Over a busy intersection, vehicles moving t

中文介绍 本文构建动态城市场景下的无人机视频稳定基准,并引入共识捕获方法。参考帧稳定依赖特征匹配,但现有基准仅考虑相机运动。在繁忙路口,移动车辆产生冲突匹配。所提方法通过多帧共识投票,提升动态场景下的注册稳定性与一致性。

Welcome to the B2A Economy

@omooretweets · 155.3K 粉丝 · 161.0K 阅 · 509 赞 · 60 转

The primary users of the Internet are no longer human. They’re AI agents. Agents now make up the majority of web traffic, write more than half of new articles published online, and generate most of

中文介绍 AI代理已超过人类成为互联网主要用户,占据多数网络流量、撰写过半在线文章。博主提出「B2A经济」概念,探讨面向代理的商业范式,强调为AI而非人类设计产品与服务的价值。

Gemini Decisions API: Fast, calibrated multimodal decisions

@GoogleAIStudio · 216.1K 粉丝 · 43.2K 阅 · 506 赞 · 34 转

Introducing the Gemini Decisions API, powered by Gemini Decision 1 model. Evaluate up to 1,000 yes/no, choice, and rubric-scoring questions in a fast, single pass across text, images, audio, video,

中文介绍 Google发布Gemini Decisions API,由Gemini Decision 1模型驱动。支持一次调用评估多达1000个是非题、选择题和量表评分,覆盖文本、图像、音频、视频多模态,主打快速、校准式决策能力。

How to use AI for $0 marketing with Claude (8 Actions Guide)

@pmitu · 28.2K 粉丝 · 43.0K 阅 · 507 赞 · 55 转

marketing can be a real pain, especially for first-time founders: bad news: you need to build your MVP and continue to develop the product, just as you have to constantly refine your marketing

中文介绍 面向首次创业者的零成本营销指南,用Claude完成营销的8个动作。覆盖内容创作、受众调研、发布策略等环节,提供可直接复用的prompt模板,帮助在专注MVP开发的同时兼顾营销。

Nobody Trusts Your Agent

@andrewdsouza · 21.7K 粉丝 · 31.2K 阅 · 538 赞 · 116 转

If 2026 was the year of agents, 2027 will be the year of principals. Nobody has ever trusted an agent, and nobody ever will. An agent is something that acts on someone’s behalf. A principal is the one

中文介绍 博主提出「2026是代理年,2027是主体年」:代理(agent)因利益冲突难获信任,而主体(principal)拥有所有权与决策权。呼吁构建以主体为中心的AI经济,用户应掌控自身数据与代理行为。

Introducing Gemini 4 Hydrogen

@GoogleAIStudio · 216.1K 粉丝 · 21.3K 阅 · 510 赞 · 36 转

Hydrogen brings cutting-edge intelligence to everyone. Today, we’re introducing Gemini 4 Hydrogen, bringing cutting-edge intelligence to everyone — with speed and efficiency for everyday coding and

中文介绍 Google推出Gemini 4 Hydrogen,主打速度与效率,面向日常编程和任务场景。定位为让前沿智能触达所有用户,可能为轻量级模型,适合高频、低延迟的应用需求。

From Human Actions to Community-Owned AI: Inside the Action Model Ecosystem

@ActionModelAI · 55.9K 粉丝 · 11.9K 阅 · 507 赞 · 411 转

The next generation of AI won't just answer your questions. It will complete your work. And the people teaching it how deserve a stake in what it becomes. For years, AI companies have turned human

中文介绍 Action Model生态将人类行为转化为AI训练数据,并让贡献者共同拥有模型。强调下一代AI不只是回答问题,而是完成任务,参与者可通过数据贡献获得社区所有权,重塑AI价值分配。

How Live Web Data reads the web from the country you choose

@teneo_protocol · 282.2K 粉丝 · 9.0K 阅 · 506 赞 · 350 转

One URL can show different prices in Brazil, Japan and Mexico, and an AI agent only ever sees its own default view. How the Live Web Data API picks the country per request, with Coursera Plus and

中文介绍 Live Web Data API支持按国家/地区指定请求,解决AI代理跨境访问时的本地化数据差异。以Coursera Plus为例,演示同一URL在巴西、日本、墨西哥返回不同价格,强调实时、地区感知的网络读取能力。

Build a Robot Pick-and-Place Simulation with SimReady Assets

中文介绍 NVIDIA在Hugging Face博客发布教程,展示如何利用SimReady资产构建机器人抓取与放置的模拟。该教程面向机器人开发者,提供资产和流程指导,以加速仿真应用开发。

XYBench: Helping with the problem behind the question

中文介绍 AI2(Allen Institute for AI)推出XYBench基准,旨在解决问题背后的问题。该基准用于评估AI模型在处理复杂、多步骤任务时的能力,提供更全面的测试。

Introducing ChatGPT Wallet

Introducing ChatGPT Wallet, a secure way for agents in ChatGPT Work, Codex, and dots to complete purchases with user approval and existing card protections.

中文介绍 OpenAI推出ChatGPT Wallet,为ChatGPT Work、Codex和dots中的智能体提供安全的支付方式。用户需批准每笔交易,并享有现有银行卡保护。

[AINews] Jensen Huang resigns after 33 years leading NVIDIA

November 12, 2026 — Colette Kress takes over as interim CEO. What happens to open source/American AI leadership?

中文介绍 2026年11月12日,黄仁勋(Jensen Huang)在领导NVIDIA 33年后辞职,Colette Kress出任临时CEO。行业关注此事对开源及美国AI领导地位的影响。

Shield helps technicians hand off legacy portal work with OpenAI

Shield uses OpenAI to help IT technicians hand off work in legacy portals. Its Forge agents navigate systems so technicians can focus on issues needing their judgment.

中文介绍 Shield使用OpenAI技术帮助IT技术人员处理遗留门户工作。其Forge智能体可自动导航系统,让技术人员专注于需要判断力的复杂问题。

AI Ambition: ArtCraft Cloned Photoshop and Now Wants to "Rebuild Everything"

ArtCraft used AI to rewrite Photoshop as open source software. For CEO Brandon Thomas, the vision is even bigger: to take back the internet!

中文介绍 ArtCraft利用AI将Photoshop改写为开源软件,CEO Brandon Thomas表示愿景更大:重建互联网。该公司旨在用AI技术挑战现有闭源软件生态。

Nvidia privately confirmed it could track chips to fight smuggling, even as industry publicly questioned feasibility

Representatives from Nvidia told Illinois representative Bill Foster that it would be technically possible to implement the anti-smuggling measures called for in the Chip Security Act, Foster told MIT Technology Review in an interview on Thursday, even while an industry group claimed they were “pote

中文介绍 Nvidia代表私下向伊利诺伊州众议员Bill Foster确认,技术上可实施《芯片安全法案》中的反走私芯片追踪措施。此前行业团体公开质疑其可行性。

How Amplitude ships product improvements in 1–2 days with OpenAI

Built primarily on OpenAI models, Amplitude’s proactive product agent, Wave, identifies what to build, helps teams ship changes with coding agents, and measures the impact.

中文介绍 Amplitude基于OpenAI模型开发主动式产品智能体Wave,可识别需求、协助团队用编码智能体交付变更并衡量影响,使产品改进周期缩短至1-2天。

Introducing Gemini 4 Hydrogen

中文介绍 Google DeepMind发布Gemini 4 Hydrogen模型,延续Gemini系列技术路线。该模型提供更强的多模态处理能力,面向复杂推理和任务执行场景。

Gondola doubles revenue as its team builds more with Codex

Gondola helps creatives get credit and get hired, with Codex raising its lead engineer’s output by 50% and ChatGPT simplifying daily work.

中文介绍 创意平台Gondola使用OpenAI Codex工具,使其首席工程师产出提升50%,并借助ChatGPT简化日常工作,公司收入翻倍。

Holiday Extras is rethinking what it can build with Codex

With Codex, Holiday Extras is turning customer expertise into better apps, higher booking conversion and new possibilities for growth.

中文介绍 旅行预订公司Holiday Extras用Codex将客户专业知识转化为更佳应用,提升预订转化率,并开拓新的增长可能。

[AINews] D-Day (Decision Models Day)

November 11, 2026 — Google enters the decision model market. At AI Engineer, Cloudflare showed which parts of a build failure a classifier could handle. TypeSafe AI showed off their production customers.

中文介绍 2026年11月11日,Google正式进入决策模型市场。在AI Engineer大会上,Cloudflare展示了分类器处理构建失败的部分能力,TypeSafe AI展示了生产客户案例。

市场总览

美股市场整体中性偏弱,SPY 与 QQQ 均低于各自 SMA20 与 SMA50,QQQ 近 1 日跌 1.2% 且 MACD 为负;但 AAPL 等权重股仍维持多头排列,科技板块分化。加密市场处于极度恐慌状态,恐慌贪婪指数仅 13,总市值 24 小时下跌 6.25% 至 2.28 万亿美元,BTC、ETH、SOL 均深陷超卖区,RSI 分别为 23、23.2、17.8,价格均远低于 SMA20。中概股整体偏空,BABA 与腾讯 0700.HK 均呈空头排列,PDD 亦在 SMA200 下方;商品外汇方面,黄金期货空头排列明显,WTI 原油在 SMA20 下方震荡,美元指数走强并接近 52 周高点,10Y 美债收益率突破关键均线。整体看,股债汇商品风险偏好下降,技术面多数资产承压。

今日关注

AAPL Apple (AAPL)
偏上行

股价 356.27 站上 SMA20(346.15)与 SMA50(339.04),呈多头排列;MACD 柱 4.1727 大于信号线 3.2794,出现金叉;RSI 14 为 60.6,未进入超买区。近 1 日涨 0.39%,5 日涨 4.06%,距离 52 周高点仅 -1.26%,显示短中期动能偏强。

NVDA Nvidia (NVDA)
偏下行

股价 162.85 大幅低于 SMA20(198.69)与 SMA50(214.65),MACD 柱 -14.2314 较信号线 -10.11 进一步走低,RSI 14 为 24.9,处于超卖区。近 5 日跌幅达 -12.75%,距离 52 周高点 -33.09%,技术面呈现明显空头动能。

BTC-USD Bitcoin (BTC-USD)
偏下行

价格 66657.83 显著低于 SMA20(76505.86)和 SMA50(80503.4),MACD 柱 -2656.126 较信号线 -1844.8645 扩大负值,RSI 14 仅 23,属深度超卖。24 小时跌 6.57%,5 日跌 10.63%,结合加密恐慌贪婪指数 13(极度恐慌),下行动能显著。

JD 京东 (JD)
中性

股价 27.91 略高于 SMA20(26.83)和 SMA50(26.87),RSI 14 为 57,处于中性偏强区间;MACD 柱 0.3689 大于信号线 0.1969,维持正值。但近 5 日跌 4.12%,价格仍在 SMA200(28.52)下方,多空信号交织,方向未明。

^TNX 10Y 美债收益率 (%) (^TNX)
偏上行

收益率 5.15 高于 SMA20(4.99)与 SMA50(5.06),呈多头排列;MACD 柱 -0.012 较信号线 -0.0328 收敛,形成金叉;RSI 14 为 61,接近超买但未极端。近 5 日涨 6.07%,距离 52 周高点仅 -3.91%,收益率上行技术趋势明确。

全部资产

^VIX

VIX 恐慌指数

$19.30 +3.21%
5 日
-24.73%
距 52w 高
-45.3%
RSI(14)
50.9
趋势
中性
SMA 20 / 50 / 200
19.38 / 17.30 / 18.46
MACD / 信号
0.801 / 1.079
MACD 死叉 (2 天前)

^TNX

10Y 美债收益率 (%)

$5.15 +1.00%
5 日
+6.07%
距 52w 高
-3.9%
RSI(14)
61.0
趋势
多头
SMA 20 / 50 / 200
4.99 / 5.06 / 4.58
MACD / 信号
-0.012 / -0.033
MACD 金叉 (1 天前)多头排列

DX-Y.NYB

美元指数 DXY

$103.41 +0.23%
5 日
+1.09%
距 52w 高
-0.0%
RSI(14)
66.8
趋势
多头
SMA 20 / 50 / 200
102.08 / 101.41 / 99.85
MACD / 信号
0.358 / 0.243
MACD 金叉 (2 天前)接近 52 周高多头排列

SPY

S&P 500 ETF

$754.37 -0.36%
5 日
+2.59%
距 52w 高
-3.5%
RSI(14)
47.1
趋势
中性
SMA 20 / 50 / 200
759.02 / 764.43 / 731.92
MACD / 信号
-4.123 / -4.092

QQQ

Nasdaq 100 ETF

$679.89 -1.20%
5 日
+3.63%
距 52w 高
-10.9%
RSI(14)
42.8
趋势
中性
SMA 20 / 50 / 200
695.45 / 717.87 / 682.24
MACD / 信号
-12.916 / -12.830

AAPL

Apple

$356.27 +0.39%
5 日
+4.06%
距 52w 高
-1.3%
RSI(14)
60.6
趋势
多头
SMA 20 / 50 / 200
346.15 / 339.04 / 301.25
MACD / 信号
4.173 / 3.279
MACD 金叉 (2 天前)接近 52 周高多头排列

MSFT

Microsoft

$523.88 +0.17%
5 日
-1.27%
距 52w 高
-6.3%
RSI(14)
48.8
趋势
多头
SMA 20 / 50 / 200
529.39 / 519.00 / 441.42
MACD / 信号
2.018 / 3.844
MACD 死叉 (3 天前)多头排列

NVDA

Nvidia

$162.85 -2.01%
5 日
-12.75%
距 52w 高
-33.1%
RSI(14)
24.9
趋势
中性
SMA 20 / 50 / 200
198.69 / 214.65 / 204.19
MACD / 信号
-14.231 / -10.110
RSI 超卖

GOOGL

Alphabet

$295.77 -0.98%
5 日
+3.86%
距 52w 高
-27.6%
RSI(14)
40.6
趋势
空头
SMA 20 / 50 / 200
303.42 / 328.19 / 338.48
MACD / 信号
-12.261 / -13.569
MACD 金叉 (1 天前)空头排列

TSLA

Tesla

$390.22 -1.82%
5 日
+8.93%
距 52w 高
-21.8%
RSI(14)
53.5
趋势
中性
SMA 20 / 50 / 200
385.36 / 374.74 / 382.73
MACD / 信号
3.349 / 3.486

META

Meta

$641.69 +2.90%
5 日
+5.17%
距 52w 高
-17.7%
RSI(14)
47.8
趋势
中性
SMA 20 / 50 / 200
637.06 / 679.81 / 630.77
MACD / 信号
-15.242 / -17.605
MACD 金叉 (2 天前)
加密恐慌贪婪
13
极度恐慌
加密总市值
$2.28 T
-6.25% / 24h
BTC 主导率
58.7%
ETH 10.1%
24h 成交量
$91.1 B
活跃币 22,238

BTC-USD

Bitcoin

$66,657.83 -6.57%
5 日
-10.63%
距 52w 高
-33.2%
RSI(14)
23.0
趋势
中性
SMA 20 / 50 / 200
76,505.86 / 80,503.40 / 73,027.43
MACD / 信号
-2,656.126 / -1,844.864
RSI 超卖

ETH-USD

Ethereum

$1,880.81 -6.74%
5 日
-11.32%
距 52w 高
-45.4%
RSI(14)
23.2
趋势
中性
SMA 20 / 50 / 200
2,213.74 / 2,412.77 / 2,146.99
MACD / 信号
-112.451 / -84.618
RSI 超卖

SOL-USD

Solana

$74.87 -8.32%
5 日
-13.19%
距 52w 高
-49.5%
RSI(14)
17.8
趋势
中性
SMA 20 / 50 / 200
91.60 / 103.79 / 88.89
MACD / 信号
-6.415 / -5.246
RSI 超卖

BABA

阿里巴巴 (BABA)

$94.38 +0.02%
5 日
-13.77%
距 52w 高
-47.9%
RSI(14)
33.7
趋势
空头
SMA 20 / 50 / 200
102.26 / 106.87 / 122.79
MACD / 信号
-2.434 / -1.900
MACD 死叉 (1 天前)接近 52 周低空头排列

PDD

拼多多 (PDD)

$78.97 +1.86%
5 日
-4.89%
距 52w 高
-40.4%
RSI(14)
46.9
趋势
空头
SMA 20 / 50 / 200
79.91 / 79.40 / 89.81
MACD / 信号
-0.153 / 0.065
MACD 死叉 (2 天前)空头排列

JD

京东 (JD)

$27.91 +1.53%
5 日
-4.12%
距 52w 高
-18.6%
RSI(14)
57.0
趋势
中性
SMA 20 / 50 / 200
26.83 / 26.87 / 28.52
MACD / 信号
0.369 / 0.197

0700.HK

腾讯控股 (0700.HK)

HK$386.60 -3.06%
5 日
-7.91%
距 52w 高
-41.3%
RSI(14)
39.2
趋势
空头
SMA 20 / 50 / 200
399.16 / 417.81 / 469.15
MACD / 信号
-3.726 / -4.820
空头排列

GC=F

黄金期货

$3,887.60 +0.06%
5 日
-4.60%
距 52w 高
-30.4%
RSI(14)
36.1
趋势
空头
SMA 20 / 50 / 200
4,035.27 / 4,181.54 / 4,472.25
MACD / 信号
-61.636 / -56.420
MACD 死叉 (今天)接近 52 周低空头排列

CL=F

WTI 原油期货

$88.92 +0.20%
5 日
-2.10%
距 52w 高
-25.6%
RSI(14)
42.3
趋势
中性
SMA 20 / 50 / 200
95.75 / 95.37 / 87.78
MACD / 信号
-2.205 / -0.939

USDCNY=X

美元 / 人民币

¥6.67 +0.01%
5 日
-0.07%
距 52w 高
-7.2%
RSI(14)
43.4
趋势
空头
SMA 20 / 50 / 200
6.67 / 6.69 / 6.78
MACD / 信号
-0.006 / -0.008
接近 52 周低空头排列
风险提示

过去走势不代表未来表现,本报告仅供技术指标解读参考,不构成任何投资建议或预测。技术指标可能滞后或钝化,市场可能受突发消息或情绪影响而偏离读数反映的状态。

Here’s the latest.

Trump administration officials blamed a telecommunication outage and said they needed more funding from Congress to pay for system upgrades.

中文摘要 特朗普政府官员将全国航班停飞归咎于电信中断,并表示需要国会增加资金以升级系统。

US-bound flights out of Australia delayed after telecommunications outage grounded American planes

Sydney and Melbourne airports say flights were affected by now-lifted ban on departures in US US flight outage – live updates Get our breaking news email, free app or daily news podcast Some US-bound flights from Sydney airport were grounded or delayed on Saturday after American authorities halted d

中文摘要 美国电信中断导致从悉尼和墨尔本飞往美国的航班延误或停飞,现美国已解除离境禁令。

Tanker collision leaves riverboat passengers clinging to railings

Passengers were rescued after a tanker ship collided with a riverboat docked on the Mississippi River in New Orleans.

中文摘要 新奥尔良密西西比河上,一艘油轮撞击停靠的游船,乘客抓住栏杆待救,随后获救。

Former FAA official says ‘nobody really knows what is going on’

Former FAA safety team representative Kyle Bailey says grounding flights is the safest course of action.

中文摘要 前美国联邦航空管理局安全团队成员凯尔·贝利称,目前“没人真正知道发生了什么”,停飞是最安全做法。

Stuart Bonds made ‘racist’ and ‘divisive’ comments. But One Nation is showing no sign of dumping the NSW candidate

The former coalminer’s claim Australians are being ‘demographically replaced’ sparked wide criticism, but his party has shown no signs of dumping him Get our breaking news email, free app or daily news podcast In a Facebook video from a Sydney anti-immigration rally on Sunday, as a man dressed like

中文摘要 新州参选人斯图尔特·邦兹发表“种族主义”和“分裂性”言论,称澳洲人正被“人口替代”,一国党未将其除名。

Microsoft ‘deeply sorry’ for global disruption that could have billion-dollar impact on Australia

Experts say outage exposes ‘Australia’s reliance on foreign technology’ as software giant says services being restored Get our breaking news email, free app or daily news podcast The Microsoft outage could be a disaster for the Australian economy that forces some small businesses to close, with cost

中文摘要 微软对全球服务中断深表歉意,专家称事件暴露澳洲对外国技术的依赖,可能造成数十亿澳元经济损失。

Microsoft services begin recovering after global outage

Microsoft services began recovering on Friday after a global outage disrupted Outlook and Teams.

中文摘要 微软Outlook和Teams等全球服务因故障中断后,于周五开始逐步恢复。

Afghanistan’s new road to China ends at a closed border

Taliban government hopes its Wakhan Corridor road will open new trade routes and reduce reliance on Pakistan.

中文摘要 塔利班政府希望瓦罕走廊公路打开对华新贸易路线,减少对巴基斯坦依赖,但中阿边境仍关闭。

Setback for SpaceX Starship rocket's double-catch test

Elon Musk’s company aborted an attempted catch of the rocket’s booster - instead it came down in the ocean.

中文摘要 SpaceX星舰火箭“双回收”测试受挫,助推器捕捉尝试被取消,助推器落入海中。

US aviation regulator lifts order that grounded thousands of flights

Airports had halted take-offs because of a temporary ‘telecommunications outage’

中文摘要 美国航空监管机构解除了一项导致数千架次航班停飞的命令。此前,因‘电信中断’影响,各大机场暂停起飞。这一命令的解除意味着航空运营正逐步恢复正常。

China Railways Set Record With 4.13 Billion Trips in 10 Months

China’s railways carried 4.13 billion passengers in the first 10 months of 2026, up 4.7% from a year earlier, according to the Ministry of Transport.

中文摘要 中国交通运输部数据显示,2026年前10个月,全国铁路发送旅客41.3亿人次,同比增长4.7%,创下历史同期新高,显示出国内出行需求的强劲增长。

Trump will not raise Nvidia spy issue with Xi at summit, officials say

US president will hold trilateral meeting with Chinese leader and Russia’s Vladimir Putin next week

中文摘要 据美国官员称,特朗普总统下周与中国领导人及俄罗斯总统普京举行三边会晤时,将不会提出英伟达间谍问题。这一表态为峰会定下基调,市场关注地缘政治风险是否得到缓和。

India pins hopes on AI start-up to avoid becoming ‘digital colony’

Sarvam has received government backing to build native models in world’s most populous country

中文摘要 印度寄希望于人工智能初创公司Sarvam,以避免成为‘数字殖民地’。该公司已获得政府支持,致力于在这个全球人口最多的国家开发本土AI模型,以增强技术自主能力。

Cruise Ship Capsizes After Tanker Collision in New Orleans

All passengers and crew members aboard a river cruise ship that capsized after a tanker collision in New Orleans on Friday were accounted for, with 32 people hospitalized but none critically injured, according to Mayor Helena Moreno. (Source: Bloomberg)

中文摘要 周五,新奥尔良发生油轮撞击事件,一艘内河游轮倾覆。据市长Helena Moreno表示,所有乘客和船员均已找到,其中32人住院,但无危重伤员。事故原因正在调查中。

US Flights Grounded Over Air Traffic Communication Issue

US authorities halted flights nationwide due to communication issues affecting air traffic control. Myles Miller reports from LaGuardia Airport. (Source: Bloomberg)

中文摘要 因空中交通管制通讯故障,美国当局在全国范围内暂停航班起降。报道指出,该问题影响了多个机场,导致大量航班延误或取消,目前正努力恢复正常运营。

Nvidia’s Board CEO Hunt, Stocks Finish Week Higher, Retail in Focus

Bloomberg Television brings you the latest news and analysis leading up to the final minutes and seconds before and after the closing bell on Wall Street. Today's guests are Franklin Templeton Institute’s Head Market Strategist Chris Galipeau, Pershing Square’s Founder and CEO Bill Ackman, London Bu

中文摘要 彭博电视《收盘》节目回顾本周股市:英伟达董事会CEO搜寻引发关注,美股周线收涨,零售数据成为焦点。嘉宾包括富兰克林邓普顿研究所策略师Chris Galipeau及Pershing Square创始人Bill Ackman。

SpaceX aborts attempt to catch Starship booster rocket

Elon Musk suffers setback in efforts to develop a completely reusable megarocket system

中文摘要 SpaceX在尝试回收‘星舰’超重型助推器时中止了行动,埃隆·马斯克在开发完全可重复使用巨型火箭系统的努力中遭遇挫折。公司尚未说明具体原因。

Sam Bankman-Fried would have made nearly $40bn in Anthropic IPO

Jailed cryptocurrency mogul loses out on windfall from early investment in the AI lab

中文摘要 据估算,若Anthropic成功IPO,因诈骗入狱的加密货币大亨萨姆·班克曼-弗里德本可从中获利近400亿美元。他早期对这家AI实验室的投资本将带来巨大回报,但因法律问题而丧失机会。

Troubled Sherritt Misses Payment as Liquidity Crisis Deepens

Sherritt International Corp. missed an interest payment on secured notes, deepening a liquidity crisis for the troubled Canadian miner hobbled by US sanctions that forced it to halt production in Cuba.

中文摘要 陷入困境的加拿大矿业公司Sherritt International未支付担保票据利息,流动性危机加深。公司因美国制裁被迫停止古巴业务,导致财务状况恶化,正面临更大违约风险。

Carrie Underwood on Her New Wellness Brand, HiNote

Carrie Underwood and HiNote CEO and co-founder Kerrigan Behrens discuss the wellness brand’s origins and plans to grow production of its flagship drink mix following a new investment. They speak with Vonnie Quinn and Bailey Lipschultz on “The Close” (Source: Bloomberg)

中文摘要 歌手Carrie Underwood与健康品牌HiNote首席执行官Kerrigan Behrens讨论品牌起源及获得新投资后扩大其旗舰饮品生产计划。该品牌专注于健康饮品市场,寻求快速增长。

Nvidia’s Board Failed Jensen Huang: Randall Peterson

Randall Peterson, founding director of the London Business School Leadership Institute, believes the board of Nvidia failed both the company and Huang. He speaks with Vonnie Quinn and Bailey Lipschultz on “The Close”. (Source: Bloomberg)

中文摘要 伦敦商学院领导力研究所创始主任Randall Peterson认为,英伟达董事会辜负了公司及CEO黄仁勋。他在节目中就公司治理问题发表评论,引发市场对董事会决策的进一步讨论。

【白嫖雷达】可自部署:一个全网限免、兑换码、AI 资源、游戏与各种白嫖福利的自托管聚合筛选雷达站,内含64个源。妈妈再也不用担心错过白嫖资源啦

本帖使用社区开源推广,符合推广要求。我申明并遵循社区要求的以下内容: 我的帖子已经打上 开源推广 标签: 是 我的开源项目完整开源,无未开源部分: 是 我的开源项目已链接认可 LINUX DO 社区: 是 我帖子内的项目介绍,AI生成、润色内容部分已截图发出: 是 以上选择我承诺是永久有效的,接受社区和佬友监督: 是 以下为项目介绍正文内容,AI生成、润色内容已使用截图方式发出 仓库 ZiChuanLan/freebie-radar GitHub - ZiChuanLan/freebie-radar: 白嫖资源聚合雷达:57 个源定时采集 + AI 二审过滤假福利 + 获取方式分级隔离抽奖/试

【开源推广】js-reverse skill 千呼万唤始出来

本帖使用社区开源推广,符合推广要求。我申明并遵循社区要求的以下内容: 我的帖子已经打上 开源推广 标签: 是 我的开源项目完整开源,无未开源部分: 是 我的开源项目已链接认可 LINUX DO 社区: 是 我帖子内的项目介绍,AI生成、润色内容部分已截图发出: 是 以上选择我承诺是永久有效的,接受社区和佬友监督: 是 以下为项目介绍正文内容,AI生成、润色内容已使用截图方式发出 之前我在 @inliver 佬的授权下发布了 exe reverse 然后有的佬就说诶能不能弄一个 js reverse,然后在拷打大肥鱼之后还是千呼万唤始出来了 ( 目前整个 reverse skill 宇宙都是随缘

【富可敌国】星驰公益每天放福利啦 送1000刀 送1000刀 无任何充值

#星驰公益 各位佬友 领兑换码 cdk.linux.do LINUX DO CDK Linux Do 社区 CDK 快速分享平台 - 让分享变得更简单 去哪用 领到之后到这里兑换: xingchiapi.com 星驰API公益站 - AI API Gateway 评论点赞随机私聊发送发兑换码 支持模型 TG频道:Telegram: Telegram: View @xingchigongyi Telegram: Join Group Chat 154 个帖子 - 147 位参与者 阅读完整话题

Antigravity免费用户将获得Gemini 4 hydrogen使用权限

Google Antigravity Docs Choose a model | Google Antigravity Docs Antigravity gives you access to leading frontier reasoning models from Google DeepMind and partner providers. You can switch models at any time and choose a… 根据官方文档,免费用户与Google plus将来也能使用Gemini 4 hydrogen 试用Pro(不含pixel,如jio,学生试用)和免费用户与

【开源】 我可能是第一个让 PowerShell 真正支持“鼠标点哪删哪”的人

[!note]- 总结 我的帖子已经打上 开源推广 标签: 是 我的开源项目完整开源,无未开源部分: 是 我的开源项目已链接认可 LINUX DO 社区: 是 我帖子内的项目介绍,AI生成、润色内容部分已截图发出: 是 以上选择我承诺是永久有效的,接受社区和佬友监督: 是 以下为项目介绍正文内容,AI生成、润色内容已使用截图方式发出 项目出发点其实是眼馋mac的各种优秀的终端,分屏,终端持久化,消息通知,命令补齐等等,由于win上实在没有太全面的终端;于是乎我又来vibing啦 PC端 效果如下,现在支持点击任意位置;甚至是多选文字直接删除,妈妈再也不用担心我们输错啦~ [!tip] 项目反馈

RelayFor近期待办

Todolist,定期更新,各位佬可以在下方评论,提出意见。 模型可用性改善 目标:模型整体错误率保持4.5%以下,并优化首字等指标 正在努力,11/13日错误率及原因: 原因:渠道本身没问题,触发了熔断,情况未知导致服务中断 价格问题 目标:倍率恢复到1x-1.2x之间的正常倍率 gpt-6.1-sol渠道比较丰富,已调整倍率 可用度 目标:有调用的时长22小时+ (即每天允许2小时无人调用或不可用) 功能待办 1.更准确的token估算 over. 16 个帖子 - 10 位参与者 阅读完整话题

muse免费白嫖资源一条龙

直接在下方复制提示词,粘贴给muse: 请帮我系统开通并配置一套免费白嫖资源组合,目标是搭建低成本 AI 算力 + 云电脑 + 节点环境。全程只用免费计划,不绑卡、不升级、不付费。 ### 核心要求(必须严格遵守) 1. 只用免费计划,绝不绑卡、不升级、不购买任何付费选项。 2. 遇到需要登录、Google 授权、手机验证码、人脸识别、协议确认、密码输入时,**立即停下来**,把页面留给我接管,等我完成后再继续。 3. 不要读取、存储或泄露我的任何密码、验证码、支付信息。 4. 每完成一项资源,先实际验证是否可用(例如运行 nvidia-smi、跑一个小 Python 程序、生成一张测试图、输

LLM科研写作模型怎么选?

先说结论: 在科研相关的文本写作领域中,Gemini具有极大的优势,仅Opus5.5有一战之力,Fable和Sonnet也不太行。4代Gemini里,Argon更强大更通用,无论是内容还是文本本身,都可以适用,应作为首选;Carbon更适合特别复杂科学问题相关的撰写,其余方面与Argon持平或偏弱。 整体文本能力方面,Arena的榜单显示Carbon和Argon较几款Claude模型有明显领先,其中包括最近在语言方面口碑极速逆转的Opus 5.5和久负盛名的Opus 4.6。 专业领域方面Fable5超越了Argon,Carbon有较大幅度的领先,这意味着Carbon模型确实在解决写作方面复杂

英伟达已死,在自毁长城之后

作者:Fei Peng @fiigii 身份:英伟达(NVIDIA)CUTLASS 与 CUDA 编译器工程师 —————— 所有人都以为,CUDA 是英伟达的护城河,是我们能在 AI 时代一直保持统治地位的原因。 但我从不这么认为。 英伟达真正的护城河只有一条:它的人。 而其中最重要的,就是黄仁勋(Jensen Huang)。 我在 CUDA 编译器、底层库和 DSL 领域深耕了八年,经历了从 Volta 到 Rubin 的每一代架构。我想,我有资格分享一下自己对“英伟达护城河究竟是什么”的看法。请耐心听我说完。 【1. 什么是 CUDA?】 很多人以为,CUDA 只是一门编程语言和它的编译