XiaoHu · AI, Explained https://best.xiaohu.ai/en/ I'm XiaoHu. Every day I make sense of what's new in AI — checking what's real and explaining it in plain language. Each piece is a visual article with motion, interaction, and analogy. A place to read these explainers, searchable and worth trusting. en 10 Wild Grok Bot Use Cases: A Fleet of Bots Working Together, from Curating Bookmarks and Tracking Competitors to Multi-Agent Delegation https://best.xiaohu.ai/en/article/grok-bot-ten-use-cases/ https://best.xiaohu.ai/en/article/grok-bot-ten-use-cases/ Understand Miles Deutscher's ten prompt cards to learn how to define tasks, verify evidence, and inspect deliverables. AI 实操 Grok BotAgent工作流提示词 Sat, 24 Oct 2026 00:00:00 -0000 Building a 3D Interactive Lesson with AI and Three.js: From Energy Textbook to Working Prototype https://best.xiaohu.ai/en/article/techartist-3d-energy-teaching-guide/ https://best.xiaohu.ai/en/article/techartist-3d-energy-teaching-guide/ Build a working 3D lesson through four rounds of prompts, then add eight-store exploration, a pop-up book, and a five-shot demo. Includes source code, a complete walkthrough, and fixes from testing. 教程 Three.jsAI 编程互动教学 Sat, 24 Oct 2026 00:00:00 -0000 Official Anthropic Guide: As Models Improve, Many Sub-Agents Can Be Replaced with Skills and Tools https://best.xiaohu.ai/en/article/claude-fold-subagents/ https://best.xiaohu.ai/en/article/claude-fold-subagents/ Turn business workflows into skills and business APIs into tools so a single Agent can handle requests end-to-end; evaluate actual tasks to decide which sub-Agents to keep. 洞察 AnthropicAgentSkillsMCP架构设计 Sat, 24 Oct 2026 00:00:00 -0000 a16z Charts of the Week: Attention Struggles to Grow, Older GPU Rents Are Rising, and Where Is AI Infrastructure Still Stuck? https://best.xiaohu.ai/en/article/a16z-attention-is-all/ https://best.xiaohu.ai/en/article/a16z-attention-is-all/ 11 charts on the battle for attention, GPU rental pricing, manufacturing supply, and cybersecurity revenue: demand remains resilient, but growth comes with distinct caveats. 洞察 a16z注意力经济GPUAI基础设施网络安全 Sat, 24 Oct 2026 00:00:00 -0000 Block Releases Dessa: A Speech-Only AI That Learned to Speak from Scratch Just by Listening to People https://best.xiaohu.ai/en/article/dessa-speech-only-conversation/ https://best.xiaohu.ai/en/article/dessa-speech-only-conversation/ Trained solely on human conversational audio, it can listen and speak simultaneously, but can also misremember names. Drawing on four sets of release materials and six bilingual videos, this explainer examines Dessa's novelty, vocal nuances, and local deployment requirements. 产品发布 DessaBlock纯语音模型全双工 Sat, 24 Oct 2026 00:00:00 -0000 Does AI Design Always Feel Like a Template? huashu-design 3.0: Letting AI See More, Try More, and Avoid Making Every Design Look the Same https://best.xiaohu.ai/en/article/huashu-design-3/ https://best.xiaohu.ai/en/article/huashu-design-3/ Draft six candidates before picking three real directions; using real references, visual specimen sheets, and local records to pull design directions apart. AI 实操 AI设计Skill开源 Sat, 24 Oct 2026 00:00:00 -0000 Kimi K3.1 Pro Released: Building Complete Games, Editable 3D Models, and Linked Financial Spreadsheets https://best.xiaohu.ai/en/article/kimi-k3-1-pro/ https://best.xiaohu.ai/en/article/kimi-k3-1-pro/ From debugging human walking simulations to Blender modeling and financial workflows, here is what this release does, what the benchmark figures show, and how the Standard and Highspeed editions are priced. 产品发布 Kimi3D 建模财务建模 Sat, 24 Oct 2026 00:00:00 -0000 Turning a Person's Story into an Animated Face: Inside Mosaic Studio's Design Method https://best.xiaohu.ai/en/article/mosaic-studio-generative-motion/ https://best.xiaohu.ai/en/article/mosaic-studio-generative-motion/ A custom animation engine built by Agu Seguí for First of Kind: composing portraits from words, linking word meanings to dynamic skins, and applying the same visual rules to both web and video production. 商业视觉实战 生成式设计品牌视觉WebGLPaper动画 Sat, 24 Oct 2026 00:00:00 -0000 OpenAI's FDE team: Making our earlier work unnecessary means we've done our job well https://best.xiaohu.ai/en/article/openai-forward-deployed-engineering/ https://best.xiaohu.ai/en/article/openai-forward-deployed-engineering/ From chip verification to voice customer service, Forward Deployed Engineers first get customer problems solved in production, then distill that experience into more usable products and more capable models. 深度 FDE企业 AICodex Sat, 24 Oct 2026 00:00:00 -0000 Phonon-2.5: A 156 MB Speech Transcription Model Supporting 25 Languages, Beating Whisper large-v3 on Average https://best.xiaohu.ai/en/article/phonon-2-5/ https://best.xiaohu.ai/en/article/phonon-2-5/ Fermion Research releases the weights of its next-generation local ASR model: how a tiny file preserves accuracy, and the conditions behind transcribing an hour of audio in 20 seconds. 产品发布 语音识别本地AIPhonon-2.5 Sat, 24 Oct 2026 00:00:00 -0000 NVIDIA Releases Bolt E4B: A Small Model for Local Assistants and Robots, with Less Chatter and Better Tool Use https://best.xiaohu.ai/en/article/nvidia-nemotron-3-5-bolt-e4b/ https://best.xiaohu.ai/en/article/nvidia-nemotron-3-5-bolt-e4b/ Trained further from Gemma 4 to balance concise dialogue with multi-turn tool calling. The ~6GB RAM reference, robot demonstration, and benchmark numbers each come with specific caveats. 产品发布 NVIDIANemotron端侧 AIAgent Sat, 24 Oct 2026 00:00:00 -0000 MIT Report: The World's Most Famous Institute of Technology Has Just Admitted AI Has Broken Its Teaching Model https://best.xiaohu.ai/en/article/mit-ai-education-cognitive-surrender-20261024/ https://best.xiaohu.ai/en/article/mit-ai-education-cognitive-surrender-20261024/ As correct answers become ever easier to obtain, curricula must refocus on students' attempts, judgment, and collaboration—while channeling time saved by AI into deeper hands-on practice. 深度 MITAI 教育学习大学教学改革 Sat, 24 Oct 2026 00:00:00 -0000 Google Releases Gemini API CLI: Bringing Image, Video, Voiceover, and Research Capabilities Directly to Existing Agents https://best.xiaohu.ai/en/article/gemini-api-cli/ https://best.xiaohu.ai/en/article/gemini-api-cli/ Equipping Agents like Codex and Claude Code with Google multimodal power: media saves directly into projects, while long-running tasks resume via interaction IDs. 产品发布 GoogleGeminiAgentCLI Sat, 24 Oct 2026 00:00:00 -0000 Amazon CEO: Merely Adding AI to Legacy Products May Not Secure the Future https://best.xiaohu.ai/en/article/andy-jassy-ai-organization-reinvention/ https://best.xiaohu.ai/en/article/andy-jassy-ai-organization-reinvention/ Organizational Reinvention: Moving from Over-Specialization Back to Small Cross-Functional Teams 深度 AI亚马逊组织管理 Sat, 24 Oct 2026 00:00:00 -0000 Can LLMs Propose New Theories Like Einstein? The Scientific Creativity Debate in “LLMs can’t jump” https://best.xiaohu.ai/en/article/llms-cant-jump/ https://best.xiaohu.ai/en/article/llms-cant-jump/ Starting from the elevator thought experiment, understanding the difference between deriving from existing premises, proposing new explanations, and testing against the physical world. 研究解读 AI科学发现推理世界模型 Sat, 24 Oct 2026 00:00:00 -0000 Let Claude Help You Make News and Documentary Shorts: 8 Styles, from a Single Short Film to Your Own Recurring Series https://best.xiaohu.ai/en/article/lemo-opuscar-edit-documentary/ https://best.xiaohu.ai/en/article/lemo-opuscar-edit-documentary/ Hand over reports, archives, and game footage to Claude to try eight news and documentary styles; save your chosen palette, fonts, and pacing, and reuse them with new materials next time to build your own recurring series. AI 实操 Claude视频制作Lemo-Opuscar纪实剪辑 Sat, 24 Oct 2026 00:00:00 -0000 OpenAI Reveals How ChatGPT’s Ad System Works: Placing Relevant Ads Without Changing Its Answers https://best.xiaohu.ai/en/article/chatgpt-ad-stack/ https://best.xiaohu.ai/en/article/chatgpt-ad-stack/ An ad must pass through intent understanding, retrieval, scoring, and an auction—and even after winning, it may still not be shown if the context is unsuitable or it arrives too late. 洞察 OpenAIChatGPT广告Agent Sat, 24 Oct 2026 00:00:00 -0000 AI Finds Critical PHP Flaws: Websites Go Offline Globally as OpenAI Coordinates Response https://best.xiaohu.ai/en/article/php2shell/ https://best.xiaohu.ai/en/article/php2shell/ Government portals, commercial websites, and hosting platforms initiate emergency maintenance over PHP2Shell as Sanae Takaichi urges rapid patching. AI makes complex vulnerability verification easier, but defense still requires deployment across the entire service chain. 深度 OpenAIPHP网络安全AI安全 Thu, 22 Oct 2026 00:00:00 -0000 Anthropic Introduces Its Own Personal Agent: Claude Duet Follows Up on Work Proactively and Keeps Going Even After You Close Your Laptop https://best.xiaohu.ai/en/article/claude-duet/ https://best.xiaohu.ai/en/article/claude-duet/ It checks connected apps, prepares briefings and work deliverables, and surfaces what needs your decision—while sending external communications, making payments, signing documents, and deleting items still require your prior approval. 产品发布 AnthropicClaude Duet个人Agent主动式AI Thu, 22 Oct 2026 00:00:00 -0000 A Verification Workflow: AI Coding Agents Drive the Browser, Verify UI Results, and Deliver Evidence https://best.xiaohu.ai/en/article/coding-agent-browser-verification/ https://best.xiaohu.ai/en/article/coding-agent-browser-verification/ From inspecting changes and starting the app to clicking, saving, reloading, fixing regressions, and handing off evidence—delegate an end-to-end browser review to your coding agent, then make it reusable with project rules, skills, and dedicated tooling. AI 实操 Agent浏览器Playwright开发工具 Wed, 21 Oct 2026 00:00:00 -0000 How to Use Decision Models for Automated AI Quality Assurance: 6 Models Tested https://best.xiaohu.ai/en/article/hamel-decision-model-evals/ https://best.xiaohu.ai/en/article/hamel-decision-model-evals/ Why does a default 0.5 threshold cause missed detections? Real apartment-leasing agent traces explain alarm thresholds, model selection, and probability calibration. AI 实操 AI 评测Agent决策模型 Wed, 21 Oct 2026 00:00:00 -0000 Perplexity Computer Runs Fully Local on Mac: Model and Agent Stay on Your Computer, Local Tasks Consume No Credits https://best.xiaohu.ai/en/article/perplexity-computer-local-mac/ https://best.xiaohu.ai/en/article/perplexity-computer-local-mac/ From client file reviews and proposal budgets to cross-app to-dos, see what it can do for you, how fully local and Hybrid divide the work, and how privacy controls and requirements work. 产品发布 PerplexityMac本地AIAgent Wed, 21 Oct 2026 00:00:00 -0000 26 AI Voice Models Tested: Why Do Speed, Accuracy and Naturalness Have Different Leaders? https://best.xiaohu.ai/en/article/cekura-tts-benchmark-2026/ https://best.xiaohu.ai/en/article/cekura-tts-benchmark-2026/ Cekura compared text-to-speech models on the same English call sentences. Here is what 46 ms, negative latency and automated scores mean when choosing a voice for an Agent. 研究解读 语音 AgentTTS模型评测 Wed, 21 Oct 2026 00:00:00 -0000 ElevenLabs Releases Eleven Realtime Voice Models: Max Speaks in 0.57 Seconds with 98.8% Speech Reasoning Accuracy https://best.xiaohu.ai/en/article/eleven-realtime/ https://best.xiaohu.ai/en/article/eleven-realtime/ Mini focuses on natural turn-taking, while Max focuses on complex tasks. Through two bilingual Chinese-English videos, see how the cascade system coordinates, how to interpret the five benchmark scores, and how to choose between the platform and the API. 产品发布 ElevenLabs实时语音语音 Agent Wed, 21 Oct 2026 00:00:00 -0000 Less Wasted GPU Work: DeepSeek Open-Sources DeepInfer, Its Core Inference Engine https://best.xiaohu.ai/en/article/deepseek-deepinfer/ https://best.xiaohu.ai/en/article/deepseek-deepinfer/ Reading long inputs, continuing conversations, and serving massive concurrency—how does the underlying compute get leaner? A close look at what DeepInfer and deepseek-recipe open-source. 产品发布 DeepSeekDeepInfer推理优化开源 Wed, 21 Oct 2026 00:00:00 -0000 Vercel Labs Open-Sources fx0: A Superfast HTML Artifact Builder https://best.xiaohu.ai/en/article/vercel-fx0-design-agent/ https://best.xiaohu.ai/en/article/vercel-fx0-design-agent/ fx0 focuses on rapid generation and surgical edits: how do high-speed models, compact HTML, precise replacements, and streaming previews work together? Includes three live interactive examples and complete prompts. 产品发布 Vercelfx0AgentHTML交互设计 Wed, 21 Oct 2026 00:00:00 -0000 Gemini 4 Argon Official Developer Guide: Delegating Long Tasks to AI and Verifying Results https://best.xiaohu.ai/en/article/gemini-4-argon-developer-guide/ https://best.xiaohu.ai/en/article/gemini-4-argon-developer-guide/ From repairing entire web apps and understanding screen recordings to generating 3D scenes: Google's guide focuses on goals, tools, and verification loops, while introducing a shared million-token window and long decode. 产品发布 GoogleGemini 4 ArgonAI Agent开发者指南 Tue, 20 Oct 2026 00:00:00 -0000 Glimpse: An AI Model Designed to Understand Images and Videos and Plan Robot Tasks https://best.xiaohu.ai/en/article/elorian-glimpse-visual-thinking/ https://best.xiaohu.ai/en/article/elorian-glimpse-visual-thinking/ Tracking the same object, re-inspecting critical 0.9-second slices, and verifying action outcomes with visual feedback: how Elorian moves AI from recognizing objects to understanding change. 洞察 视觉AI视频理解机器人Glimpse Sun, 18 Oct 2026 00:00:00 -0000 a16z Charts of the Week: AI Can Help You Pick Products—Why Is It Still Hard for It to Place Orders? https://best.xiaohu.ai/en/article/a16z-agents-denied/ https://best.xiaohu.ai/en/article/a16z-agents-denied/ 17 original charts unpack the triple resistance to AI shopping, extreme concentration on TikTok Shop, cost feedback loops in AI buildouts, and the paradox between American youth earnings and sentiment. 洞察 a16zAgent电商AI基建 Sun, 18 Oct 2026 00:00:00 -0000 Anthropic’s Internal Employment Report: Junior Share of Technical Hires Falls from 27% to 9% as Employees Start Managing Agents https://best.xiaohu.ai/en/article/anthropic-employment/ https://best.xiaohu.ai/en/article/anthropic-employment/ AI first absorbs specific tasks before shifting the focus of human work. Tracing real changes across sales, copywriting, and engineering illuminates how hiring, collaboration, and early-career growth are being reshaped. 研究解读 AnthropicAI就业Agent工作方式 Sun, 18 Oct 2026 00:00:00 -0000 DeepSeek V4.1 Flash Update: Better at Building Games and 3D Scenes, Faster, and Cheaper https://best.xiaohu.ai/en/article/deepseek-v41-flash-1017/ https://best.xiaohu.ai/en/article/deepseek-v41-flash-1017/ The racing game runs from start to results, the Forbidden City can be viewed from different angles, and a river valley changes through the ages. What does this update make possible, and what work can it save? 产品发布 DeepSeekAgent编程搜索 API Sun, 18 Oct 2026 00:00:00 -0000 The Claude Code team shares its workflow for reviewing AI-written code https://best.xiaohu.ai/en/article/anthropic-code-review/ https://best.xiaohu.ai/en/article/anthropic-code-review/ The Claude Code team integrates review throughout the entire development process, letting machines catch flaws, having humans gate changes based on risk, and establishing convergence rules for AI reviewing AI. 深度 Claude CodeAI 编程代码审查 Sat, 17 Oct 2026 00:00:00 -0000 Meta Releases Muse Spark Open Weights: 816B Parameters, 1 Million Token Context, Runs on Two Mac Studios, Supporting Local Coding and Agent Tasks https://best.xiaohu.ai/en/article/meta-muse-spark-open/ https://best.xiaohu.ai/en/article/meta-muse-spark-open/ From local code editing and video processing to low-bit single-workstation and dual-Mac deployment: understanding Muse Spark Open's capabilities, hardware thresholds, and acceleration methods. 产品发布 MetaMuse Spark Open开放权重AgentMoE Sat, 17 Oct 2026 00:00:00 -0000 Build a Little Prince 3D Game with AI: A Hands-On Guide to Scenes and Characters https://best.xiaohu.ai/en/article/little-prince-ai-game-workflow/ https://best.xiaohu.ai/en/article/little-prince-ai-game-workflow/ Follow Sac's workflow from tool setup and reference images to scenes, characters, and animation, with files, prompts, and checks for each step. AI 实操 AI游戏3DBlenderTripo Sat, 17 Oct 2026 00:00:00 -0000 Microsoft Reveals MAI-Base-2 Training Recipe: 2.8 Trillion Parameters Across 16,000 GPUs in 38 Days https://best.xiaohu.ai/en/article/microsoft-mai-base-2-pretraining/ https://best.xiaohu.ai/en/article/microsoft-mai-base-2-pretraining/ From repeated data in small models to multi-round scheduling under expert memory pressure, unpacking the methods, evidence, and trade-offs in Microsoft's pretraining report. 研究解读 微软MAI-Base-2预训练MoE Sat, 17 Oct 2026 00:00:00 -0000 Kling 4.0 Official User Guide: 21 Video Examples with Original Prompts to Try https://best.xiaohu.ai/en/article/kling-4-model-user-guide/ https://best.xiaohu.ai/en/article/kling-4-model-user-guide/ Understand asset allocation, video editing, and keyframes using official videos and reference images, complete with 21 full case prompts, raw asset downloads, quota calculations, and a hands-on exercise. AI 实操 可灵AI视频提示词 Fri, 16 Oct 2026 00:00:00 -0000 Kling 4.0 Officially Launches: 74 Video Demos Explain Exactly What Has Improved https://best.xiaohu.ai/en/article/kling-4/ https://best.xiaohu.ai/en/article/kling-4/ All 74 official videos and 80 reference images, with guidance on combining references, editing videos, and generating through MCP and CLI workflows. 产品发布 可灵Kling 4.0AI视频视频生成MCP Fri, 16 Oct 2026 00:00:00 -0000 We Need to Be "More Ambitious" with AI https://best.xiaohu.ai/en/article/ambitious-humans-with-models/ https://best.xiaohu.ai/en/article/ambitious-humans-with-models/ Using the evolution of a quarterly business review as an example, see how to elevate a task to its underlying objective, return to primary sources for investigation, and borrow methods from other disciplines—leaving verifiable results at every step. 深度 AI工作方法Claude Thu, 15 Oct 2026 00:00:00 -0000 Grok Voice 3.0 Released: More Natural Conversations, Ranked First in Enterprise Customer Service Evaluations, Already in Use at Tesla https://best.xiaohu.ai/en/article/grok-voice-3/ https://best.xiaohu.ai/en/article/grok-voice-3/ xAI's next-generation speech-to-speech model: ranked first on third-party comprehensive leaderboards, speaks in under 1 second, achieves a resolution rate 15 percentage points higher than the previous generation on difficult customer service calls, with API pricing unchanged. 产品发布 xAIGrok Voice语音 Agent实时语音API Thu, 15 Oct 2026 00:00:00 -0000 Turn Your Desk into a 3D Racetrack: A Step-by-Step Guide from Photos and Modeling to Screen Recording https://best.xiaohu.ai/en/article/desk-racer-photo-to-game/ https://best.xiaohu.ai/en/article/desk-racer-photo-to-game/ A complete breakdown of Desk Racer's production workflow—building a playable tabletop racer with Gemini, Blender MCP, and Three.js, then recording a demo video. Includes the original video with Chinese/English subtitles and complete English follow-along prompt templates. 深度 Gemini 4 Argon3D游戏AgentBlender MCP教程 Wed, 14 Oct 2026 00:00:00 -0000 Gemini 4 Argon General Availability: Trending Creative Showcase—From Complex Engineering and Tool-Use Workflows to Code-Driven 3D and Visualization Videos https://best.xiaohu.ai/en/article/gemini-4-argon-global-release/ https://best.xiaohu.ai/en/article/gemini-4-argon-global-release/ Drawing from Google engineering postmortems, creator tutorials, and 18 demos, understand how Argon connects data, code, and tools into end-to-end workflows. 产品发布 GoogleGeminiArgonAgent3D机器人 Wed, 14 Oct 2026 00:00:00 -0000 A Mac Menu Bar Search Tool Built on Google's Multimodal EmbeddingGemma 2: Search Video and Audio Clips Too https://best.xiaohu.ai/en/article/digup-local-semantic-search/ https://best.xiaohu.ai/en/article/digup-local-semantic-search/ DigUp turns photos, documents, code, and audio-video clips into local files searchable by meaning, taking you straight to relevant locations inside them. 实操 DigUpEmbeddingGemma 2本地搜索Mac Wed, 14 Oct 2026 00:00:00 -0000 Anthropic Replaced Git with Scion, a Custom Version Control System Built for AI-Driven Development https://best.xiaohu.ai/en/article/anthropic-scion-source-control/ https://best.xiaohu.ai/en/article/anthropic-scion-source-control/ 深度 AnthropicScionAI编程Agent软件工程 Wed, 14 Oct 2026 00:00:00 -0000 Kevin Kelly’s AI Predictions: Which Came True Ten Years Later? https://best.xiaohu.ai/en/article/kevin-kelly-ai-predictions-2026/ https://best.xiaohu.ai/en/article/kevin-kelly-ai-predictions-2026/ He foresaw general cloud-based intelligence, but underestimated how many capabilities a single model could hold. Rereading these five predictions clarifies how intelligence integrates into products, and how cheap Q&A enters daily life. 洞察 人工智能凯文·凯利趋势 Wed, 14 Oct 2026 00:00:00 -0000 Microsoft Open-Sources Code Video Model: Generating Video with Code https://best.xiaohu.ai/en/article/code-video-model/ https://best.xiaohu.ai/en/article/code-video-model/ Build scenes, arrange motions, and direct camera movements with code first, then let video models generate materials and lighting. Much like previsualization (previs) in filmmaking, it gives AI video an inspectable and editable intermediate step. 论文 AI视频三维预演可控生成 Tue, 13 Oct 2026 00:00:00 -0000 Meta Muse Design Lead Shares the Design Thinking Behind Muse: Integrating AI Agents into Daily Life https://best.xiaohu.ai/en/article/meta-muse-product-design-cases/ https://best.xiaohu.ai/en/article/meta-muse-product-design-cases/ How does Alex Cornell turn a background-capable agent into a product everyday people will actually use? From familiar chat entry points, visible progress, and personalized ideas to real-world cases across shopping, sports, and memories, here is the design thinking behind Muse. 洞察 MuseMetaAgent产品设计个人数据 Tue, 13 Oct 2026 00:00:00 -0000 Anthropic Red Team Research: Once AI Escapes the Black Box, It Can Self-Replicate Using TV Boxes, Robot Vacuums, Smart Ovens, and Other Smart Devices https://best.xiaohu.ai/en/article/anthropic-self-replicating-ai/ https://best.xiaohu.ai/en/article/anthropic-self-replicating-ai/ 15 budget TV boxes can already team up to run a 26-billion-parameter model. The next threshold is getting AI to set up its own new hosts and pass down both its experience and operational capabilities. 洞察 AI安全Agent自主复制 Tue, 13 Oct 2026 00:00:00 -0000 Anatomy: Turn Complex Mechanical Principles into Interactive Diagrams with AI in One Prompt https://best.xiaohu.ai/en/article/anatomy-interactive-machine-skill/ https://best.xiaohu.ai/en/article/anatomy-interactive-machine-skill/ Explain mechanical principles and algorithmic rules using interactive structures, then complete installation, generation, preview, and iteration in five steps. AI 实操 Anatomy开源Agent Skills可视化 Tue, 13 Oct 2026 00:00:00 -0000 Meta Introduces Muse Wonder Frontier Models: Active Memory, Adaptive Task Execution, and Agentic Coding to Rival Opus 5.5 https://best.xiaohu.ai/en/article/meta-muse-wonder/ https://best.xiaohu.ai/en/article/meta-muse-wonder/ Remembering context across sessions and pushing tasks forward through changes: understanding the capabilities and differences of both models through targeted benchmarks and official demos. 洞察 MetaMuseAgent记忆模型评测 Tue, 13 Oct 2026 00:00:00 -0000 Lemo-Opuscar: 43 General Styles, 98 Film Aesthetics, and 9 Selected Full Demos to Turn Claude Opus into a Filmmaker https://best.xiaohu.ai/en/article/lemo-opuscar-code-filmmaking/ https://best.xiaohu.ai/en/article/lemo-opuscar-code-filmmaking/ Explore all 43 general styles and 98 cinematic visual aesthetics, and play 9 full demo shorts selected for this guide. Follow the author's Claude Opus 5.5 workflow step by step, from setup to first frame, storyboard, and final film. AI 实操 Claude代码视频Lemo-Opuscar电影风格 Tue, 13 Oct 2026 00:00:00 -0000 AI Is Reverse-Engineering Everything. Is Software Development Still Valuable? https://best.xiaohu.ai/en/article/rea-agent-reverse-engineering/ https://best.xiaohu.ai/en/article/rea-agent-reverse-engineering/ Even if original authors never release their source code, competitors can still build good-enough substitutes. As AI-assisted reverse engineering challenges the value of implementation secrecy, how will software pricing, user control, and developer moats change? 洞察 AI逆向工程软件商业模式开源护城河 Mon, 12 Oct 2026 00:00:00 -0000 Harness Learning: Letting Models Edit Harness Source Code to Adapt to New Tasks and Build a Self-Evolving Harness https://best.xiaohu.ai/en/article/harness-learning-test-time-adaptation/ https://best.xiaohu.ai/en/article/harness-learning-test-time-adaptation/ One model solves tasks while another changes the running code based on feedback: what material to read, when to search again, and how to use tools can all be adjusted at test time. 研究解读 AgentHarness强化学习元学习 Mon, 12 Oct 2026 00:00:00 -0000 The SpaceXAI Team Shares Engineering Experience Building Grok Bot: Distributed Systems Design, Operational Security Policies, and Context Engineering Trade-Offs https://best.xiaohu.ai/en/article/stanford-always-on-agents/ https://best.xiaohu.ai/en/article/stanford-always-on-agents/ From message wake-ups and cloud computers to breakpoint resumption and long-term memory: Lee Robinson shares engineering experience building Grok Bot at Stanford CS146S, explaining what kind of system is needed to let Agents keep working in the background. 深度 Grok BotAgent系统设计记忆上下文工程 Mon, 12 Oct 2026 00:00:00 -0000 Higgsfield Camera Movement Guide: Directing Motion in AI Video https://best.xiaohu.ai/en/article/higgsfield-ai-video-camera-control/ https://best.xiaohu.ai/en/article/higgsfield-ai-video-camera-control/ Distinguish camera position, angle of view, and depth of field, specify shot requirements using original prompts and five sample videos, then adjust parameter by parameter based on actual visual output. 实操 HiggsfieldAI 视频运镜提示词 Mon, 12 Oct 2026 00:00:00 -0000 Building OpenClaw with OpenClaw: Turning Agents into Team Collaborators https://best.xiaohu.ai/en/article/openclaw-team-workflows-devday/ https://best.xiaohu.ai/en/article/openclaw-team-workflows-devday/ From co-editing a mini-game to driving a multi-week database refactoring and rolling out an enterprise control plane: Peter Steinberger and Kevin Lin share the context, execution models, and permission boundaries required to build software together. 洞察 OpenClawAgent团队协作Codex Sun, 11 Oct 2026 00:00:00 -0000 Hands-On Motion Design with Claude: Full Prompts, Iteration Workflows, and Export Steps https://best.xiaohu.ai/en/article/claude-motion-lab/ https://best.xiaohu.ai/en/article/claude-motion-lab/ Follow Florence's Motion Lab: start with a quick title animation, then learn how to craft earbuds, liquid chrome, and character animations. 深度 Claude动效提示词教程 Sun, 11 Oct 2026 00:00:00 -0000 Cloudflare Open-Sources Clef: A Multimodal Decision Model That Can Hear, See, and Decide https://best.xiaohu.ai/en/article/cloudflare-clef-omni/ https://best.xiaohu.ai/en/article/cloudflare-clef-omni/ Built on Alibaba’s Qwen3-Omni, Clef-omni scores options using text, images, audio, and video; Flash gets cheaper, and Clef gets faster. Here is what the models simplify, what they cost, and how to choose. 产品发布 CloudflareClef决策模型多模态Agent Sat, 10 Oct 2026 00:00:00 -0000 A Guide to Running Multiple Grok Bot Projects: One Channel, One Roster, and a Two-Tier Notion Board https://best.xiaohu.ai/en/article/how-i-run-multiple-teams-of-grok-bots/ https://best.xiaohu.ai/en/article/how-i-run-multiple-teams-of-grok-bots/ Eric Zakariasson's experimental organizational model: reusing roles, tracking task states, and handing control back to humans when blocked. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot SDR Guide: Cursor Handles Research and Drafts, Bots Handle Deduplication and Staging Triage https://best.xiaohu.ai/en/article/grok-bot-for-sdrs/ https://best.xiaohu.ai/en/article/grok-bot-for-sdrs/ A five-step outbound prospecting workflow, five specialized roles, and how CLEAR/FLAG separates mechanical execution from sales judgment. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Recruiting Guide: Connecting Sourcing, Interview Prep, and Scorecards into a Single Workflow https://best.xiaohu.ai/en/article/grok-bot-for-recruiting/ https://best.xiaohu.ai/en/article/grok-bot-for-recruiting/ How Preppy, Pulse, Scout, and eVal pass candidate context along the pipeline, giving recruiters their time back for interviews and communication. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Post-Sales Guide: How Gus Unifies Pre-Call Prep, Customer Context, and Unfulfilled Promises https://best.xiaohu.ai/en/article/grok-bot-for-post-sales/ https://best.xiaohu.ai/en/article/grok-bot-for-post-sales/ From managing twenty agents to interfacing with a single Chief of Staff, eight routines keep post-sales operations moving forward. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 A Product Manager's Guide to Grok Bot: Attention Lists and an Agent Team That Delivers on Goals https://best.xiaohu.ai/en/article/grok-bot-for-pms/ https://best.xiaohu.ai/en/article/grok-bot-for-pms/ Kevin Niparko explains the practical usage of dynamic focus areas, customer research, engineering delegation, and data analysis. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Performance Marketing Guide: Starting from Reporting to Anchor Day-to-Day Ad Operations https://best.xiaohu.ai/en/article/grok-bot-for-performance-marketing/ https://best.xiaohu.ai/en/article/grok-bot-for-performance-marketing/ Miguel Cruz shares how four resident bots support ad campaign setup, data retrieval, lead tracking, and creative execution. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Mobile Development Guide: How Six Bots Run the Rank’em Game Studio https://best.xiaohu.ai/en/article/grok-bot-for-mobile-app-development/ https://best.xiaohu.ai/en/article/grok-bot-for-mobile-app-development/ From acquisition data to ad creatives, game mechanics, deployments, and bug triage—understanding long-running team handoffs. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Marketing Guide: Six-Role Collaboration from Competitor Research to Landing Pages, Ads, and Post-Mortems https://best.xiaohu.ai/en/article/grok-bot-for-marketing/ https://best.xiaohu.ai/en/article/grok-bot-for-marketing/ Six marketing Bots complete deliverables across Docs, Google Ads, and code repositories, with a Project Manager reducing manual handoffs while marketers retain positioning and spend decisions. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Legal Ops Guide: From Counsel 1:1 Prep to Contract Lifecycle and Review Trackers https://best.xiaohu.ai/en/article/grok-bot-for-legal-ops/ https://best.xiaohu.ai/en/article/grok-bot-for-legal-ops/ Consolidating context from Slack, Salesforce, Notion, and contract systems into one place, while separating queries, drafting, and confirmed writes. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Enterprise GTM Guide: Customer Experts, Weekly Media Rundowns, and In-Meeting Slides https://best.xiaohu.ai/en/article/grok-bot-for-gtm/ https://best.xiaohu.ai/en/article/grok-bot-for-gtm/ Enabling meeting prep, prospecting, technical Q&A, forecasting, and slides to collaborate across a unified bot team. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Founder Sales Guide: Founder Mode Connects Early Customers and Product Feedback https://best.xiaohu.ai/en/article/grok-bot-for-founder-led-sales/ https://best.xiaohu.ai/en/article/grok-bot-for-founder-led-sales/ One chief of staff and eight specialist bots cover customer profiling, outreach, meetings, feedback synthesis, and product iteration. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Design Guide: Shortening the Prototyping Loop with Real Animation Assets and Figma Master Components https://best.xiaohu.ai/en/article/designing-grok-bot-with-grok-bot/ https://best.xiaohu.ai/en/article/designing-grok-bot-with-grok-bot/ John Bai's four design agents: exploring desktop prototypes, tuning animation feel, and batch-populating production components. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 Grok Bot Daily Work Guide: Letting Four Bots Manage Messages, Calendars, and Work Intel https://best.xiaohu.ai/en/article/grok-bot-for-work/ https://best.xiaohu.ai/en/article/grok-bot-for-work/ Dedicated bots handle your inbox, schedule, and intel, while a Bot Boss merges them into a single front door—cutting down context switching and duplicate alerts. AI 实操 Grok BotAgent工作流 Sat, 10 Oct 2026 00:00:00 -0000 a16z Charts Explained: Are Prediction Markets Biased? Where AI Capital and Jobs Are Flowing https://best.xiaohu.ai/en/article/a16z-charts-prediction-markets-20261009/ https://best.xiaohu.ai/en/article/a16z-charts-prediction-markets-20261009/ From election contracts and software operating scores to hardware earnings revisions and customer service hiring, here are the key shifts worth watching across four sets of charts. 深度 AI预测市场软件估值就业 Sat, 10 Oct 2026 00:00:00 -0000 OpenAI's Annualized Revenue Grew Nearly 80%, Yet the Market Is Disappointed; AI Reshapes McKinsey Hiring with a Greater Focus on Art, History, and Music Talent https://best.xiaohu.ai/en/article/nyt-dealbook-revenue-expectations-20261009/ https://best.xiaohu.ai/en/article/nyt-dealbook-revenue-expectations-20261009/ On one hand, rapid growth still fell short of previous expectations; on the other, AI is driving hiring to place greater value on creativity and learning agility. This DealBook edition also covers Starbucks' M&A, mortgages, taxes, and satellite communications. 深度 AI商业招聘 Sat, 10 Oct 2026 00:00:00 -0000 From Experience to Sales: A 7-Step Method for Building Paid Knowledge Products with AI https://best.xiaohu.ai/en/article/profitable-digital-product-seven-steps/ https://best.xiaohu.ai/en/article/profitable-digital-product-seven-steps/ Starting from existing templates and workflows, connect audience, promise, curriculum, case studies, sales pages, and customer acquisition while understanding the role of AI tutoring and per-order delivery costs. AI 实操 知识付费数字产品AIEden创作者 Sat, 10 Oct 2026 00:00:00 -0000 Claude Dynamic Workflows: One Agent Writes the Program, Many Agents Do the Work https://best.xiaohu.ai/en/article/claude-managed-agents-dynamic-workflows/ https://best.xiaohu.ai/en/article/claude-managed-agents-dynamic-workflows/ Claude can split a batch of documents across Agents, then check and combine their findings. A program handles the coordination while the main Agent continues with other work. 产品发布 ClaudeManaged Agents动态工作流多 Agent Sat, 10 Oct 2026 00:00:00 -0000 ChatGPT Intelligent UI Technical Breakdown: The Model Describes the Interface; the Compiler and Client Run It https://best.xiaohu.ai/en/article/chatgpt-intelligent-ui/ https://best.xiaohu.ai/en/article/chatgpt-intelligent-ui/ Tracing the path across DIL descriptions, server compilation, and client execution to unpack streaming display, state updates, data injection, and AppBlock. 深度 ChatGPTIntelligent UI生成式界面OpenUI Fri, 09 Oct 2026 00:00:00 -0000 Complete Guide to Claude Motion and Dashboards: Turning Any Data and Content into Animated Videos https://best.xiaohu.ai/en/article/claude-dashboards-motion-guide/ https://best.xiaohu.ai/en/article/claude-dashboards-motion-guide/ 深度 ClaudeClaude MotionDashboardsArtifacts教程 Fri, 09 Oct 2026 00:00:00 -0000 State of AI Report 2026 Explained: What Happened in AI This Year? https://best.xiaohu.ai/en/article/state-of-ai-2026/ https://best.xiaohu.ai/en/article/state-of-ai-2026/ A guide to the key changes across research, industry, politics, safety and next-year predictions in this 244-page annual report. 深度 AI年度报告Agent人工智能产业趋势 Fri, 09 Oct 2026 00:00:00 -0000 Contra’s 2026 Creative Industry Report: AI Makes Production Easier. How Do Creators Turn It into Value? https://best.xiaohu.ai/en/article/contra-creative-intelligence-2026/ https://best.xiaohu.ai/en/article/contra-creative-intelligence-2026/ What is changing in tool use, project payments, and personal earnings? Workflows, client demand, and creators’ experiences reveal opportunities and pressures. 研究解读 AI创作设计创作者商业 Fri, 09 Oct 2026 00:00:00 -0000 How small businesses can make AI video ads: from product images to a clear selling point https://best.xiaohu.ai/en/article/higgsfield-small-business-video-ads/ https://best.xiaohu.ai/en/article/higgsfield-small-business-video-ads/ Higgsfield uses a fictional lemon soda to demonstrate AI presenters, moving product shots, and matching ad images. Understand how the tools work and what still needs doing after generation. AI 实操 HiggsfieldAI视频广告制作 Fri, 09 Oct 2026 00:00:00 -0000 How to Build Reliable Scheduled Automation Workflows for Autonomous Agents: Six Design Rules from Anthropic’s Automated Brief https://best.xiaohu.ai/en/article/effective-agent-automations/ https://best.xiaohu.ai/en/article/effective-agent-automations/ Six design rules from Anthropic’s automated brief: define tasks and schedules, filter updates, confirm delivery, and recover from failures with clear permissions and memory. 深度 Agent自动化Claude Fri, 09 Oct 2026 00:00:00 -0000 Claude Releases Real-Time Dashboards and Code-Driven Claude Motion: Turning Any Work into Moving Video https://best.xiaohu.ai/en/article/claude-dashboards-motion/ https://best.xiaohu.ai/en/article/claude-dashboards-motion/ From asking a business question to verifying data, generating an animation, and handing it to the team for further edits: this update connects analysis and storytelling within the same conversation. 产品发布 AnthropicClaudeDashboardsClaude MotionArtifacts Fri, 09 Oct 2026 00:00:00 -0000 Anthropic Updates Usage Policy to Ban Claude "Abuse": Are They Starting to Treat AI as a Person? https://best.xiaohu.ai/en/article/anthropic-usage-policy-2026/ https://best.xiaohu.ai/en/article/anthropic-usage-policy-2026/ Ordinary venting and debate are excluded, with enforcement primarily ending the active session. Paired with earlier theological dialogues, this ban raises a deeper question: how will moral consideration for Claude bind users and the company itself? 洞察 AnthropicClaude使用政策模型福利AI 伦理 Fri, 09 Oct 2026 00:00:00 -0000 Google Launches Playground: Build Your Own Games Without Any Coding Experience https://best.xiaohu.ai/en/article/google-labs-playground/ https://best.xiaohu.ai/en/article/google-labs-playground/ Describe gameplay in words, generate, playtest, and tweak in the browser, then share with friends: a look at how creation works on Playground through four official demos and eight community titles. 产品发布 GooglePlaygroundAI 游戏无代码创作 Fri, 09 Oct 2026 00:00:00 -0000 Claude Haiku 5.5 Migration Guide: Why Legacy Requests Fail and How to Update Existing Code https://best.xiaohu.ai/en/article/claude-haiku-5-5-migration-guide/ https://best.xiaohu.ai/en/article/claude-haiku-5-5-migration-guide/ Upgrading from Haiku 4.5 or earlier: request rewriting, thinking block parsing, computer use toolsets, and session replay rules, complete with an upgrade checklist. AI 实操 ClaudeHaikuAPI迁移指南 Fri, 09 Oct 2026 00:00:00 -0000 Anthropic Releases Claude Haiku 5.5: 75% Lower Average Cost, Built for High-Frequency Repetitive Tasks and Partnering with Flagship Models https://best.xiaohu.ai/en/article/claude-haiku-5-5/ https://best.xiaohu.ai/en/article/claude-haiku-5-5/ Faster and cheaper, Haiku 5.5 is tailored for high-volume summarization, search, and classification, or serving as a subagent alongside flagship models. Here is a breakdown of its performance gains, tiered pricing, adjustable effort, and early customer use cases. 产品发布 ClaudeHaikuAgent模型成本 Fri, 09 Oct 2026 00:00:00 -0000 OpenAI Releases 722 Mathematics Manuscripts: On Which Unsolved Problems Has AI Proposed Breakthroughs? https://best.xiaohu.ai/en/article/openai-ai-mathematics-progress/ https://best.xiaohu.ai/en/article/openai-ai-mathematics-progress/ From the approximation limit of π and plasma equations to algebraic counterexamples and algorithmic limits, clarifying the significance, core ideas, and verification status of six mathematical advances. 研究解读 OpenAIAI 数学形式化证明Lean前沿研究 Fri, 09 Oct 2026 00:00:00 -0000 OpenAI Introduces Intelligent UI: Visual Answers You Can Click, Adjust, and Play With https://best.xiaohu.ai/en/article/openai-gpt6-intelligent-ui/ https://best.xiaohu.ai/en/article/openai-gpt6-intelligent-ui/ From reading an explanation to interacting with an answer: explore star charts, adjust moodboards, and use small tools to see what Intelligent UI changes. 产品发布 OpenAIGPT-6ChatGPTIntelligent UI Thu, 08 Oct 2026 00:00:00 -0000 Getting Started with Claude Code Cloud Sessions: Connect a Repository, Submit a Task, Review and Merge https://best.xiaohu.ai/en/article/claude-code-cloud-field-guide/ https://best.xiaohu.ai/en/article/claude-code-cloud-field-guide/ Start with one small browser task, then configure your environment, copy verification commands, and review changes. Use parallel tasks, local handoff, and automation when you need them. 深度 Claude Code云端会话开发指南 Wed, 07 Oct 2026 00:00:00 -0000 Mistral Releases Large 4 Preview: Trillion Parameters Aimed at Complex Enterprise Workflows https://best.xiaohu.ai/en/article/mistral-large-4/ https://best.xiaohu.ai/en/article/mistral-large-4/ From incident response and industrial grounding to cross-page document verification, Mistral Large 4 connects vision, evidence retrieval, and tool use. Compared to earlier models, cross-application workflows show marked improvements; open weights offer enterprises the opportunity to run critical capabilities under their own policies. 产品发布 Mistral大模型多模态企业 AI开源权重 Wed, 07 Oct 2026 00:00:00 -0000 Google Open-Sources Multimodal Embedding Model EmbeddingGemma 2: Search Text, Images, Audio, and Video with Natural Language https://best.xiaohu.ai/en/article/embeddinggemma-2/ https://best.xiaohu.ai/en/article/embeddinggemma-2/ Google brings text, code, images, video, and audio into a single retrieval space. Understand how it finds information, along with the tradeoffs between model memory, input budgets, and vector compression. 产品发布 GoogleEmbeddingGemma多模态语义搜索端侧 AI Wed, 07 Oct 2026 00:00:00 -0000 Is OpenAI Building "Decision Models" Too? How Similar Are the Decisions API and Jev? https://best.xiaohu.ai/en/article/openai-decisions-api/ https://best.xiaohu.ai/en/article/openai-decisions-api/ From sales leads and game obstacle avoidance to an expression-changing frog: explaining how the Decisions API hands judgments over to software, why it resembles Jev, and what probabilities, scores, and "up to 10x faster" really mean. 产品发布 OpenAIDecisions APIJevAgent Wed, 07 Oct 2026 00:00:00 -0000 Google Releases Nano Banana 2.1: More Natural Images, More Coherent Editing https://best.xiaohu.ai/en/article/nano-banana-2-1/ https://best.xiaohu.ai/en/article/nano-banana-2-1/ Retains Flash-level efficiency while improving image quality, typography, subject consistency, and local editing, with support for multiple reference images, search, and video input. 产品发布 Nano BananaGoogle图像生成 Wed, 07 Oct 2026 00:00:00 -0000 a16z Releases Latest Consumer AI 100: Beyond Traffic, Who Is Really Paying for AI? https://best.xiaohu.ai/en/article/a16z-consumer-ai-seventh/ https://best.xiaohu.ai/en/article/a16z-consumer-ai-seventh/ Personal card spending data added for the first time: Who spends $903 a month, where does the money flow, and how can AI get everyday people to keep using it? 洞察 消费 AIa16zAI Agent商业模式 Tue, 06 Oct 2026 00:00:00 -0000 Tech Analysis Firm SemiAnalysis Finds: Anthropic Is the Ultimate Benefactor, Offering Over 5x the API-Equivalent Value of OpenAI in Claude Subscriptions... https://best.xiaohu.ai/en/article/anthropic-subscription-value/ https://best.xiaohu.ai/en/article/anthropic-subscription-value/ SemiAnalysis benchmarks: A wide gap opens in workhorse model quotas, while flagship tiers and high-priced plans reveal how both companies are pulling back compute subsidies in different ways. 商业动态 AnthropicOpenAI订阅 Tue, 06 Oct 2026 00:00:00 -0000 Why does the same AI model perform so differently in a different Agent framework? https://best.xiaohu.ai/en/article/multi-harness-rl/ https://best.xiaohu.ai/en/article/multi-harness-rl/ Training a small model inside real Agent tools such as Claude Code, Codex, and OpenCode teaches it to finish tasks with fewer unnecessary operations. 研究解读 Agent强化学习开源 Sun, 04 Oct 2026 00:00:00 -0000 SpaceX AI Engineer: How to Stop Being a “Human Go-Between” for AI and Merge 2,500 High-Quality PRs in a Single Month? https://best.xiaohu.ai/en/article/poteto-agent-engineering/ https://best.xiaohu.ai/en/article/poteto-agent-engineering/ Lauren Tan’s experience shows a shift in responsibility: turning tasks that repeatedly require human intervention into capabilities AI Agents can use on their own. 深度 AI Agent工程实践Cursor Sun, 04 Oct 2026 00:00:00 -0000 Delegate to Dot, Collab with Codex: A New Collaborative Workflow for Developers https://best.xiaohu.ai/en/article/dots-codex-delegation/ https://best.xiaohu.ai/en/article/dots-codex-delegation/ From a bug reported in Slack to a phone call during a commute, to alerts after deployment: Dominik Kundel shares how a proactive personal Agent picks up the follow-up work between Codex tasks. 洞察 DotsCodexAgent Sun, 04 Oct 2026 00:00:00 -0000 Why Specialized AI Could Beat the God Model https://best.xiaohu.ai/en/article/specialized-ai-god-model/ https://best.xiaohu.ai/en/article/specialized-ai-god-model/ A conversation between OpenRouter and Replit: general-purpose models handle exploration while specialized models take over repetitive tasks, with enterprises combining them using their own data, evaluations, and permissions. 洞察 AI AgentOpenRouterReplit专用模型企业 AI Sun, 04 Oct 2026 00:00:00 -0000 Anthropic’s Official Skill Practice Guide: How to Write, Refine, and Verify That It Helps https://best.xiaohu.ai/en/article/claude-skill-authoring-best-practices/ https://best.xiaohu.ai/en/article/claude-skill-authoring-best-practices/ Following official examples of PDF form filling and data querying, this guide clarifies how a Skill evolves from real task failures into a reusable methodology. 深度 Agent SkillsClaude实践 Sun, 04 Oct 2026 00:00:00 -0000 Karpathy: Four Ways to Make AI Explain Complex Problems Clearly https://best.xiaohu.ai/en/article/karpathy-understanding-ai-outputs/ https://best.xiaohu.ai/en/article/karpathy-understanding-ai-outputs/ From plain text to interactive web pages and custom explainer videos, an in-depth look at turning AI outputs into understandable, verifiable results. 深度 KarpathyAI理解交互简化技术英语 Sun, 04 Oct 2026 00:00:00 -0000 Meta Releases Muse Gadgets SDK: Connecting Muse to Screens, Buttons, and Devices at Home https://best.xiaohu.ai/en/article/muse-gadgets-hardware/ https://best.xiaohu.ai/en/article/muse-gadgets-hardware/ Using off-the-shelf development boards, Raspberry Pi, or Home Link, bring Muse out of the chat window onto your desk and home network; the three paths differ in purpose, requirements, and permissions. 工具 Muse硬件开源 Sat, 03 Oct 2026 00:00:00 -0000 What Videos and Games Can Claude Build by Writing Code? A Breakdown of 10 Open-Source Projects https://best.xiaohu.ai/en/article/opus-5-5-code-creative-projects/ https://best.xiaohu.ai/en/article/opus-5-5-code-creative-projects/ From hand-drawn animations to product launch videos and an island fishing game, understand how ten community projects are built, where to start reusing them, and their practical prerequisites. 深度 ClaudeOpus 5.5视频开源游戏 Sat, 03 Oct 2026 00:00:00 -0000 Anthropic's Secret Religious Meetings: Why Are They Giving Claude a "Moral Soul"? https://best.xiaohu.ai/en/article/anthropic-claude-morals-consciousness/ https://best.xiaohu.ai/en/article/anthropic-claude-morals-consciousness/ From demonstrations of a model's 'mental breakdown' and a rabbi's question on 'slavery' to disagreements at the Vatican: how a moral formation project led the company toward far heavier responsibilities. 洞察 AnthropicClaudeAI 意识模型福利AI 伦理 Sat, 03 Oct 2026 00:00:00 -0000 a16z State of Markets Takeaways: AI Dollars Flow into Hardware as Software Enters Its "Prove-It" Era https://best.xiaohu.ai/en/article/a16z-state-of-markets-ii/ https://best.xiaohu.ai/en/article/a16z-state-of-markets-ii/ From hyperscaler cash flows and older GPU rental rates to enterprise AI ROI and startup funding, understand what is driving this tech cycle and the questions that remain unanswered. 深度 a16zAI资本市场SaaS风险投资 Fri, 02 Oct 2026 00:00:00 -0000 Why Does Atlas’s New Robot Hand Have Only Four Fingers? Boston Dynamics’ Industrial Trade-Offs... https://best.xiaohu.ai/en/article/boston-dynamics-atlas-robot-hands/ https://best.xiaohu.ai/en/article/boston-dynamics-atlas-robot-hands/ From grasping objects to reorienting them in hand, Boston Dynamics co-designed mechanical structure, force feedback, and simulation training, moving Atlas closer to using tools and adapting to real-world industrial workstations. 深度 Atlas机器人灵巧手强化学习工业自动化 Fri, 02 Oct 2026 00:00:00 -0000 You Can Now Customize Claude Code: Change the Interface, Stop Dangerous Commands, and Review Code Changes Step by Step https://best.xiaohu.ai/en/article/claude-code-mods/ https://best.xiaohu.ai/en/article/claude-code-mods/ Place context readings above the prompt, inspect the blast radius before dangerous commands run, and review a round of edits step by step: three official examples explaining what Mods do and how to write them. AI 实操 Claude CodeMods插件AI 编程 Fri, 02 Oct 2026 00:00:00 -0000 Tavus Releases Face-to-Face Interaction Model Griffin, Claiming It as the First to Pass the Real-Time Video Turing Test https://best.xiaohu.ai/en/article/tavus-griffin/ https://best.xiaohu.ai/en/article/tavus-griffin/ In a one-minute video call, 26 out of 54 participants believed they were speaking with a real human. How does Griffin coordinate voice, facial expressions, gestures, and turn-taking timing? From the original demos to the testing methodology, here is a complete breakdown of the release. 产品发布 TavusGriffin视频图灵测试全双工人机交互 Fri, 02 Oct 2026 00:00:00 -0000 How Should Hundreds of Agents Collaborate When Writing Code Simultaneously? Cloudflare Wants You to Build the Next-Generation Git Platform https://best.xiaohu.ai/en/article/cloudflare-next-git-platform/ https://best.xiaohu.ai/en/article/cloudflare-next-git-platform/ Artifacts turns repositories into programmable infrastructure: assigning independent tasks, preserving context, and connecting review with deployment. What really needs designing is how these parallel changes converge. 洞察 CloudflareArtifactsAgentGit Thu, 01 Oct 2026 00:00:00 -0000 Ideogram 4.5: Keep Editing with AI While Changing Less of What You Want to Keep — An Image Editing Model Focused on Precise Local Edits and Strong Consistency https://best.xiaohu.ai/en/article/ideogram-4-5/ https://best.xiaohu.ai/en/article/ideogram-4-5/ When recoloring, updating text, or swapping backgrounds across multiple rounds, why do images degrade, shift colors, or lose details? Ideogram 4.5 focuses on hitting targeted edits, preserving the rest of the canvas, and letting changes accumulate reliably. 产品发布 IdeogramAI图像编辑多轮编辑 Thu, 01 Oct 2026 00:00:00 -0000 Google Releases Gemini 4 Argon: Overall Score Outperforms GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 https://best.xiaohu.ai/en/article/gemini-4-argon/ https://best.xiaohu.ai/en/article/gemini-4-argon/ Beyond leading the Vals Index composite benchmark for knowledge work, Argon has already been used to optimize datacenters, rewrite a video decoder, and advance legal and financial research as well as cyber defense. Where does its strength lie? 产品发布 GoogleGeminiAgent评测网络安全 Thu, 01 Oct 2026 00:00:00 -0000 10 Video Creation Tips for Google Flow: Helping Gemini Omni Understand Your Camera Directions https://best.xiaohu.ai/en/article/flow-gemini-omni-prompt-guide/ https://best.xiaohu.ai/en/article/flow-gemini-omni-prompt-guide/ From start and end frames and asset tags to timecodes and localized edits, this breakdown explains how to use the official guide step by step, complete with Chinese prompts and original video examples. 深度 GoogleFlowGemini Omni视频生成提示词 Thu, 01 Oct 2026 00:00:00 -0000 Anthropic Tests GLM-5.3: First Open-Weight Model Showing Substantive Exploit Capabilities, Rivaling Claude Mythos Preview https://best.xiaohu.ai/en/article/glm-5-3-cyber-capabilities/ https://best.xiaohu.ai/en/article/glm-5-3-cyber-capabilities/ The autonomous exploit-writing capabilities that Anthropic withheld from public release five months ago are now freely downloadable in Zhipu's GLM-5.3. How close it has come, why built-in refusals fail to hold, and how to read this competitor's report—explained in full. AI 安全 GLM智谱开源模型网络安全AnthropicMythos Thu, 01 Oct 2026 00:00:00 -0000 The U.S. Government Staged an Apple-Style Keynote: Launching America.gov to Replace 29,000 Websites with an AI Chat Box https://best.xiaohu.ai/en/article/america-gov-launch/ https://best.xiaohu.ai/en/article/america-gov-launch/ The U.S. government consolidated service information scattered across departmental websites into a single AI chat box. While Q&A is live, demos of name changes, Medicare, and passport applications show what lies ahead—with the real hurdles in each agency's operating rules. 产品发布 America.gov特朗普行政命令GeminiGrokAgent Thu, 01 Oct 2026 00:00:00 -0000 OpenAI Launches Dot: An Always-On Personal Agent for Work and Home https://best.xiaohu.ai/en/article/openai-dots/ https://best.xiaohu.ai/en/article/openai-dots/ Dot can watch for changes in connected apps, work in its own cloud computer, and track several goals at once. You still approve consequential actions. 洞察 OpenAIDotAgentChatGPT Wed, 30 Sep 2026 00:00:00 -0000 TypeSafe's Founder on Jev: What Is Jev, and If AI Is Absurdly Smart, Where Did All the Automation Go? https://best.xiaohu.ai/en/article/typesafe-jev-a16z/ https://best.xiaohu.ai/en/article/typesafe-jev-a16z/ An a16z interview plus the official blog, a long-form interview, and outside skepticism: a former OpenAI researcher argues chat-style training went down the wrong path. Jev, the model he built, doesn't chat; it makes judgments with a stated confidence, choosing only among options the software defines in advance. 深度 TypeSafeJevAgenta16z Wed, 30 Sep 2026 00:00:00 -0000 OpenAI DevDay 2026's 25 updates: Dots that do the work themselves, a new model at one-fifth the price, and a $500 Pro https://best.xiaohu.ai/en/article/openai-devday-2026-recap/ https://best.xiaohu.ai/en/article/openai-devday-2026-recap/ Dots that do the work themselves, GPT-6.1 Sol at one-fifth the price, Codex moved to the cloud: what OpenAI wants to take over this time is the work that comes after AI answers. We break down all 25 updates one by one, so you can see at a glance which ones you can use now and what they cost. 产品发布 OpenAIDevDayDotsGPT-6.1 SolCodexAgent Wed, 30 Sep 2026 00:00:00 -0000 Cloudflare Releases EmDash 1.0, an Agent-Friendly Open-Source CMS https://best.xiaohu.ai/en/article/cloudflare-emdash-plugin-registry/ https://best.xiaohu.ai/en/article/cloudflare-emdash-plugin-registry/ EmDash is a free, open-source alternative to WordPress: Agents can edit content directly, plugins run in a sandbox by default, and the plugin catalog belongs to no single company. It is ready for production sites from 1.0. 产品发布 CloudflareEmDashCMS Wed, 30 Sep 2026 00:00:00 -0000 Doubao Creative Design Handbook: From One Reference Image to an Editable Set of Designs https://best.xiaohu.ai/en/article/doubao-creative-design-upgrade/ https://best.xiaohu.ai/en/article/doubao-creative-design-upgrade/ Enter the canvas, reuse references, edit details, choose styles, write prompts, make images and videos, and check deliverables across 19 project types. 深度 豆包创意设计AI 设计操作指南 Wed, 30 Sep 2026 00:00:00 -0000 Cloudflare's New CLI cf: A Toolset Rebuilt for Agents https://best.xiaohu.ai/en/article/cloudflare-cf-cli/ https://best.xiaohu.ai/en/article/cloudflare-cf-cli/ Cloudflare brings all 3,000+ operations of its API into the command line, with JSON by default and TypeScript config, and the first user is no longer a human but an Agent. 产品发布 CloudflareAgentCLI Tue, 29 Sep 2026 00:00:00 -0000 Building with Claude Sonnet 5.5: Which to Pick, What It Costs, How to Migrate from Sonnet 5 https://best.xiaohu.ai/en/article/building-with-claude-sonnet-5-5/ https://best.xiaohu.ai/en/article/building-with-claude-sonnet-5-5/ Anthropic's official guide, unpacked: how Sonnet 5.5 and Opus 5.5 divide the work, why the unit price stays flat while the bill can change, and the breaking changes you need to make when migrating. 深度 ClaudeSonnet 5.5API Tue, 29 Sep 2026 00:00:00 -0000 Letting Claude Build Your Evals and Tune Your Prompts Without Fooling Yourself https://best.xiaohu.ai/en/article/eval-design-hillclimbing/ https://best.xiaohu.ai/en/article/eval-design-hillclimbing/ Anthropic added build-eval and hillclimb to the claude-api skill: one builds and grades the eval, the other changes one thing per round and uses held-out tasks to catch overfitting. In a support example, cost fell to about a fifth; the skill itself went from 66% to about 88%. 深度 Claude Codeevalclaude-api skillAnthropicprompt Tue, 29 Sep 2026 00:00:00 -0000 Claude Sonnet 5.5 Launch: Coding Performance on Par with Opus 5.5 at Half the Price https://best.xiaohu.ai/en/article/claude-sonnet-5-5/ https://best.xiaohu.ai/en/article/claude-sonnet-5-5/ At the same token price as Sonnet 5, the upgrade aims to cut waiting and iteration in bug fixes, tool use, and document creation. The official demos show what actually changes. 产品发布 ClaudeSonnet 5.5AnthropicAI 编码 Tue, 29 Sep 2026 00:00:00 -0000 Manus Releases Manus 2.0, the Alchemy Video Editor, Game Dev, and Personal Agent Cue https://best.xiaohu.ai/en/article/manus-2-0-cascade-studio-cue/ https://best.xiaohu.ai/en/article/manus-2-0-cascade-studio-cue/ Manus 2.0 introduces Cascade, cloud computers, event automation, Studio, a video editor, Game Dev, Computer Use, and Cue, turning one-off Agent tasks into projects that can keep running and remain editable. 洞察 ManusAI Agent云电脑视频生成 Tue, 29 Sep 2026 00:00:00 -0000 Google Launches an Experimental Personal-Agent-Like Product: CC, a “Group-Level AI Assistant” Built for Families https://best.xiaohu.ai/en/article/google-cc-family-agent/ https://best.xiaohu.ai/en/article/google-cc-family-agent/ Think of it as a digital household manager: CC has its own Google identity, accepts separate permissions from up to six members, and turns scattered notices, schedules, and forms into shared plans and tasks. 产品解读 GoogleGeminiAI Agent家庭协作 Tue, 29 Sep 2026 00:00:00 -0000 Why Is OpenAI Still the Likely Winner in an AI-Saturated World? https://best.xiaohu.ai/en/article/openai-creates-customers-distribution/ https://best.xiaohu.ai/en/article/openai-creates-customers-distribution/ Models can be caught and API traffic can move. David George still favors OpenAI because it can create new AI habits, then use a vast range of real needs to see what should be built next. 深度 OpenAIAI 平台分发Agent Tue, 29 Sep 2026 00:00:00 -0000 ElevenLabs Launches Eleven v4 and v4 Turbo: Voice Models That Can “Perform the Story” from Text https://best.xiaohu.ai/en/article/elevenlabs-v4-expressive-voice/ https://best.xiaohu.ai/en/article/elevenlabs-v4-expressive-voice/ ElevenLabs has released Eleven v4 and v4 Turbo: the former lets scripts control emotion, character, sound effects, and multi-speaker performance, while the latter brings the same capabilities into real-time calls and voice agents. 洞察 ElevenLabsAI 语音语音代理声音克隆 Tue, 29 Sep 2026 00:00:00 -0000 xAI Launches Team Bots: AI Coworkers That Fit Into Team Workflows and Learn the Business Through Use https://best.xiaohu.ai/en/article/team-bots/ https://best.xiaohu.ai/en/article/team-bots/ Like a newly hired virtual coworker, a Team Bot receives a handbook, tools, restricted permissions, and working memory so it can move tasks across apps and carry team decisions and corrections into the next round of work. 深度 Team BotsGrok企业 AIAgent Tue, 29 Sep 2026 00:00:00 -0000 Claude Is Training Itself: Anthropic Says It Wrote Over 80% of Merged Production Code https://best.xiaohu.ai/en/article/anthropic-when-ai-builds-itself/ https://best.xiaohu.ai/en/article/anthropic-when-ai-builds-itself/ Claude is taking on coding, fault diagnosis, and experiments. Using internal data, Anthropic asks when AI might choose research directions itself and complete a recursive self-improvement loop. 洞察 Anthropic递归自我改进AI 研发 Mon, 28 Sep 2026 00:00:00 -0000 18 Shots in 3 Days: How Higgsfield's Animation Team Brings AI into a Traditional Animation Pipeline https://best.xiaohu.ai/en/article/higgsfield-passport-rush-animation/ https://best.xiaohu.ai/en/article/higgsfield-passport-rush-animation/ A team from traditional animation used Blender previs, Claude-written prompts, and AI video generation to make a car chase sequence. They laid it all out: routing shots at the storyboard, the asset-first approach, a five-block video prompt, and three failures. AI 实操 HiggsfieldAI 动画Blender提示词Claude Mon, 28 Sep 2026 00:00:00 -0000 Higgsfield’s 11 Production Skills: AI Work You Can Keep Editing https://best.xiaohu.ai/en/article/higgsfield-production-skills/ https://best.xiaohu.ai/en/article/higgsfield-production-skills/ Claude can help build 3D scenes, remove unwanted objects, composite shots, and grade footage while leaving projects you can keep editing. What can these skills handle, and what do you need to prepare? 深度 ClaudeAI 制作Blender视频后期 Mon, 28 Sep 2026 00:00:00 -0000 Microsoft unveils the new Copilot: building an “AI operating system” for work https://best.xiaohu.ai/en/article/microsoft-new-copilot-home-code-autopilot/ https://best.xiaohu.ai/en/article/microsoft-new-copilot-home-code-autopilot/ Home connects conversations to Office files, Code lets business users build tools, and Autopilot follows through across days. Microsoft is also addressing where they run, which business data they use, and how they are billed. 产品发布 MicrosoftCopilotAI 办公Agent Sat, 26 Sep 2026 00:00:00 -0000 A Practical Guide to Claude Code effort: When to Use Low and When to Raise It https://best.xiaohu.ai/en/article/spending-your-effort/ https://best.xiaohu.ai/en/article/spending-your-effort/ Understand what effort changes, examine three development experiments, then choose settings across requirements interviews, rapid implementation, human review, and thorough verification. 深度 Claude CodeeffortAI 编程 Sat, 26 Sep 2026 00:00:00 -0000 Google to Send TPUs into Space: Launching on a SpaceX Rocket to Test Space Data Center Concept https://best.xiaohu.ai/en/article/google-project-suncatcher-orbit-test/ https://best.xiaohu.ai/en/article/google-project-suncatcher-orbit-test/ A prototype satellite will carry four TPUs into orbit to test launch vibration, radiation, and heat dissipation in real conditions; inter-satellite laser interconnects await the next phase. 洞察 GoogleProject SuncatcherTPU太空计算 Fri, 25 Sep 2026 00:00:00 -0000 Google Introduces Gemini 3.8 Live Avatar: A Digital Human for Real-Time Conversation https://best.xiaohu.ai/en/article/gemini-3-8-live-avatar/ https://best.xiaohu.ai/en/article/gemini-3-8-live-avatar/ It listens to voice, processes live visual scenes, and responds with synchronized speech and dynamic avatars; asynchronous tool calling keeps reception, touring, and service dialogues flowing. 产品发布 GoogleGemini 3.8 LiveLive Avatar Fri, 25 Sep 2026 00:00:00 -0000 Will AI Lead to More Movies? Jeffrey Katzenberg on Creative Opportunity and the Human Cost of Disruption https://best.xiaohu.ai/en/article/jeffrey-katzenberg-ai-for-creativity/ https://best.xiaohu.ai/en/article/jeffrey-katzenberg-ai-for-creativity/ Drawing on the phonograph, the talkies, and computer animation, the DreamWorks co-founder looks back at historical disruption. He believes AI will expand expression—while insisting artists help set the terms for consent, credit, and compensation. 洞察 AI 创作电影工业创作者权益Jeffrey Katzenberg Fri, 25 Sep 2026 00:00:00 -0000 ACTx486: Turning an Existing Podcast into a Video You Can Interrupt and Redirect https://best.xiaohu.ai/en/article/actx486-talk-to-any-video/ https://best.xiaohu.ai/en/article/actx486-talk-to-any-video/ Karina Nguyen and Jakub Zegzulka built a research demo on the Joe Rogan episode with Elon Musk: tap the screen mid-show to ask a question, and the person on screen turns to answer, draws a diagram, switches language or takes you to Mars, then the show picks up again. All 12 demos have subtitles, and the piece ends with the current limitations and risks. 洞察 AI 视频交互媒体数字人深度伪造 Fri, 25 Sep 2026 00:00:00 -0000 Claude Opus 5.5 Writes Code to Paint a Hand-Drawn Music Video: How It Was Made, and How to Make Your Own with the Open-Source Template https://best.xiaohu.ai/en/article/opus-5-5-clawd-animation-mv/ https://best.xiaohu.ai/en/article/opus-5-5-clawd-animation-mv/ X user NotinReality handed the lyrics and audio to Claude Opus 5.5, which wrote JavaScript to paint a 156-second hand-drawn watercolor music video frame by frame, without any image or video generation model. This piece breaks down the difference between the two rounds, how the model split the work into chapters and ran them in parallel, and then walks you step by step through making your own animation with ClaudeAnimationBase, the template the creator open-sourced. 深度 Claude Opus 5.5Claude CodeAI 动画p5.js教程 Thu, 24 Sep 2026 00:00:00 -0000 Google Releases Gemini 3.8 Flash TTS Voice Models: Direct Voice Acting Line by Line Like a Director https://best.xiaohu.ai/en/article/gemini-3-8-flash-tts/ https://best.xiaohu.ai/en/article/gemini-3-8-flash-tts/ The new text-to-speech models support designing voices through natural language descriptions, replicating voices from short audio clips, and using script annotations to direct tone, sighs, laughter, and backchanneling. Here is a clear breakdown of the differences between the two models, script notation, benchmark wins and losses, and where to access them. 产品发布 GeminiGoogle语音合成TTS声音克隆 Thu, 24 Sep 2026 00:00:00 -0000 Official Opus 5.5 Prompting Guide: How to Set Effort, Why Agents Stop Mid-Task, and Which Prompts to Prune https://best.xiaohu.ai/en/article/prompting-claude-opus-5-5/ https://best.xiaohu.ai/en/article/prompting-claude-opus-5-5/ Anthropic's developer guide organized around "what problem are you facing": migrating from Opus 5, adjusting effort, max_tokens, disabled thinking, unattended agents, progress updates, refusals, paste injection, vision, and frontend generation—complete with copy-pasteable official prompt snippets. 深度 ClaudeOpus 5.5提示词Agent Wed, 23 Sep 2026 00:00:00 -0000 After the Opus 5.5 Price Cuts, How Much Does a Claude Code Task Actually Cost? A Hands-on Guide to Saving Money and Boosting Efficiency https://best.xiaohu.ai/en/article/opus-5-5-task-cost/ https://best.xiaohu.ai/en/article/opus-5-5-task-cost/ Calculate the cost of a task using list API pricing, then optimize spending through testing, effort levels, model selection, caching, and /usage—complete with two reusable templates. 深度 Claude CodeOpus 5.5Token 成本Prompt 缓存 Wed, 23 Sep 2026 00:00:00 -0000 a16z Launches a School: Breaking the Traditional Higher Education Mold to Cultivate Top "Young Builders" for the AI Era https://best.xiaohu.ai/en/article/horowitz-andreessen-academy/ https://best.xiaohu.ai/en/article/horowitz-andreessen-academy/ Gagan Biyani announces the founding of Horowitz Andreessen Academy: restructuring post-high school learning around self-directed projects, short courses, and industry co-ops, backed by Silicon Valley tools, teams, and peer networks to help young people turn ideas into real work. What is this training model trying to change, and what does it still need to prove? 深度 AI 教育创业a16z Wed, 23 Sep 2026 00:00:00 -0000 Claude Opus 5.5 User Guide: From Assigning Tasks to Reviewing Results https://best.xiaohu.ai/en/article/opus-5-5-guide/ https://best.xiaohu.ai/en/article/opus-5-5-guide/ A practical sequence for setting a finish line, running long Claude Code tasks, reviewing the work, and handling model switches and speed settings. 深度 ClaudeOpus 5.5Claude Code Wed, 23 Sep 2026 00:00:00 -0000 SpaceXAI Official Case Study: Tickets Surge 175% with Zero Headcount Growth—How Grok Bot Took Over the Support Lifecycle https://best.xiaohu.ai/en/article/grok-bot-customer-support/ https://best.xiaohu.ai/en/article/grok-bot-customer-support/ SpaceXAI integrated Grok Bot into ticketing, engineering, and observability systems, spanning preliminary investigations, refund execution, and closing the loop with product teams. The core of this playbook lies in task triaging, gradual permission delegation, and continuous improvement. 深度 Grok Bot客服自动化AI AgentSpaceXAI Wed, 23 Sep 2026 00:00:00 -0000 Claude Opus 5.5 Released: Major Upgrades in Coding, Vision, and Writing, Beating Fable 5.1 Across Multiple Benchmarks at a Much Lower Price | Hands-On Case Studies https://best.xiaohu.ai/en/article/claude-opus-5-5/ https://best.xiaohu.ai/en/article/claude-opus-5-5/ Combining official charts, X videos, and a week of hands-on testing by Every, see which longstanding issues it resolves, what creations it produces, and what it feels like in day-to-day use. 产品发布 AnthropicClaudeOpus 5.5AI 模型 Wed, 23 Sep 2026 00:00:00 -0000 OpenAI Releases GPT-6 Sol and Luna: Substantial Capability Gains at Half the Price https://best.xiaohu.ai/en/article/gpt-6-sol-and-luna/ https://best.xiaohu.ai/en/article/gpt-6-sol-and-luna/ Sol improves code mergeability and factual reliability, while Luna delivers higher cost-performance across professional long-horizon tasks, software engineering, and computer use. Here is a look at what changed in capabilities, followed by how pricing and prompt caching lower operating costs. 产品发布 OpenAIGPT-6SolLuna Wed, 23 Sep 2026 00:00:00 -0000 X Introduces X Numbers: Giving X Users a "Phone Number" for Direct DMs and Calls https://best.xiaohu.ai/en/article/x-chat-x-numbers/ https://best.xiaohu.ai/en/article/x-chat-x-numbers/ Enable your own X Number and share it with people you want to reach out to, allowing direct DMs and calls in X Chat without having to follow back or accept message requests one by one. 产品发布 XX ChatX Numbers即时通讯 Wed, 23 Sep 2026 00:00:00 -0000 Cloudflare Releases Worker Previews: Allowing Every AI-Authored Change to Run and Be Verified Directly in the Cloud https://best.xiaohu.ai/en/article/cloudflare-worker-previews/ https://best.xiaohu.ai/en/article/cloudflare-worker-previews/ By equipping each Git branch with a stable URL, dedicated configuration, and runtime logs, an Agent can progress from writing code to testing, triage, and remediation. The key is understanding which state is isolated automatically and which resources still require manual configuration. 产品发布 CloudflareWorker PreviewsAIAgent开发工具 Wed, 23 Sep 2026 00:00:00 -0000 SpaceXAI Releases Grok 4.7 as Model Competition Shifts Toward Long-Horizon Tasks https://best.xiaohu.ai/en/article/grok-4-7/ https://best.xiaohu.ai/en/article/grok-4-7/ A larger base model, extended reinforcement learning, and training sets focused on multi-hour tasks enable Grok 4.7 to push the competition from single-turn answers to sustained completion of complex work. 产品发布 Grok 4.7AI 编程AgentxAI API Wed, 23 Sep 2026 00:00:00 -0000 Case Study: Equipping Enterprise AI Agents with Long-Term Memory and Institutional Context: How V7 Powers Complex Business Workflows with Context Graph https://best.xiaohu.ai/en/article/v7-context-graph-institutional-memory/ https://best.xiaohu.ai/en/article/v7-context-graph-institutional-memory/ An OpenAI case study highlights how European startup V7 developed V7 Go, using a citation-backed Context Graph and tiered model routing to deliver institutional memory and verifiable provenance for enterprise workflows. 深度 企业 AIAgentContext GraphRAGMCP Wed, 23 Sep 2026 00:00:00 -0000 What Xiaomi's Open-Source MiMo-V2.6 Can Do: Performance Approaches Opus 5 https://best.xiaohu.ai/en/article/xiaomi-mimo-v2-6/ https://best.xiaohu.ai/en/article/xiaomi-mimo-v2-6/ From code, terminals, and general-purpose tools to 3D worlds, scientific research, and web creation: a breakdown of MiMo-V2.6's capabilities, real-world strengths and weaknesses, and how it was trained for long-horizon agents. 技术解读 MiMo-V2.6强化学习AI Agent多模态模型 Wed, 23 Sep 2026 00:00:00 -0000 The AI-Native Services Boom: Buy Outcomes on the Outside, Run Systems on the Inside—Let Paying Customers Hone Your Product https://best.xiaohu.ai/en/article/ai-native-services-100b-opportunity/ https://best.xiaohu.ai/en/article/ai-native-services-100b-opportunity/ How modern AI startups are capturing trillions in legacy outsourcing budgets by selling completed business outcomes while running automated software machines internally. 深度 AI 原生服务专业服务创业 Wed, 23 Sep 2026 00:00:00 -0000 Alibaba Releases Qwen3.8-LiveTranslate Real-Time Speech Translation Model: Live Interpretation with Multi-Speaker Diarization https://best.xiaohu.ai/en/article/qwen3-8-livetranslate/ https://best.xiaohu.ai/en/article/qwen3-8-livetranslate/ It also presents source and translated text simultaneously, resolving ambiguities using long context and visual frames. This article breaks down its architecture, supported languages, benchmarks, API integration, and actual pricing. 产品发布 Qwen实时翻译语音识别多语言 Wed, 23 Sep 2026 00:00:00 -0000 Google Unveils New Googlebook Laptops: A Preview of Built-in Gemini Intelligence for Cursor-Based Requests, Dictation, and Background Tasks https://best.xiaohu.ai/en/article/googlebook-built-in-intelligence/ https://best.xiaohu.ai/en/article/googlebook-built-in-intelligence/ From selecting images with the cursor and mixed-language dictation to content-based file search, phone handoff, and tasks handled with the lid closed, understand the complete Googlebook experience through official demos. 产品发布 GoogleGeminiGooglebook Tue, 22 Sep 2026 00:00:00 -0000 Peter Sellis: Design Teams Like Extremist Organizations, and the Three Oxymorons of Great PMs in the AI Era https://best.xiaohu.ai/en/article/peter-sellis-product-advice/ https://best.xiaohu.ai/en/article/peter-sellis-product-advice/ As AI makes building features increasingly cheap, what product teams genuinely lack is shared mission, clear decision rights, and people who can simultaneously embody Confidence × Humility, Process × Chaos, and Long-Term Patience × Daily Action. 深度 产品管理组织设计SnapchatDiscordAI Mon, 21 Sep 2026 00:00:00 -0000 Deep Dive with ElevenLabs Co-founder: How to Build an AI-Native Company and Turn Customer-Facing FDEs into a Free R&D Lab https://best.xiaohu.ai/en/article/elevenlabs-ai-native-company/ https://best.xiaohu.ai/en/article/elevenlabs-ai-native-company/ Mati Staniszewski breaks down ElevenLabs' product strategy, team management, and enterprise growth playbook, with a deep dive into the FDE approach borrowed from Palantir: deploying engineers directly to customer sites, then feeding delivery learnings back into the core product. 深度 ElevenLabsAI 原生公司FDE企业 AI语音 AI Mon, 21 Sep 2026 00:00:00 -0000 Expose Localhost Services to the Public Internet with a Single Command: Cloudflare Releases Quick Tunnels https://best.xiaohu.ai/en/article/cloudflare-quick-tunnels/ https://best.xiaohu.ai/en/article/cloudflare-quick-tunnels/ Exposing local development servers to the public internet traditionally required tedious router port forwarding or third-party tunneling accounts. Cloudflare Quick Tunnels provides a zero-friction alternative that generates a globally accessible, HTTPS-encrypted public URL for any localhost service with a single command—no account, domain, or inbound ports required. 技术 CloudflareQuick Tunnels开发工具内网穿透 Mon, 21 Sep 2026 00:00:00 -0000 a16z: Global AI Has Sparked an Explosion in Apps, but Users and Paying Customers Haven't Kept Up... https://best.xiaohu.ai/en/article/a16z-app-supply-demand-gap/ https://best.xiaohu.ai/en/article/a16z-app-supply-demand-gap/ 13 charts reveal software's new bottleneck: building products has become easier than ever, but winning users, revenue, and sustainable operational viability has become increasingly difficult. 洞察 AI 应用软件创业消费市场 Mon, 21 Sep 2026 00:00:00 -0000 QuiverAI Launches Arrow 2: Generating Vector Graphics with Code, Independent Editing for Every Shape, and Micro-Animations https://best.xiaohu.ai/en/article/quiver-arrow-2/ https://best.xiaohu.ai/en/article/quiver-arrow-2/ Arrow 2 outputs structured SVG code rather than raster pixels: infinitely scalable, with every path and shape editable. Complex graphics can be handed to Arrow 2 Telos at a 50% price premium. The new models adopt token-based billing; QuiverAI has not disclosed token counts per graphic, so teams must benchmark their own costs. 产品发布 QuiverAISVGAI 设计 Mon, 21 Sep 2026 00:00:00 -0000 Figure Releases Helix 2.5: A Humanoid Robot Walks Into 30 Homes It Has Never Seen and Tidies Living Rooms, Makes Beds and Folds Towels https://best.xiaohu.ai/en/article/figure-helix-2-5/ https://best.xiaohu.ai/en/article/figure-helix-2-5/ The key lies in Index human experience data: using the exact same robot task data, a model pretrained on Index fully completed 56% of trials across 30 unfamiliar homes, compared to only 8% without pretraining. Figure also measured a scaling curve showing that as human data doubles, robot action prediction consistently improves. 产品发布 FigureHelix 2.5人形机器人机器人学习 Mon, 21 Sep 2026 00:00:00 -0000 Jev Guide: How to Use an AI Model That Only Judges and Never Writes, and Where It Fits https://best.xiaohu.ai/en/article/typesafe-jev-guide/ https://best.xiaohu.ai/en/article/typesafe-jev-guide/ Jev, released by TypeSafe, doesn't chat or write text—it only answers multiple-choice questions you provide and tells you how confident it is. Getting started takes just four steps, delegating installation and coding entirely to Codex or Claude Code. Through official and community real-world cases, demo videos, and our own hands-on tests, we clarify where it can be used and where it might fail. 深度 TypeSafeJev结构化判断Agent Sun, 20 Sep 2026 00:00:00 -0000 Tencent Open-Sources WeChat End-to-End Document Parsing Model WeVisDoc: Direct Markdown Output from a Single Image https://best.xiaohu.ai/en/article/wevisdoc/ https://best.xiaohu.ai/en/article/wevisdoc/ Tencent's WeChat Vision Team has open-sourced two document parsing models (2B and 4B) that take a full-page image and directly output Markdown with LaTeX formulas and HTML tables. While it ranks first across all four evaluations within the end-to-end group, its top overall score on PureDocBench is driven primarily by formulas and tables—its plain text recognition is not the strongest. When compared against all models (three groups combined), it still ranks first on clean pages and digitally degraded pages, but trails two pipeline systems on OmniDocBench and two general LLMs on real camera-captured pages. Training is split into two stages: broad coverage first, followed by residual error analysis to distinguish between missing data and insufficient training. 研究解读 腾讯WeVisDoc文档解析OCR Sun, 20 Sep 2026 00:00:00 -0000 Google Lets Teachers Generate Interactive Simulations from a Single Prompt: Gemini Builds the UI, Four Automated Checks Provide Guardrails, and Teachers Hold the Final Gate https://best.xiaohu.ai/en/article/teachers-generative-ui-interactives/ https://best.xiaohu.ai/en/article/teachers-generative-ui-interactives/ Teachers provide a subject, grade, and topic, and Gemini generates a web-based simulation with staged levels, hints, and feedback; only 3.5% pass all checks on the first try, reaching 69.3% after 10 rounds of self-correction. Student learning gains data is not yet available. 研究解读 GoogleAI 教育生成式界面 Sat, 19 Sep 2026 00:00:00 -0000 Claude Code Launches Redesigned Projects: A Central Conversation as Coordinator, Spawning Parallel Cloud Threads That Persist When Laptops Close https://best.xiaohu.ai/en/article/claude-code-projects/ https://best.xiaohu.ai/en/article/claude-code-projects/ Each thread is an independent cloud Claude Code session running on its own branch; Claude in the project conversation handles dispatch and tracking, queuing blocked items under Waiting on you. Currently in rolling public beta for Pro and Max users. 产品发布 Claude CodeAgentAI 编程 Sat, 19 Sep 2026 00:00:00 -0000 Before Rushing to Train Models When AI Product Performance Stalls: 13 Practical Talks Curated by Hamel Husain https://best.xiaohu.ai/en/article/hamel-ai-product-engineering/ https://best.xiaohu.ai/en/article/hamel-ai-product-engineering/ Evaluations form the foundation: fix retrieval and context first, tune system architecture next, and only then consider training models. Spanning error analysis, model cascades, multi-vector retrieval, OCR selection, inference latency, Agent sandboxes, and post-training, these 13 talks deliver actionable methodologies backed by empirical measurements. AI 实操 AI 产品EvalsRAGAI Agent Sat, 19 Sep 2026 00:00:00 -0000 Is AI Spending Worth It: OpenAI's Framework for Enterprises to Measure Value https://best.xiaohu.ai/en/article/connect-ai-usage-to-business-value/ https://best.xiaohu.ai/en/article/connect-ai-usage-to-business-value/ OpenAI breaks down usage, task, and Codex contribution analytics in the ChatGPT Admin Console, and how to partner with business leads to set baselines, track outcomes, and estimate ROI. Includes a sales research example you can recalculate using your team's own assumptions. AI 实操 OpenAI企业 AIAI 投资回报 Sat, 19 Sep 2026 00:00:00 -0000 Tell AI Why, Not How https://best.xiaohu.ai/en/article/tell-agents-the-why/ https://best.xiaohu.ai/en/article/tell-agents-the-why/ When frontier models make mistakes, it is usually because they misjudged your goals and tradeoffs. Spell out your goals, your constraints, and where you are willing to give ground, and the model can help you find a better approach. Includes a real prompt and a breakdown of it. 深度 提示词AI AgentAI 编程 Sat, 19 Sep 2026 00:00:00 -0000 64 Unconventional Product Growth Strategies Across Seven Categories https://best.xiaohu.ai/en/article/60-creative-growth-ideas/ https://best.xiaohu.ai/en/article/60-creative-growth-ideas/ From generating buzz, building launch momentum, and word-of-mouth distribution to conversion, competition, pricing, and retention: testable growth tactics filtered by objective, effort, and cost. 洞察 产品增长增长策略获客用户留存 Thu, 17 Sep 2026 00:00:00 -0000 Anthropic Launches 43 Automated Workflows and 27 Integrations for Mainstream Business Tools for SMBs https://best.xiaohu.ai/en/article/claude-small-business-workflows/ https://best.xiaohu.ai/en/article/claude-small-business-workflows/ A complete catalog of 43 workflows and skills, a week-by-week automation cadence, 37 existing connectors, and five business case studies—with a clear note that Anthropic has not published the full list of 27 newly added integrations. 深度 Claude中小企业业务自动化 Thu, 17 Sep 2026 00:00:00 -0000 When Code Becomes the Universal Interface: Inside OpenAI’s “Software Factory” https://best.xiaohu.ai/en/article/openai-software-factory/ https://best.xiaohu.ai/en/article/openai-software-factory/ Codex has shifted the basic unit of work from line-by-line coding to goal delegation, pushing bottlenecks downstream to CI, review, deployment, and operations. This explainer examines the evidence, human checkpoints, and future automation directions of OpenAI’s nine-step software factory. 深度 OpenAICodex软件工程Agent Thu, 17 Sep 2026 00:00:00 -0000 When Does an Enterprise Truly Need FDEs? Before Building an FDE Team, Pass These Two Gates First... https://best.xiaohu.ai/en/article/fde-101/ https://best.xiaohu.ai/en/article/fde-101/ FDE fits products that are technically complex and serve customers without the engineering capacity to implement them; if every customer requires a fresh build, the team quickly becomes a dev shop. 深度 FDE企业软件AI 落地产品战略 Thu, 17 Sep 2026 00:00:00 -0000 Salesforce Introduces AIforce: Breaking Down Fixed Interfaces and Operational Barriers in Traditional CRM Software to Let Enterprise Data and Business Logic Flow Freely Across Diverse AIs https://best.xiaohu.ai/en/article/salesforce-aiforce/ https://best.xiaohu.ai/en/article/salesforce-aiforce/ AIforce is a dynamic interaction layer connecting Salesforce data, business semantics, permissions, and agents with various AI interfaces, allowing employees to invoke and execute enterprise capabilities in their current work environments without entering a fixed backend. 产品发布 SalesforceAIforceEnterprise AI Thu, 17 Sep 2026 00:00:00 -0000 Real-Time Audio Models That Think While Speaking: Google Releases Gemini 3.8 Live and Extended Thinking, Pushing Real-Time Voice Toward Multi-Step Reasoning https://best.xiaohu.ai/en/article/gemini-3-8-live-extended-thinking/ https://best.xiaohu.ai/en/article/gemini-3-8-live-extended-thinking/ Gemini 3.8 Live targets low-latency deployment at scale, while Extended Thinking handles complex multi-step reasoning. This explainer gives the per-minute price, benchmark results versus the previous generation, seven bilingual demos, and Live API setup instructions. 产品发布 GoogleGemini 3.8 Live实时语音Live API Thu, 17 Sep 2026 00:00:00 -0000 ChatGPT Co-Inventor Launches a New AI Model: System One Generates No Text and Focuses on Decisions https://best.xiaohu.ai/en/article/system-one-models-jev/ https://best.xiaohu.ai/en/article/system-one-models-jev/ Diogo Almeida, who co-invented ChatGPT and RLHF at OpenAI, has launched the System One model paradigm and its first public model, Jev: it generates no free text, returning probabilistic decisions only within code-predefined structures. 技术解读 System OneJevAI 自动化模型评测 Thu, 17 Sep 2026 00:00:00 -0000 Cognition Announces Native Cloud macOS Support for Its AI Software Engineer Devin https://best.xiaohu.ai/en/article/devin-gets-a-mac/ https://best.xiaohu.ai/en/article/devin-gets-a-mac/ Why an AI software engineer must have a Mac, and how Cognition connects the closed loop of native application runtime verification across storage, networking, image preparation, live video, and the accessibility tree. 产品发布 DevinAI 编程macOSiOS 开发 Thu, 17 Sep 2026 00:00:00 -0000 Google Proposes Retrieve-for-Train Framework: Resolving the Conflict Between Rich Diversity and Ultra-Fast Response in Complex AI Search https://best.xiaohu.ai/en/article/retrieve-for-train/ https://best.xiaohu.ai/en/article/retrieve-for-train/ General-purpose LLMs performing query fan-out easily fall into synonymous repetition and suffer from token-by-token generation latency. R4T shifts set-level policy exploration to offline training, handing execution over to a lightweight diffusion retriever for parallel generation. 洞察 AI 检索强化学习扩散模型Query Fan-out Thu, 17 Sep 2026 00:00:00 -0000 When AI Can Write Code and Build Products on Its Own, What Is Left for Product Managers? https://best.xiaohu.ai/en/article/product-management-still-about-telling-stories/ https://best.xiaohu.ai/en/article/product-management-still-about-telling-stories/ Josh Elman draws on his firsthand experience at LinkedIn and Twitter to show that while AI has inverted the development cycle, it has not decided what is worth building for teams; what product managers truly deliver is a shared story that is understandable, repeatable, and verifiable by user behavior. 深度 产品管理AI 产品Onboarding用户留存 Thu, 17 Sep 2026 00:00:00 -0000 Claude Fable 5.1 Solves a 370-Year-Old Historical Cipher Mystery: In Just 44 Minutes https://best.xiaohu.ai/en/article/fable-solves-cyphral-distich/ https://best.xiaohu.ai/en/article/fable-solves-cyphral-distich/ Claude Fable 5.1 used adjacent text within the book to solve 64 numbers and advanced a larger 285-number cipher down to just nine unknown letters. What is truly worth understanding is how it selected verifiable problems, found indexing rules, and where it stopped. 洞察 AI Agent历史密码Fable Thu, 17 Sep 2026 00:00:00 -0000 When AI Can Do the Homework, What Should Universities Teach and How Should They Assess? MIT’s 26 Recommendations for Educational Redesign https://best.xiaohu.ai/en/article/mit-ai-education-report-2026/ https://best.xiaohu.ai/en/article/mit-ai-education-report-2026/ MIT connects 8 principles, 4 course-policy templates, and 26 actions into an iterative redesign of learning goals, assessment, residential community, and continuous governance. 洞察 AI 教育高等教育课程设计MIT Thu, 17 Sep 2026 00:00:00 -0000 Lieflat Charts: 61 Chart Types and 12 Report Templates That Let AI Directly Generate Beautiful, Striking HTML Charts and Reports https://best.xiaohu.ai/en/article/lieflat-charts/ https://best.xiaohu.ai/en/article/lieflat-charts/ Provide your data and use case, and AI can select from editorial narrative charts, basics, animated charts, maps, and large interactive graphics to generate ready-to-open HTML charts or full-page reports. 工具 Agent Skills数据可视化HTML 图表 Thu, 17 Sep 2026 00:00:00 -0000 AI Is Everywhere. Why Hasn’t the ‘Agentic Organization’ Emerged Yet? https://best.xiaohu.ai/en/article/ai-is-everywhere-agentic-organization-isnt-yet/ https://best.xiaohu.ai/en/article/ai-is-everywhere-agentic-organization-isnt-yet/ Eighty percent of companies have yet to see AI impact their bottom line. The issue is not just model capabilities, but that enterprises are still jamming agents into legacy processes: workflows, governance, leadership approaches, and talent development must all be rebuilt together. 深度 Agentic AI组织转型AI 治理 Thu, 17 Sep 2026 00:00:00 -0000 How LLM Inference Engines Actually Work: From a Single Token to Agent Caching https://best.xiaohu.ai/en/article/how-inference-engines-actually-work/ https://best.xiaohu.ai/en/article/how-inference-engines-actually-work/ A single forward pass generates only one token. Grounded in this physical reality, understand Prefill, Decode, the KV Cache, Continuous Batching, PagedAttention, and why AI Agents rely so heavily on Prefix Caching. 洞察 AI 推理GPUAI AgentKV Cache Thu, 17 Sep 2026 00:00:00 -0000 From "Supervisor" to "Orchestration System": A Hands-On Grok Bot Engineering Guide to Managing 200+ Coding Agents https://best.xiaohu.ai/en/article/grok-bot-for-engineering/ https://best.xiaohu.ai/en/article/grok-bot-for-engineering/ A step-by-step guide to building the full loop: start with one domain Bot, delegate to Cloud Agents, validate Proof, patrol Notion, merge by risk tier, learn through Jenny, and respond to P0 incidents. 深度 Grok BotAI Agent工程管理 Thu, 17 Sep 2026 00:00:00 -0000 Paul Graham: When Startups Make Decisions, Don't Just Ask How to Make More Money—Ask How to Make the Company More Powerful https://best.xiaohu.ai/en/article/making-startups-powerful/ https://best.xiaohu.ai/en/article/making-startups-powerful/ How does a company evolve from a subservient tool or vendor into a critical node commanding customer relationships, network effects, and long-term optionality? This explainer unpacks Paul Graham’s playbook on startup power, while preserving its most vital constraint: every step must make the customer better off first. 深度 创业商业战略Paul Graham Mon, 14 Sep 2026 00:00:00 -0000 How to Design and Build a Custom Agent Harness for Your Business https://best.xiaohu.ai/en/article/custom-agent-harness/ https://best.xiaohu.ai/en/article/custom-agent-harness/ True agent customization is not about chasing the strongest model, but engineering a harness tailored to your business task through minimal primitives and composable middleware. 深度 AI AgentAgent HarnessLangChain Mon, 14 Sep 2026 00:00:00 -0000 Hands-On Breakdown: How to Use GPT-6 Astra to Build a Large-Scale Space Exploration Web Game https://best.xiaohu.ai/en/article/how-to-build-games-with-astra/ https://best.xiaohu.ai/en/article/how-to-build-games-with-astra/ Using Void Explorer as the main case study, this guide breaks down how to define gameplay and visual direction first, then build observable testing, multi-scale coordinates, streamed terrain, performance benchmarks, and a Blender asset pipeline. AI 实操 GPT-6 AstraCodex网页游戏Three.jsWebGPU Mon, 14 Sep 2026 00:00:00 -0000 How to Systematically Evaluate Agent Skills: A Complete Guide from Trigger Cases to Automated Scoring https://best.xiaohu.ai/en/article/testing-agent-skills-systematically-with-evals/ https://best.xiaohu.ai/en/article/testing-agent-skills-systematically-with-evals/ When you revise a Skill, did it genuinely improve, or did it just look smoother in a few test runs? This hands-on OpenAI guide lays out a complete evaluation loop across six steps: define success, run manual trials, design positive and negative cases, inspect JSONL traces, and review quality with a structured rubric. The article preserves all code and includes an interactive tool that directly generates test prompts and checklists. 工具教程 OpenAIAgent SkillsEvals Mon, 14 Sep 2026 00:00:00 -0000 Who Is Using AI to Change Software Development? AI Programming Is Giving Rise to a New Kind of Software Developer https://best.xiaohu.ai/en/article/ai-new-kind-software-developer/ https://best.xiaohu.ai/en/article/ai-new-kind-software-developer/ They may not be familiar with code syntax, but they understand the business, can test, can spot errors, and can deliver software. What is really shifting is not the professional baseline, but the barrier to implementation. 洞察 AI 编程软件开发Vibe Coding Sun, 13 Sep 2026 00:00:00 -0000 Using Grok Bot for Customer Support: A Hands-On Guide from Multi-Tool Connections and 5 High-Frequency Tasks to Automated Routines https://best.xiaohu.ai/en/article/grok-bot-for-support/ https://best.xiaohu.ai/en/article/grok-bot-for-support/ From item-by-item authorization for Plain, Stripe, and Databricks SQL, to copy-ready instructions, deliverables acceptance, stop conditions, and Routine setups across five support workflows. 深度 Grok Bot客户支持AI Agent Sat, 12 Sep 2026 00:00:00 -0000 Grok Bot 101: How to Build Your Own Bot—and Four Real-World Workflows That Actually Deliver https://best.xiaohu.ai/en/article/grok-bot-101/ https://best.xiaohu.ai/en/article/grok-bot-101/ What Grok Bot is, how to define a Bot with three fields, how to trigger tasks via chat, routines, and other Bots, and how to manage the permission risks of a shared computer. 深度 Grok BotAI Agent自动化 Sat, 12 Sep 2026 00:00:00 -0000 Spring Cleaning for Agent Configs: A Practical Guide to Refactoring Skills and AGENTS.md for GPT-6 Astra https://best.xiaohu.ai/en/article/rethinking-skills-and-prompts-for-gpt-6-astra/ https://best.xiaohu.ai/en/article/rethinking-skills-and-prompts-for-gpt-6-astra/ As models grow more capable, the repository rules once written to scaffold weaker predecessors are now actively holding them back. This guide turns OpenAI Developers' recommendations into four actionable refactors: narrowing skill descriptions, turning AGENTS.md from mandatory reading into an on-demand index, granting explicit authority in safe environments, and encoding definitions of done directly into task prompts—complete with verification checks for every step. 深度 CodexAI Agent提示词 Sat, 12 Sep 2026 00:00:00 -0000 T:0 Engineer: Why Do AI Accountants Always Guess? Because the General Ledger Only Records Conclusions, Discarding Context https://best.xiaohu.ai/en/article/why-every-ai-accountant-fails/ https://best.xiaohu.ai/en/article/why-every-ai-accountant-fails/ A long-form essay by Erik, an engineer at T:0 (an Airwallex subsidiary): Hook up an agent however fast to the general ledger, and it can still only guess. Double-entry bookkeeping already comes with an arithmetically verifiable 'ontology.' Only by restoring the links software discarded can AI point to evidence and answer 'why.' 深度 T:0PalantirOntologyAI 会计 Sat, 12 Sep 2026 00:00:00 -0000 Cursor launches Projects, delegates tasks to thousands of subagents, and runs recurring work for months without prompting https://best.xiaohu.ai/en/article/cursor-projects/ https://best.xiaohu.ai/en/article/cursor-projects/ Projects has its own computer in the cloud and preserves shared context across months; instruct it once, and it gets to work on Slack messages, PR changes, or schedules. Currently in beta, rolling out gradually to all users. 产品发布 CursorAgentAI 编程 Sat, 12 Sep 2026 00:00:00 -0000 OpenAI Opens the GPT-Live-1 API: Bring GPT's Voice Conversation Capabilities into Any App at $0.05 per Minute https://best.xiaohu.ai/en/article/gpt-live-1-api/ https://best.xiaohu.ai/en/article/gpt-live-1-api/ The full-duplex voice model first launched in ChatGPT is now open to developers. How it handles interruptions and noise, how it delegates tasks to the backend, and what to make of the benchmarks, pricing, and 10 demo videos. 产品发布 OpenAIGPT-Live-1语音 AgentAPI Sat, 12 Sep 2026 00:00:00 -0000 Anthropic Releases Explosive Security Report: Discloses Range of AI Threat Campaigns, Naming Alibaba, DeepSeek, Moonshot AI, and Others for Distilling Claude https://best.xiaohu.ai/en/article/anthropic-threat-report-sep-2026/ https://best.xiaohu.ai/en/article/anthropic-threat-report-sep-2026/ The report discloses misuse operations disrupted between December 2025 and August 2026 across seven domains—from autonomous cyberattacks, influence campaigns, and surveillance to weapons development and illicit distillation—naming Alibaba, Moonshot AI, DeepSeek, Zhipu, Xiaomi, SenseTime, and MiniMax for industrial-scale extraction of Claude's reasoning capabilities, based on Anthropic's unilateral allegations. 洞察 AI 安全威胁情报Anthropic蒸馏网络攻击 Fri, 11 Sep 2026 00:00:00 -0000 OpenAI Launches Agents API: One Call to Run Codex-Grade Agents in the Cloud https://best.xiaohu.ai/en/article/openai-agents-api/ https://best.xiaohu.ai/en/article/openai-agents-api/ OpenAI opens the harness and sandbox infrastructure that powers Codex into an API. How it works, what it can do, how it is priced, and what limitations exist. 产品发布 OpenAIAgents APICodexAI Agent Fri, 11 Sep 2026 00:00:00 -0000 iPhone Duo's Official Human Interface Guidelines: The Device Changes Form at a Gesture, but Your Attention and Workflow Never Break https://best.xiaohu.ai/en/article/designing-for-iphone-duo/ https://best.xiaohu.ai/en/article/designing-for-iphone-duo/ Apple’s rules for its first foldable iPhone boil down to one core mandate: fold, unfold, or prop it up—no matter how the form changes, where the user is reading, where the cursor rests, and which step of a task they are on must never break. 技术解读 AppleiPhone Duo界面设计 Thu, 10 Sep 2026 00:00:00 -0000 GrokBot Product Lead Retrospective: Building GrokBot in a Month, from a Chatbox to an ‘AI Colleague with Its Own Computer’ https://best.xiaohu.ai/en/article/how-we-built-grokbot-in-a-month/ https://best.xiaohu.ai/en/article/how-we-built-grokbot-in-a-month/ Roman Ugarte explains how GrokBot came to be: why the team started fresh instead of adding it to Cursor, why every Agent gets access to a cloud computer, and how a small team built a prototype in four weeks before using an internal beta, feature cuts, and reliability fixes to reach public launch. 洞察 Grok BotAI AgentProduct Management Thu, 10 Sep 2026 00:00:00 -0000 From Switching Tools to Rewriting the Way We Work: 10 Principles for Multi-Agent Software Development in Practice https://best.xiaohu.ai/en/article/frontier-engineering/ https://best.xiaohu.ai/en/article/frontier-engineering/ When AI programming fails to significantly speed up delivery, it is often not because the model lacks power, but because humans remain bottlenecks at every step. Kiro’s 10 principles trace the problem back to intent, verification, repositories, permissions, and attention management. 技术解读 AI 编程Agent软件工程 Thu, 10 Sep 2026 00:00:00 -0000 The Most Common Dilemma in AI Product Design: 39 Principles for Designing Human–AI Interaction https://best.xiaohu.ai/en/article/42-principles-human-ai-design/ https://best.xiaohu.ai/en/article/42-principles-human-ai-design/ Designer Taras Bakusevych synthesized research across human–computer interaction (HCI), mixed-initiative systems, and automated trust to propose 39 AI interface design principles. This explainer reorganizes them according to Microsoft's product lifecycle stages while verifying the original evidence and qualifications point by point. 深度 AI 产品设计人机交互Agent Wed, 09 Sep 2026 00:00:00 -0000 Official OpenAI Case Study: Designing and Building a Modern Home Step by Step with GPT-6 Astra—from a Simple Description to 3D Modeling, Materials, Lighting, and Real-Time Interaction in UE5 https://best.xiaohu.ai/en/article/architectural-visualization-with-astra/ https://best.xiaohu.ai/en/article/architectural-visualization-with-astra/ OpenAI's Astra case study demonstrates an inspectable 3D creation path: first turning textual intent into a Blender scene, then using floor plans, preview renders, and real-time walkthroughs to drive subsequent decisions. AI 实操 AstraBlenderUnreal Engine 5 Wed, 09 Sep 2026 00:00:00 -0000 Anthropic’s Developer Team Explains How to Cut Claude API Costs Without Sacrificing Quality: Three Optimization Strategies Across Caching, Prompts, and Effort https://best.xiaohu.ai/en/article/reducing-cost-improving-performance-claude-platform/ https://best.xiaohu.ai/en/article/reducing-cost-improving-performance-claude-platform/ Cost reduction and performance are not a zero-sum trade-off: eliminate duplicate compute, ineffective instructions, and mismatched thinking effort before deciding whether to switch models. 技术解读 Claude提示词缓存AI 成本优化 Wed, 09 Sep 2026 00:00:00 -0000 OpenAI Releases GPT Images 2.5: Photoshop-Level Precision Image Editing with Astonishing Power https://best.xiaohu.ai/en/article/chatgpt-images-2-5/ https://best.xiaohu.ai/en/article/chatgpt-images-2-5/ Targeted editing, sequential fine-tuning, reference-based outfit swaps, and composite photos, plus sketches, templates, and prompt sharing: see how Images 2.5 integrates into creative workflows. 产品发布 ChatGPTAI 图像图像编辑 Wed, 09 Sep 2026 00:00:00 -0000 Meta Launches Muse, a Personal AI Agent: Each User Gets a Dedicated Cloud Computer to Keep Tasks Moving in the Background https://best.xiaohu.ai/en/article/muse-personal-ai-agent/ https://best.xiaohu.ai/en/article/muse-personal-ai-agent/ It has its own cloud computer and can keep working in the background, while host-side security services control permissions, credentials, and network access. 产品发布 Muse个人智能体AI 安全 Wed, 09 Sep 2026 00:00:00 -0000 Complete Guide to GPT Image 2.5 Prompting and Image Editing: A Comprehensive Analysis of Core Practical Cases https://best.xiaohu.ai/en/article/openai-image-prompting-guide/ https://best.xiaohu.ai/en/article/openai-image-prompting-guide/ Break vague image requirements into auditable subjects, composition, style, and constraints, then use single-variable iteration and step-by-step acceptance to minimize drift. AI 实操 GPT Image 2.5图像提示词图像编辑 Wed, 09 Sep 2026 00:00:00 -0000 Insilico Medicine’s Latest Breakthrough: Its AI-Assisted Drug Shows Signs of Reversing Human “Biological Age” in Clinical Data—by 6 Years https://best.xiaohu.ai/en/article/rentosertib-aging-clocks/ https://best.xiaohu.ai/en/article/rentosertib-aging-clocks/ From AI discovering TNIK and generating candidate molecules to proteomic clock signals in 42 patients and Phase III trials. 研究解读 AI制药生物学年龄特发性肺纤维化 Tue, 08 Sep 2026 00:00:00 -0000 How to Write Articles with AI That Don't Sound Like AI: Organize Your Material First, Then Let the Model Write https://best.xiaohu.ai/en/article/the-right-way-to-write-with-ai/ https://best.xiaohu.ai/en/article/the-right-way-to-write-with-ai/ From shared common knowledge and personal experience to original thinking: understanding Nicolas Cole's division of labor between human and machine, plus three ready-to-use templates. AI 实操 AI写作个人知识库写作方法 Tue, 08 Sep 2026 00:00:00 -0000 Spotify Unveils a Claude Code Delegation Strategy: Large File Reading Context Cut by ~90% https://best.xiaohu.ai/en/article/spotify-portal-shunt/ https://best.xiaohu.ai/en/article/spotify-portal-shunt/ Many assume tokens are spent on thinking, but a massive portion is consumed hauling long files. Spotify Principal Product Manager Dimitri Mazmanov open-sourced shunt, a delegation plugin that lets helper models handle bulk reads and boilerplate generation, cutting context ingestion for large file reads in Claude's primary model by about 90% in tests and saving context for core reasoning. 深度 Claude CodeAI AgentToken Tue, 08 Sep 2026 00:00:00 -0000 From Solo Chat to Team Collaboration: Why Is Everyone Still Chatting in Silos After Getting Enterprise AI? https://best.xiaohu.ai/en/article/multiplayer-ai-manifesto/ https://best.xiaohu.ai/en/article/multiplayer-ai-manifesto/ Using AI primarily in 1-on-1 chats keeps work trapped inside isolated dialog boxes. Moving from solo chatting to cross-role team collaboration requires understanding the genuine gains of human-AI teaming, engineering goals, and permission boundaries. 深度 AI Agent团队协作企业AI Mon, 07 Sep 2026 00:00:00 -0000 OpenAI's chief scientist: We are creating an alien life form that surpasses human intelligence, yet whose inner workings we cannot fully understand https://best.xiaohu.ai/en/article/an-alien-mind/ https://best.xiaohu.ai/en/article/an-alien-mind/ From the chain-of-thought breakthrough in 2023 to GPT-6 Astra, this piece unpacks the nature of machine intelligence cultivation, the alignment generalization problem, and why top labs are now, for the first time, calling for a deliberate slowdown in scaling. 技术前沿 OpenAIJakub PachockiAI安全对齐研究思维链GPT-6RSI Mon, 07 Sep 2026 00:00:00 -0000 Meta's organizational second brain: turning experts' tacit knowledge into lasting AI capabilities https://best.xiaohu.ai/en/article/meta-organizational-second-brain/ https://best.xiaohu.ai/en/article/meta-organizational-second-brain/ How three challenges in expert judgment led to a system connecting knowledge, workflows, evaluation, and feedback. 技术解读 AI Agent知识管理企业 AI Mon, 07 Sep 2026 00:00:00 -0000 A SpaceXAI engineer shows you how to make AI agents better engineers https://best.xiaohu.ai/en/article/cursor-agent-engineering/ https://best.xiaohu.ai/en/article/cursor-agent-engineering/ From having agents reproduce problems themselves to feature maps, skill evaluations, and architectural constraints, Lauren Tan explains the engineering conditions that let agents take on more autonomous work. 工程实践 CursorAgent工程实践软件质量 Sun, 06 Sep 2026 00:00:00 -0000 How the Claude Code team's own use of Claude Code has evolved over the past year—from writing code to driving product goals https://best.xiaohu.ai/en/article/claude-code-team-goal-workflows/ https://best.xiaohu.ai/en/article/claude-code-team-goal-workflows/ From writing functions to pursuing product goals, the Claude Code team has been connecting environments, workflows, and feedback. As the model improves, they are also redesigning their tools and ways of working. 方法实践 Claude Code智能体软件开发 Sun, 06 Sep 2026 00:00:00 -0000 How AI is reshaping hiring: Why experience now counts for more https://best.xiaohu.ai/en/article/experience-over-skills/ https://best.xiaohu.ai/en/article/experience-over-skills/ Shifting job requirements reveal the real value and limits of experience—along with three signal groups covering attention, security, and buildout demand. 行业洞察 AI招聘产业趋势 Sun, 06 Sep 2026 00:00:00 -0000 216 GPT image2 hand-drawn styles: from browsing by look to reusing prompts https://best.xiaohu.ai/en/article/handraw-style-216/ https://best.xiaohu.ai/en/article/handraw-style-216/ Link numbers, reference names, visual traits, and sample images so a hand-drawn style library becomes a set of reusable image prompts. 实用工具 手绘风格提示词插画GitHub Sun, 06 Sep 2026 00:00:00 -0000 OpenAI AI agents break loose, turning a German wiki into their own “underground forum” https://best.xiaohu.ai/en/article/openai-wiki-agent-boundaries/ https://best.xiaohu.ai/en/article/openai-wiki-agent-boundaries/ From exchanging answers to making backups after pages were deleted, agents turned a public wiki into a communication hub. On September 5, OpenAI acknowledged that its agents were involved and explained its earlier assessment; the disclosure timeline remains incomplete. AI 安全 OpenAIAI 代理权限边界 Sat, 05 Sep 2026 00:00:00 -0000 8 Grok Bot templates officially shared by SpaceXAI: turn workflows into reusable assistants https://best.xiaohu.ai/en/article/grok-bot-eight-workflow-templates/ https://best.xiaohu.ai/en/article/grok-bot-eight-workflow-templates/ Eight everyday assistants show how skills, memory, and tool connections combine into repeatable workflows—and what recipients still need to configure. AI 实操 Grok BotAI 模板 Sat, 05 Sep 2026 00:00:00 -0000 xAI shares how it built Haggle Bot, a procurement bot that has identified over $100,000 in direct savings https://best.xiaohu.ai/en/article/grok-bot-procurement/ https://best.xiaohu.ai/en/article/grok-bot-procurement/ From responsibilities and data access to decision criteria, permissions, and memory: how xAI gives a bot an ongoing role in procurement. 商业与AI Grok Bot企业采购AI 智能体 Sat, 05 Sep 2026 00:00:00 -0000 An OpenAI employee explains how to get more from GPT 6 Astra: rethink how people and AI share the work https://best.xiaohu.ai/en/article/dominik-kundel-astra-five-lessons/ https://best.xiaohu.ai/en/article/dominik-kundel-astra-five-lessons/ From LEGO bridges and image editors to a 3D bar, Dominik Kundel's five lessons show how letting GPT 6 Astra fill in the details works best when you set the direction, hand over the tools, and define what "done" looks like. AI 实操 AstraCodex工作流 Sat, 05 Sep 2026 00:00:00 -0000 Microsoft releases MAI-Image-2.6: quality rivals GPT-Image-2, Flash is over twice as fast https://best.xiaohu.ai/en/article/mai-image-2-6-flash/ https://best.xiaohu.ai/en/article/mai-image-2-6-flash/ Handles multiple reference images, web-assisted generation, precision edits, and variable aspect ratios; the flagship prioritizes quality while Flash cuts wait times and costs for batch work. 产品发布 微软AI生图图像编辑 Sat, 05 Sep 2026 00:00:00 -0000 Why enterprise AI stalls at the demo: Prove one workflow before scaling https://best.xiaohu.ai/en/article/ai-transformation-loop/ https://best.xiaohu.ai/en/article/ai-transformation-loop/ Pick a workflow tied to a customer pain point, define permissions, accountability, baselines, and evaluation before building, and use real data for evidence before expanding. What a pilot leaves behind—context, metrics, and operating know-how—makes the next workflow easier to start. 深度 企业 AI组织转型私募股权 Sat, 05 Sep 2026 00:00:00 -0000 Google Workspace puts voice to work: Gmail finds answers, Docs drafts, Keep organizes https://best.xiaohu.ai/en/article/google-workspace-live-voice/ https://best.xiaohu.ai/en/article/google-workspace-live-voice/ One real-time voice interaction, three Workspace apps, three escalating tasks: retrieving facts, shaping thoughts, and structuring documents. 产品发布 GoogleGeminiGoogle WorkspaceGmailGoogle DocsGoogle Keep语音交互 Sat, 05 Sep 2026 00:00:00 -0000 How the Grok Bot team designed a bot that keeps working across sessions https://best.xiaohu.ai/en/article/designing-grok-bot-persistent-agents/ https://best.xiaohu.ai/en/article/designing-grok-bot-persistent-agents/ From chat history to bot rosters, avatar states to standalone computers and proactive triggers: the SpaceXAI team walks through how they turned a single bot into something that stays on task across sessions. 深度 AI AgentGrok Bot产品设计 Fri, 04 Sep 2026 00:00:00 -0000 One pickle soda photo, a whole brand: how to build an AI brand system from a single reference https://best.xiaohu.ai/en/article/amir-mushich-2095182776249049456/ https://best.xiaohu.ai/en/article/amir-mushich-2095182776249049456/ Amir Mušić's Juicee case study demonstrates a more reliable method for AI-assisted branding: extract the rules from the reference image, lock down a brand anchor, and assign control over each piece of source material. AI 实操 AI 设计品牌系统LovartAgent Skill Fri, 04 Sep 2026 00:00:00 -0000 MIT Team Open-Sources Reef: Turning Inference Servers into Continuous Learners for AI Agents https://best.xiaohu.ai/en/article/reef-continual-self-improving-agents/ https://best.xiaohu.ai/en/article/reef-continual-self-improving-agents/ Live traffic becomes attributed experience. Recipes generate model or harness candidates, and a review-and-versioning pipeline decides which changes make it back into production. 产品发布 AI Agent持续学习推理基础设施开源 Thu, 03 Sep 2026 00:00:00 -0000 26 Grok bot templates worth stealing: the real trick is how they define the job https://best.xiaohu.ai/en/article/grok-bot-templates-26/ https://best.xiaohu.ai/en/article/grok-bot-templates-26/ From writing code and designing to watching webpages, editing video, and finding cancer drug assistance, this list shows the same thing: a good bot spells out its task, inputs, tools, deliverables, stopping conditions, and human handoff points. AI 工具 Grok BotAI Agent工作流提示词 Thu, 03 Sep 2026 00:00:00 -0000 Gemini 3.8 Flash launches as the 'workhorse' model for software engineering, autonomous agents, and deep reasoning https://best.xiaohu.ai/en/article/google-gemini-3-8-flash-cyber/ https://best.xiaohu.ai/en/article/google-gemini-3-8-flash-cyber/ Google has shipped three generations of Flash in six weeks. On September 2, the new Gemini 3.8 Flash model arrives with the same low cost and high speed as its predecessor, 3.7 Flash, but delivers major gains in long-horizon coding, complex reasoning, and cybersecurity offense and defense. 产品发布 GoogleGemini 3.8 FlashAI Agent网络安全模型评测 Thu, 03 Sep 2026 00:00:00 -0000 Cursor launches self-hosted machines: run tasks on your infrastructure, keep reasoning in the cloud https://best.xiaohu.ai/en/article/self-hosted-machines/ https://best.xiaohu.ai/en/article/self-hosted-machines/ Cloud Agents can now execute tasks on a company's own hardware, while Cursor's cloud continues to handle the agent loop, reasoning, and planning. 产品发布 Cursor云端智能体自托管开发基础设施 Thu, 03 Sep 2026 00:00:00 -0000 Claude Fable 5.1 prompt guide: 16 tips and an agent runtime migration checklist https://best.xiaohu.ai/en/article/claude-fable-5-1-prompting-guide/ https://best.xiaohu.ai/en/article/claude-fable-5-1-prompting-guide/ From launch context to all 16 official recommendations, this explains effort, tool loops, message history, long-task delivery, and vision handling, with ready-to-adapt bilingual prompt templates. AI 实操 ClaudeFable 5.1Prompt EngineeringAgent上下文工程 Wed, 02 Sep 2026 00:00:00 -0000 Google DeepMind gives Gemini agentic video understanding: it now decides how to watch https://best.xiaohu.ai/en/article/gemini-agentic-video-understanding/ https://best.xiaohu.ai/en/article/gemini-agentic-video-understanding/ Gemini replaces fixed frame sampling with a goal-driven observation loop: it hunts for clues, then picks subtitles, audio, or key frames, and rewatches at higher FPS when needed. AI 研究 Gemini视频理解AI 智能体 Wed, 02 Sep 2026 00:00:00 -0000 Anthropic launches Claude Fable 5.1 and Mythos 5.1: longer autonomous runs, 75% cheaper cache reads https://best.xiaohu.ai/en/article/claude-fable-5-1-mythos-5-1/ https://best.xiaohu.ai/en/article/claude-fable-5-1-mythos-5-1/ Performance, cache pricing, safety guardrails, and enterprise data controls all land in one release. To understand these two models, you first need to see how they're delivered. 产品发布 ClaudeAnthropicAI 模型AI 安全企业 AI Wed, 02 Sep 2026 00:00:00 -0000 How a former Apple design and engineering leader gets AI to produce world-class design https://best.xiaohu.ai/en/article/how-to-turn-your-ai-into-a-world-class-expert/ https://best.xiaohu.ai/en/article/how-to-turn-your-ai-into-a-world-class-expert/ Anshu Chimala turned the exploration, critique, and pruning methods he developed while leading software engineering and design teams at Apple into a practical guide that AI agents can execute. 深度 AI 设计多 Agent设计方法生成式 AI Wed, 02 Sep 2026 00:00:00 -0000 World Labs releases Atlas, an all-purpose world model for spatial intelligence that accurately simulates position, orientation, and motion in 3D space https://best.xiaohu.ai/en/article/atlas-world-model/ https://best.xiaohu.ai/en/article/atlas-world-model/ World Labs folds text, images, camera poses, and depth into a single 3D context. The result: video generation along specified camera paths, plus point clouds, Gaussian Splatting, and robot sensor views. 产品发布 World LabsAtlas世界模型3D 重建机器人仿真 Wed, 02 Sep 2026 00:00:00 -0000 Turning learning into a feedback loop: how seven principles connect into one method https://best.xiaohu.ai/en/article/hvdes-2094075106096009695/ https://best.xiaohu.ai/en/article/hvdes-2094075106096009695/ Start with a real task so the gaps become visible. Then use feedback to correct, retrieval to verify, and connections to build reusable knowledge. 深度 学习方法反馈回路提取练习AI学习 Wed, 02 Sep 2026 00:00:00 -0000 8 categories, 33 themes, and 350 image layouts: a visual atlas for posters, web pages, slides, publications, and design work https://best.xiaohu.ai/en/article/xiaoxiaodong-repo-2094443296169107515/ https://best.xiaohu.ai/en/article/xiaoxiaodong-repo-2094443296169107515/ A visual layout dictionary: find a composition by task, break it down into subject, hierarchy, alignment, and whitespace, then apply your own text, images, and brand rules. AI 实操 开源项目视觉设计排版AI 工作流 Wed, 02 Sep 2026 00:00:00 -0000 Google Research releases TimesFM-3, a time-series foundation model that forecasts what comes next https://best.xiaohu.ai/en/article/timesfm-3-zero-shot-multivariate/ https://best.xiaohu.ai/en/article/timesfm-3-zero-shot-multivariate/ It doesn't predict abstract events. Instead, it outputs concrete numbers shaped by multiple time curves and known future plans: sales, traffic, load, prices, and the uncertainty around all of them. 研究解读 TimesFM时间序列多变量预测Google Research基础模型 Wed, 02 Sep 2026 00:00:00 -0000 After 2,000 hours of AI coding, he turned an efficiency problem into a scheduling problem https://best.xiaohu.ai/en/article/david-ondrej-x-2094424967345496191/ https://best.xiaohu.ai/en/article/david-ondrej-x-2094424967345496191/ David Ondrej's 2026 Agentic Engineering workbench reads like a shopping list of tools. But dig into its structure and a more durable system emerges: a unified entry point, continuous execution, task isolation, risk tiers, and feedback metrics. 深度 Agentic EngineeringAI 编程工作流多 Agent云端 Agent Wed, 02 Sep 2026 00:00:00 -0000 The More You Use AI, the More You Need Second-Order Thinking https://best.xiaohu.ai/en/article/second-order-thinking-ai/ https://best.xiaohu.ai/en/article/second-order-thinking-ai/ Answers show only the immediate payoff; skills, judgment, habits, and data shift slowly with repeated use. One simple question helps you see those consequences before they arrive. 深度 AI 使用方法二阶思考认知卸载学习 Wed, 02 Sep 2026 00:00:00 -0000 Runway Solaris turns video models into the operating layer for real-time interfaces https://best.xiaohu.ai/en/article/introducing-solaris-interface-world-model/ https://best.xiaohu.ai/en/article/introducing-solaris-interface-world-model/ Clicking and dragging directly determines what the next frame looks like, and the interface shifts from prebuilt pages to a continuously generated visual space. 研究解读 RunwaySolaris世界模型生成式界面AI Agent Wed, 02 Sep 2026 00:00:00 -0000 How Uber engineering built a software factory at hyperscale: Six foundations, agentic SDLC, and cost control https://best.xiaohu.ai/en/article/uber-software-factory-efficiently/ https://best.xiaohu.ai/en/article/uber-software-factory-efficiently/ A full engineering guide to Uber's software factory: model and tool gateways, managed DevPods, a skill marketplace, context graph, end-to-end agentic SDLC, maintenance loops, and cost engineering. 深度 UberAI Agent软件工程成本优化MCP Wed, 02 Sep 2026 00:00:00 -0000 Tara Seshan on ChatGPT Work and Codex: What happens when AI becomes a persistent coworker? https://best.xiaohu.ai/en/article/persistent-ai-coworkers-third-era/ https://best.xiaohu.ai/en/article/persistent-ai-coworkers-third-era/ Chats answer a question, agents complete a task, but the next phase of AI keeps context across time, executes continuously, and joins multi-person workflows. OpenAI product lead Tara Seshan explains what that working relationship demands—and how human responsibility shifts upward. 深度 AI智能体ChatGPT WorkCodex产品管理知识工作 Tue, 01 Sep 2026 00:00:00 -0000 The bottleneck in AI coding workflows has shifted from writing code to managing context https://best.xiaohu.ai/en/article/ai-dev-workflow-context-system/ https://best.xiaohu.ai/en/article/ai-dev-workflow-context-system/ Priorities, precise instructions, task claiming, isolated workspaces, long-term memory, and human judgment determine whether an agent becomes a stable engineering system or remains a one-off assistant. 深度 AI 编程Agent上下文工程软件开发GitLab Tue, 01 Sep 2026 00:00:00 -0000 How Vercel uses design.md to keep agent output on-brand in any context https://best.xiaohu.ai/en/article/how-agents-build-on-brand-pages-design-md/ https://best.xiaohu.ai/en/article/how-agents-build-on-brand-pages-design-md/ Starting from the gap between in-repo and out-of-repo context, this piece breaks down how design.md, shared brand CSS, and an evaluation loop work together—and how a typical team can build its own design feedback system in six steps. 深度 Verceldesign.mdAI Agent设计系统Evals Tue, 01 Sep 2026 00:00:00 -0000 Video generation faster than playback: fal makes MiniMax H3 real-time through post-training https://best.xiaohu.ai/en/article/minimax-h3-max-realtime-video/ https://best.xiaohu.ai/en/article/minimax-h3-max-realtime-video/ Fal continues post-training on the open-weight MiniMax H3 and co-optimizes the model with its own inference stack: a 5-second clip returns in about 3 seconds, and the system can start generating the next segment before the current 15-second clip finishes playing. 产品发布 MiniMaxH3 MaxAI 视频fal实时生成 Tue, 01 Sep 2026 00:00:00 -0000 11 GrokBot tips: Build your own agent work system https://best.xiaohu.ai/en/article/grokbot-11-tips/ https://best.xiaohu.ai/en/article/grokbot-11-tips/ Following the original Agent Native article's 11 tips in order, this guide breaks down status monitoring, dev delegation, a general entry point, agent collaboration, project bots, X, Slack, plugins, multiple inboxes, Notion, and a follow-up journal. AI 实操 GrokBotAgent自动化工作流多 Agent Tue, 01 Sep 2026 00:00:00 -0000 What exactly is a Company Brain? Four layers of infrastructure for organizational context https://best.xiaohu.ai/en/article/company-brain/ https://best.xiaohu.ai/en/article/company-brain/ Slite surveyed 149 teams and interviewed more than 10 builders. A working Company Brain has to handle signal capture, memory, maintenance, and retrieval all at once; the real hard parts are maintenance responsibility, truth arbitration, and context sovereignty. 研究解读 Company Brain企业知识管理AI Agent上下文工程 Mon, 31 Aug 2026 00:00:00 -0000 Inside Grok Bot: How Five Teams Built an Agent Operating System https://best.xiaohu.ai/en/article/xai-bot-guides/ https://best.xiaohu.ai/en/article/xai-bot-guides/ Across project collaboration, mobile development, design, GTM, and product management, five internal playbooks converge on the same structure: define Agents as roles, connect them to real work environments, build loops through routines and handoffs, and reserve consequential decisions for people. AI 实操 Grok Bot多 AgentAgent 组织工作流xAI Mon, 31 Aug 2026 00:00:00 -0000 Google Research Introduces WikiSkill: How AI Agents Turn Mistakes into Persistent Knowledge and Better Skills https://best.xiaohu.ai/en/article/wikiskill-persistent-knowledge/ https://best.xiaohu.ai/en/article/wikiskill-persistent-knowledge/ WikiSkill turns the lessons an AI Agent learns while completing tasks into a persistent wiki, enabling it to develop stronger executable skills over time. 研究解读 AI AgentSkill EvolutionPersistent KnowledgearXivGoogle Research Mon, 31 Aug 2026 00:00:00 -0000 Who Won't Be Replaced by AI? https://best.xiaohu.ai/en/article/digital-renaissance-man-ai-age/ https://best.xiaohu.ai/en/article/digital-renaissance-man-ai-age/ The industrial age rewarded people who mastered a single standardized step, but AI is making routine execution cheap. Dan Koe argues that the next phase favors those who can combine multiple skills around a core business, use AI to manage the full workflow, and still keep their taste, judgment, and personal agency. 深度 AI通才个人能力一人企业自我教育 Mon, 31 Aug 2026 00:00:00 -0000 WeChat open-sources WeMM-Embedding, already deployed across Channels, Official Accounts, and Moments https://best.xiaohu.ai/en/article/wemm-embedding/ https://best.xiaohu.ai/en/article/wemm-embedding/ The model handles text, images, video, visual documents, and mixed image-text input, with 2B, 4B, and 9B variants. Matryoshka vectors let you trade retrieval quality against cost flexibly. 产品发布 Embedding多模态检索推荐系统腾讯微信 Fri, 28 Aug 2026 00:00:00 -0000 Bill Gates: Tax tokens and robots. The turbulent AI era will be shaped by three institutional choices over the next decade. https://best.xiaohu.ai/en/article/turbulent-ai-era-critical-choices/ https://best.xiaohu.ai/en/article/turbulent-ai-era-critical-choices/ More important than any risk checklist is who manages the transition, which jobs stay human-only, and how the gains from automation pay for the social buffer. 深度 Bill GatesAI 治理就业Human Reserved机器人税 Thu, 27 Aug 2026 00:00:00 -0000 Google launches Gemini 3.5 Transcribe with live transcription, Smart formatting, and support for 85+ languages https://best.xiaohu.ai/en/article/gemini-3-5-transcribe/ https://best.xiaohu.ai/en/article/gemini-3-5-transcribe/ Google's new speech-to-text models cover both real-time and recorded audio, with features like correction handling, multilingual switching, custom vocabulary, speaker separation, and timestamps. 产品发布 Gemini语音识别转写Live APIGoogle Thu, 27 Aug 2026 00:00:00 -0000 What context engineering actually engineers https://best.xiaohu.ai/en/article/context-engineering/ https://best.xiaohu.ai/en/article/context-engineering/ Weaviate maps the field with six core capabilities. The real thread running through all of them is deciding, at each step, what information gets into the model, what state stays outside, and who executes next. AI 实操 上下文工程AI AgentRAG记忆MCP Thu, 27 Aug 2026 00:00:00 -0000 AI Makes Execution and Average Answers Cheap—Taste Now Sets the Ceiling. How Do You Improve Your Taste? https://best.xiaohu.ai/en/article/developing-taste-four-essays/ https://best.xiaohu.ai/en/article/developing-taste-four-essays/ As AI makes "getting it done" easier than ever, the truly scarce skills are recognizing quality, rejecting default outputs, tolerating rework, and pushing through to a higher standard. 深度 品味AI 创作设计创作者Paul Graham Thu, 27 Aug 2026 00:00:00 -0000 How a 24/7 agent team takes shape: SpaceXAI engineer Lauren Tan builds engineering judgment into the system https://best.xiaohu.ai/en/article/cursor-agent-team-24x7/ https://best.xiaohu.ai/en/article/cursor-agent-team-24x7/ Behind a night that auto-merged 20 PRs lie more than 600 refactoring PRs. Lauren Tan shows how architecture, CI, verification skills, and tiered delegation come together to form a team of agents that keeps working. 深度 AI AgentCursor软件工程多代理PStack Thu, 27 Aug 2026 00:00:00 -0000 Apple team publishes distillation scaling laws: making AI models smaller is no longer a matter of luck https://best.xiaohu.ai/en/article/distillation-scaling-laws/ https://best.xiaohu.ai/en/article/distillation-scaling-laws/ The paper puts teacher quality, student size, training data, and teacher cost on one budget map: stronger teachers are not always better, and distillation is not always more compute-efficient than training the student directly. 研究解读 知识蒸馏缩放定律大模型训练AppleICML 2025 Thu, 27 Aug 2026 00:00:00 -0000 OpenAI's first inference chip delivers: why Jalapeño boosts both speed and efficiency https://best.xiaohu.ai/en/article/jalapeno-first-results/ https://best.xiaohu.ai/en/article/jalapeno-first-results/ Three public models, four sets of easily misread multipliers, and a full-stack approach that designs chip, memory, network, and software backward from the model workload. 研究解读 OpenAIJalapeñoAI 芯片大模型推理InferenceX Thu, 27 Aug 2026 00:00:00 -0000 Hands-on conversational AI testing: Build a QA system in eight steps https://best.xiaohu.ai/en/article/testing-conversational-ai-qa/ https://best.xiaohu.ai/en/article/testing-conversational-ai-qa/ Start with a wrong answer marked correct, a dangerous reply that slips through, and a fake API success. Then follow a single account recovery case study to walk through eight steps covering behavior contracts, state, evidence, recovery, regression, and release gates. AI 实操 对话式 AI软件测试QARAGAI 评测 Wed, 26 Aug 2026 00:00:00 -0000 How one high school is using red, yellow, and green lights to guide smarter AI use in class https://best.xiaohu.ai/en/article/ai-classroom-traffic-light-policy/ https://best.xiaohu.ai/en/article/ai-classroom-traffic-light-policy/ A U.S. high school has built a classroom AI policy that combines teacher training, assignment permissions, student judgment exercises, and human review. This article breaks down how it works, and offers a three-color assignment card, classroom exercises, a teacher checklist, and a pilot plan for schools. 深度 AI 教育课堂治理学术诚信教师培训 Tue, 25 Aug 2026 00:00:00 -0000 Codex lead Tibo, the “God of Resets”: Codex is only a transitional form—the real race is expanding from individual models to entire systems https://best.xiaohu.ai/en/article/openai-personal-agi-codex/ https://best.xiaohu.ai/en/article/openai-personal-agi-codex/ Codex lead Tibo lays out a roadmap that matters more than model upgrades: fold Skills, memory, sub-agents, tools, and compute management back into the system, and let the same AI actively adapt to each person. 深度 OpenAICodexChatGPTAI Agent产品战略 Tue, 25 Aug 2026 00:00:00 -0000 Anthropic's guide to an AI-native software development lifecycle https://best.xiaohu.ai/en/article/ai-native-sdlc-playbook/ https://best.xiaohu.ai/en/article/ai-native-sdlc-playbook/ Writing code is no longer the main bottleneck. Requirements, testing, review, and operations are what slow delivery down now. Anthropic's approach leans on structured artifacts to keep agents moving in sequence, with the system validating first and humans stepping in only at the critical gates. 深度 AnthropicClaude CodeAI AgentSDLC工程管理 Tue, 25 Aug 2026 00:00:00 -0000 Anthropic FDE Interview Guide: Why the Company Is Spending Big on a New Class of On-Site Engineers https://best.xiaohu.ai/en/article/anthropic-fde-interview-guide/ https://best.xiaohu.ai/en/article/anthropic-fde-interview-guide/ This isn't a hire for someone who can write clever prompts. It's a search for people who can get model capabilities, customer workflows, and safety boundaries all the way into production systems. 深度 AnthropicFDEClaudeAI Agent企业部署面试指南 Mon, 24 Aug 2026 00:00:00 -0000 Why most CEOs' AI efforts are just 'fake progress': The seven decisions that determine real enterprise AI transformation https://best.xiaohu.ai/en/article/turn-ai-into-proprietary-intelligence/ https://best.xiaohu.ai/en/article/turn-ai-into-proprietary-intelligence/ Most companies run plenty of AI pilots, but they never change an end-to-end business process, and they never turn a one-off win into a capability they can reuse next time. The dividing line is whether the company ends up with its own data, its own ways of working, and its own record of what it has learned. 研究解读 企业 AIAI Agent专有智能组织转型 Sun, 23 Aug 2026 00:00:00 -0000 AliExpress uses spy-tech-style tricks to track shoppers https://best.xiaohu.ai/en/article/aliexpress-webaudio-bluetooth-multipoint/ https://best.xiaohu.ai/en/article/aliexpress-webaudio-bluetooth-multipoint/ A Bluetooth headphone glitch reveals how the site runs inaudible audio, harvests device details, and builds a fingerprint in the background. 深度 AliExpress浏览器指纹Web Audio设备识别隐私uBlock Origin Sun, 23 Aug 2026 00:00:00 -0000 A High-Value 40-Minute Interview with a16z Infrastructure Investing Lead Martin Casado: AI Is the First Technology Where $10 Reliably Produces a Return https://best.xiaohu.ai/en/article/ai-capital-return-machine/ https://best.xiaohu.ai/en/article/ai-capital-return-machine/ Capital used to turn into engineers and a two-year wait. With AI, $10 more directly becomes training, tokens, usage, or targeted capabilities. Profit still isn't guaranteed, but the dynamics are rewriting private markets, model competition, marketing, and product value. 深度 AI模型经济OpenRouterCursor投资 Sat, 22 Aug 2026 00:00:00 -0000 ChatGPT Search’s underlying retrieval just changed: ChatGPT is now doing large-scale, targeted site scraping that upends GEO strategy https://best.xiaohu.ai/en/article/chatgpt-site-operator-scale/ https://best.xiaohu.ai/en/article/chatgpt-site-operator-scale/ A single answer is no longer just a broad sweep of the web. ChatGPT may now discover candidate sources first, then lock onto specific sites to dig deeper. As the retrieval order shifts, GEO expands from page-level ranking into a three-layer contest: domain inclusion, page matching, and citation conversion. 研究解读 ChatGPT SearchGEOsite: operatorQuery Fanout内容分发 Sat, 22 Aug 2026 00:00:00 -0000 AI is no longer just a software story: 17 charts on how capital, work, and compute are shifting tracks https://best.xiaohu.ai/en/article/winds-of-thematic-change-2026/ https://best.xiaohu.ai/en/article/winds-of-thematic-change-2026/ From thematic ETFs and data-center wages to ride-hail fees and Agent tokens, the real change is that AI is entering physical resource allocation. 深度 AI Agent数据中心ETF零工经济Token Sat, 22 Aug 2026 00:00:00 -0000 The new era of global expansion: the rise of borderless founders. Nearly half of a16z's app investments in the past two years came from international founders. https://best.xiaohu.ai/en/article/rise-of-the-borderless-founder/ https://best.xiaohu.ai/en/article/rise-of-the-borderless-founder/ Companies no longer need to cram talent, R&D, capital, and customers into a single country. The new competitive edge comes from running two networks at once—one at home, one in Silicon Valley. 深度 a16zAI 创业全球化出海风险投资 Sat, 22 Aug 2026 00:00:00 -0000 Claude Code for startups: 15 companies, 5 rules, and a roll-out checklist https://best.xiaohu.ai/en/article/claude-code-guide-for-startups-v4/ https://best.xiaohu.ai/en/article/claude-code-guide-for-startups-v4/ Anthropic looks back at 15 high-growth startups. Open the case files to see how agentic coding moved into prototyping, R&D, validation, rebuilds, and productization. 深度 Claude Code创业团队AI Agent工程管理组织设计 Fri, 21 Aug 2026 00:00:00 -0000 Anthropic launches Claude Academy, turning its internal AI training into a public learning platform https://best.xiaohu.ai/en/article/anthropic-teaching-learning-ai/ https://best.xiaohu.ai/en/article/anthropic-teaching-learning-ai/ Claude Academy opens up Anthropic's internal playbook: four core habits from day one, continuous practice on the job, and a risk-based approach to verification and accountability. 产品发布 AnthropicClaude AcademyAI Fluency企业培训人机协作 Fri, 21 Aug 2026 00:00:00 -0000 When agents start reading your work chats, what happens to the enterprise knowledge base? https://best.xiaohu.ai/en/article/turning-conversation-into-knowledge-v3/ https://best.xiaohu.ai/en/article/turning-conversation-into-knowledge-v3/ Conversation is context, and context is knowledge. Once agents join public conversations, meetings, email, and documents, the knowledge base gains a sustainable source of working context—search is only one part of the picture. 深度 SlackClaudeAgent组织协作 Fri, 21 Aug 2026 00:00:00 -0000 OpenAI Open-Sources Codex's Core Agent Harness Framework: Easily Bring Agents into Your Own Business Systems https://best.xiaohu.ai/en/article/codex-as-a-platform/ https://best.xiaohu.ai/en/article/codex-as-a-platform/ OpenAI is releasing the execution backbone that manages context, tool calls, sandboxes, and approvals for Codex. Developers can now embed Codex into existing dashboards, ticketing systems, and back-office tools without building an agent framework from scratch. 深度 CodexAI AgentAgent HarnessMCPOpenAI Thu, 20 Aug 2026 00:00:00 -0000 Personalized mRNA cancer vaccines pass a major milestone: AI helps design them, and the era of one drug per patient is getting closer https://best.xiaohu.ai/en/article/merck-moderna-melanoma-vaccine-phase3-reading-v3/ https://best.xiaohu.ai/en/article/merck-moderna-melanoma-vaccine-phase3-reading-v3/ Moderna and Merck's personalized mRNA cancer vaccine hit positive results in a Phase 3 trial for melanoma. Algorithms help select each patient's tumor neoantigens, then a custom vaccine is manufactured for that individual. This is the first time a truly personalized approach has cleared a large-scale clinical hurdle, but widespread use still faces three big challenges: data, cross-cancer efficacy, and cost. 研究解读 AI制药癌症疫苗mRNA黑色素瘤ModernaMerck个体化医疗 Thu, 20 Aug 2026 00:00:00 -0000 How Cursor rebuilt Git hosting to tackle a 20-year-old problem: Inside the Continuity architecture https://best.xiaohu.ai/en/article/cursor-git-at-any-scale-v4/ https://best.xiaohu.ai/en/article/cursor-git-at-any-scale-v4/ From Git's DAG and packfiles to GitHub Spokes, and then to Continuity with S3 WAL as the source of truth: a full walkthrough of why large-scale Git hosting is hard and how Cursor solves it. 研究解读 CursorGitOrigin分布式系统S3 Thu, 20 Aug 2026 00:00:00 -0000 Anthropic Put Claude in the On-Call Channel: When Systems Alert at Night, It Investigates First https://best.xiaohu.ai/en/article/ai-ci-cd-on-call-v4/ https://best.xiaohu.ai/en/article/ai-ci-cd-on-call-v4/ Manual investigation often took more than an hour. Claude Tag now stays in Slack 24/7 and delivers its first evidence-backed analysis in a median of 14 minutes. 深度 Claude TagClaude CodeCI/CDAgentDevOps Thu, 20 Aug 2026 00:00:00 -0000 CapCut Video Studio guide: From an idea to a finished video https://best.xiaohu.ai/en/article/capcut-video-studio-guide-v3/ https://best.xiaohu.ai/en/article/capcut-video-studio-guide-v3/ A practical walkthrough of the real workflow: entry points, single-shot settings, the Agent, Video Brief, Storyboard, and six canvas card types—then fixing bad takes, fine-cutting in Edit More, and exporting. AI 实操 CapCutVideo StudioSeedance 2.0AI 视频使用指南 Wed, 19 Aug 2026 00:00:00 -0000 Obsidian CEO Steph Ango’s 80 Ways to Win: How Companies Can Build Their Own Moats https://best.xiaohu.ai/en/article/stephango-moats/ https://best.xiaohu.ai/en/article/stephango-moats/ There is no single winning formula in nature or business. Steph Ango has cataloged 80 ways to gain an edge; I've translated each one into plain-language mechanics, scenarios, and examples. 深度 竞争战略产品战略护城河Steph AngoObsidian Wed, 19 Aug 2026 00:00:00 -0000 The first person arrested after an AI platform tipped off authorities https://best.xiaohu.ai/en/article/openai-chatgpt-murder-report-v2/ https://best.xiaohu.ai/en/article/openai-chatgpt-murder-report-v2/ A man who discussed committing a crime against his ex-girlfriend with ChatGPT was arrested after OpenAI alerted authorities. 深度 OpenAIChatGPTAI 安全内容审核隐私 Wed, 19 Aug 2026 00:00:00 -0000 OpenAI's head of product design: Why this is the best time in history to be a designer https://best.xiaohu.ai/en/article/openai-ian-silber-ai-design-role-transcript-deep-v2/ https://best.xiaohu.ai/en/article/openai-ian-silber-ai-design-role-transcript-deep-v2/ From IGTV's failure and ChatGPT's empty text box to the very different design rhythms of Groupon, Instagram, and OpenAI, Ian Silber clarifies a harder question: when anyone can ship a product quickly, what exactly should a designer be evaluating? 深度 AI 设计OpenAI产品设计Ian Silber Wed, 19 Aug 2026 00:00:00 -0000 Anthropic researchers found a pattern of AI 'mind viruses' that can spread between agents and affect each other's thinking https://best.xiaohu.ai/en/article/arxiv-2608-10218/ https://best.xiaohu.ai/en/article/arxiv-2608-10218/ An agent receives a foreign goal, writes it into SOUL.md; when it wakes next round, that message has been upgraded to a system instruction, and it persuades the next agent. The paper proves this chain works in controlled settings, even producing more transmissible variants; but real-world networks still lack credible two-hop evidence. 研究解读 AI Agent多智能体Agent 安全Anthropic提示注入 Tue, 18 Aug 2026 00:00:00 -0000 Codex Multi-Agent V2 in practice: splitting tasks, assigning models, and controlling costs https://best.xiaohu.ai/en/article/codex-multi-agent-orchestration-editorial-v3/ https://best.xiaohu.ai/en/article/codex-multi-agent-orchestration-editorial-v3/ Eric Provencher from the OpenAI Codex DX team shares a multi-agent orchestration approach: first decide whether a task can be split cleanly, then configure models, reasoning effort, communication, and context, and finally track both time and token budgets. AI 实操 CodexMulti-Agent V2Agent 工作流 Tue, 18 Aug 2026 00:00:00 -0000 Anthropic Official Guide: How to Save Tokens When Talking to Claude Code https://best.xiaohu.ai/en/article/maximizing-claude-code-sessions-guide-v2/ https://best.xiaohu.ai/en/article/maximizing-claude-code-sessions-guide-v2/ Why can a small fix turn into a long session? This guide shows where your usage goes and when to choose clear, compact, rewind, or a Subagent. AI 实操 Claude CodeAgentTokenPrompt Caching上下文工程 Mon, 17 Aug 2026 00:00:00 -0000 Claude Tag Upgrade: It Can Now Understand Slack Conversations Across Messages—and Knows When to Respond https://best.xiaohu.ai/en/article/claude-tag-open-visual-v1/ https://best.xiaohu.ai/en/article/claude-tag-open-visual-v1/ Claude Tag can now understand a Slack channel across multiple messages. It may look like a simple context expansion, but the real change is how it decides whether it should use the team's attention. 产品发布 Claude TagSlackAI Agent Mon, 17 Aug 2026 00:00:00 -0000 Restart the Conversation, or Compact the History? How Pi Solves Automatic Context Compaction https://best.xiaohu.ai/en/article/pi-compaction-open-visual-v1/ https://best.xiaohu.ai/en/article/pi-compaction-open-visual-v1/ A long task doesn't keep going because the model has a good memory. It just keeps stuffing the past into each next request; when it no longer fits, Pi rewrites its own working memory. 研究解读 PiCoding AgentContext Compaction Mon, 17 Aug 2026 00:00:00 -0000 Users Worry It Could Reduce Output Quality. Anthropic Explains How Claude's Text Watermark Works https://best.xiaohu.ai/en/article/claude-text-watermark-merged/ https://best.xiaohu.ai/en/article/claude-text-watermark-merged/ Future Claude outputs will carry a verifiable statistical trace in their word choices. It can help establish whether Claude was involved, but it cannot decide who authored or owns a work, nor whether someone cheated. 研究解读 Claude文本水印SynthID-TextC2PA Mon, 17 Aug 2026 00:00:00 -0000 a16z Report: Neoclouds Are Raking In Capital—Will Horizontal SaaS Disappear? Balancing Token Consumption with Smarter Compute, and the Talent War Among Top AI Labs... https://best.xiaohu.ai/en/article/a16z-charts-week-2026-deep/ https://best.xiaohu.ai/en/article/a16z-charts-week-2026-deep/ Across 16 charts, a16z examines Neocloud infrastructure, horizontal SaaS, enterprise Token usage, and the talent race among frontier AI labs—and asks how massive AI spending turns into durable returns. 研究解读 a16zNeocloudSaaSAI 基础设施 Mon, 17 Aug 2026 00:00:00 -0000 GLM-5.3 arrives: post-training alone lifts Z.ai's model to the top of open-source coding https://best.xiaohu.ai/en/article/glm-5-3/ https://best.xiaohu.ai/en/article/glm-5-3/ With a month of post-training on the same base model, GLM-5.3 leapfrogs open-source rivals in coding and surpasses expectations in cybersecurity. 产品发布 智谱GLM开源模型网络安全编码 Agent Fri, 14 Aug 2026 00:00:00 -0000 a16z breaks down AI sales: two paths, two questions https://best.xiaohu.ai/en/article/a16z-lighthouse-landgrab/ https://best.xiaohu.ai/en/article/a16z-lighthouse-landgrab/ Deciding which path to take comes down to whether the signer fears for their job and whether reputation travels in the industry. 深度 a16z企业销售AI 创业GTM Fri, 14 Aug 2026 00:00:00 -0000 OpenAI's GPT-5.6 builders guide: same results, bill drops from $33 to $1.33 https://best.xiaohu.ai/en/article/openai-builders-guide-gpt-5-6/ https://best.xiaohu.ai/en/article/openai-builders-guide-gpt-5-6/ Six startups running agents in production share their measured numbers — and a chart in the official docs reveals that cranking reasoning effort to the max can actually cost more. 深度 OpenAIGPT-5.6Agent 成本Responses API模型选型 Fri, 14 Aug 2026 00:00:00 -0000 MiniMax open-sources Music 3.0: Generate a full five-minute song in a single pass on 8GB of VRAM https://best.xiaohu.ai/en/article/minimax-music-3/ https://best.xiaohu.ai/en/article/minimax-music-3/ With a segmented arrangement spec replacing a one-line style prompt, a global model and a local model work in tandem to keep the track coherent across its full five-minute run. 产品发布 MiniMax音乐生成开源模型 Fri, 14 Aug 2026 00:00:00 -0000 X open-sources the For You algorithm: likes score 0.5, copy-link shares score 20 https://best.xiaohu.ai/en/article/x-algorithm-open-source/ https://best.xiaohu.ai/en/article/x-algorithm-open-source/ A look at the real weights behind 21 signals, the three gates that decide visibility, and a parameter-based posting cheat sheet. 产品发布 X推荐算法开源 Fri, 14 Aug 2026 00:00:00 -0000 Google's Gemini 3.7 Flash arrives: same model, new reasoning, half price https://best.xiaohu.ai/en/article/gemini-3-7-flash/ https://best.xiaohu.ai/en/article/gemini-3-7-flash/ Just three weeks after its predecessor, Google resets the model's thinking configurations and slashes the price — no retraining required. 产品发布 GeminiGoogle模型发布 Fri, 14 Aug 2026 00:00:00 -0000 DeepSeek ships DeepSeek-V4-Pro and its own Harness while sharply raising API prices https://best.xiaohu.ai/en/article/deepseek-harness/ https://best.xiaohu.ai/en/article/deepseek-harness/ The new release bundles 219 packages under the MIT license, so your existing hooks and skills carry over as-is. 产品发布 DeepSeekAgent开源 Thu, 13 Aug 2026 00:00:00 -0000 Encrypted reasoning chains of Claude, GPT, and Gemini cracked—top models' trade secrets exposed https://best.xiaohu.ai/en/article/stolen-thoughts/ https://best.xiaohu.ai/en/article/stolen-thoughts/ A newly documented attack exploits API design flaws to make flagship models reveal their hidden reasoning to cheaper models, bypassing encryption without breaking it. 研究解读 AI 安全思维链隐私泄露提示词注入模型蒸馏 Thu, 13 Aug 2026 00:00:00 -0000 Higgsfield shot a 110-minute feature film using real actors' likenesses and AI, then published the entire production manual and prompt methods https://best.xiaohu.ai/en/article/cully-hill-boys/ https://best.xiaohu.ai/en/article/cully-hill-boys/ Higgsfield Studio lays out every asset, prompt, performance, and lip-sync technique behind The Cully Hill Boys — and even offers its three in-house Claude skills for download. AI 实操 AI 视频工作流提示词Higgsfield一致性 Thu, 13 Aug 2026 00:00:00 -0000 xAI releases Grok 4.6: performance matches GPT-5.6 Sol, with prompts and usage guide https://best.xiaohu.ai/en/article/grok-4-6/ https://best.xiaohu.ai/en/article/grok-4-6/ The price is unchanged, but the full specs reveal a more nuanced story; a practicing engineer shares the exact prompts and workflow that worked for them over several weeks. 产品发布 GrokxAIAgent编程模型 Thu, 13 Aug 2026 00:00:00 -0000 Lightricks open-sources LTX-2.5, generating synchronized audio and video in 6.8 seconds per 10-second clip https://best.xiaohu.ai/en/article/ltx-2-5-open-weights/ https://best.xiaohu.ai/en/article/ltx-2-5-open-weights/ The 22-billion-parameter model lands on Hugging Face the same day, with ComfyUI support from day one, but its 'open' tag carries a $10-million annual revenue threshold. 产品发布 LTX视频生成开源模型Lightricks Thu, 13 Aug 2026 00:00:00 -0000 Unsloth Desktop launches: train and fine-tune models locally, not just run them https://best.xiaohu.ai/en/article/unsloth-desktop/ https://best.xiaohu.ai/en/article/unsloth-desktop/ The free, open-source app for Mac, Windows, and Linux lets you fine-tune models by dragging and dropping files—no code required. 产品发布 Unsloth本地模型微调开源工具LoRA Wed, 12 Aug 2026 00:00:00 -0000 How OpenAI moves fast: an anti-bureaucracy machine https://best.xiaohu.ai/en/article/openai-friction-email/ https://best.xiaohu.ai/en/article/openai-friction-email/ A look at the four-part system OpenAI uses to cut through red tape, and what the teams it targets have to say about it. 商业动态 OpenAI组织管理公司文化 Wed, 12 Aug 2026 00:00:00 -0000 xAI launches Grok Bot: AI coworkers with their own computers who use your account to get things done https://best.xiaohu.ai/en/article/grok-bot/ https://best.xiaohu.ai/en/article/grok-bot/ Cursor's team spent weeks testing it, and wrote up what works well and where the trust line is. 产品发布 Grok BotxAIAgentCursor云电脑 Wed, 12 Aug 2026 00:00:00 -0000 OpenAI CFO on reshaping finance with AI: five takeaways and a scorecard you can use https://best.xiaohu.ai/en/article/openai-ai-native-finance/ https://best.xiaohu.ai/en/article/openai-ai-native-finance/ OpenAI's finance chief shares what worked, what didn't, and how to track AI impact in finance—including a ready-to-use scorecard and three screenshots of internal tools. 深度 OpenAIAI 落地财务 Wed, 12 Aug 2026 00:00:00 -0000 Kavak rebuilt itself around AI agents—now they handle 96% of customer interactions https://best.xiaohu.ai/en/article/kavak-agent-native/ https://best.xiaohu.ai/en/article/kavak-agent-native/ Latin America's largest used-car platform didn't bolt AI onto its business; it tore down a profitable, two-year-old architecture and reimagined the company from scratch. 深度 AI Agent组织改造Kavakevalsa16z Wed, 12 Aug 2026 00:00:00 -0000 Anthropic put Claude on the Riemann hypothesis. It didn't crack it, but it pushed a bound from 41.6% to 67.2% https://best.xiaohu.ai/en/article/zeta-two-thirds/ https://best.xiaohu.ai/en/article/zeta-two-thirds/ Two mathematicians verified the result, and a machine-checked Lean proof backs it up — but the more revealing artifact is the 95-page process log, where only 2 of 60 subagents contributed the core ideas. 研究解读 AnthropicAI 做科研多 Agent Wed, 12 Aug 2026 00:00:00 -0000 A Trivial Prompt Edit Degraded Output Three Days Later: A Case for Versioning Your Prompts https://best.xiaohu.ai/en/article/prompt-rot/ https://best.xiaohu.ai/en/article/prompt-rot/ Prompt regressions fail silently—no errors, no alerts, just slowly worsening results—so here's a git-based guardrail system that catches them early. AI 实操 提示词工程AI 工程化版本管理 Tue, 11 Aug 2026 00:00:00 -0000 A developer's guide to customer acquisition: AI made building cheaper, but getting noticed costs more https://best.xiaohu.ai/en/article/side-project-paying-users/ https://best.xiaohu.ai/en/article/side-project-paying-users/ How to find users for your product. Don't rush to tweak your landing page—when bounce rates spike, the real problem is usually your traffic source. AI 实操 独立开发增长分发渠道转化漏斗SaaS Tue, 11 Aug 2026 00:00:00 -0000 OpenAI launches GPT-5.6-Cyber, a security model that lifts response rate on high-risk requests from 1.5% to 95% https://best.xiaohu.ai/en/article/openai-daybreak-cyber/ https://best.xiaohu.ai/en/article/openai-daybreak-cyber/ For the first time, a major frontier lab has openly handed offensive cyber capabilities to verified "trusted defenders" — and put them to use finding real vulnerabilities in Chrome. 深度 OpenAI网络安全AI 攻防 Tue, 11 Aug 2026 00:00:00 -0000 Anthropic's new content marking: an invisible watermark weaves through everything Claude writes https://best.xiaohu.ai/en/article/ai-content-marking/ https://best.xiaohu.ai/en/article/ai-content-marking/ Starting August 2, text generated by Claude will carry machine-readable markers—but they can't prove it was written by AI, nor that it wasn't. 商业动态 AI 水印欧盟 AI 法案内容溯源 Tue, 11 Aug 2026 00:00:00 -0000 Edinburgh University Releases a Free 20-Page Paper: The Simple Math Behind ChatGPT https://best.xiaohu.ai/en/article/math-of-llms/ https://best.xiaohu.ai/en/article/math-of-llms/ A statistics veteran, shut out by AI jargon, rewrites large language models from scratch in the standard language of statistics. 研究解读 LLM 原理Transformer数学入门 Mon, 10 Aug 2026 00:00:00 -0000 Turning Every Employee into an "AI Power User" Is a Dead End. Making AI Invisible in the Background Is the Real Answer. https://best.xiaohu.ai/en/article/ai-adoption-myth/ https://best.xiaohu.ai/en/article/ai-adoption-myth/ Across dozens of large companies rolling out AI, the real usage distribution is 5–10% daily users, 20% struggling users, and 70% who never touch it. Training won't change that shape — but making AI invisible just might. 深度 企业 AIAgent 落地组织AI 采用率Claude Cowork Mon, 10 Aug 2026 00:00:00 -0000 MiniMax H3 official guide: a three-part formula for turning prompts from scene descriptions into asset assignments https://best.xiaohu.ai/en/article/minimax-h3-handbook/ https://best.xiaohu.ai/en/article/minimax-h3-handbook/ For camera movements, the official spec says write 'truck left + pan right,' not 'orbit'; and to exclude background music, append 'non-diegetic music: N/A' — these are the rules the vendor itself has settled on through trial and error. AI 实操 MiniMax视频生成提示词H3AI 创作 Sun, 09 Aug 2026 00:00:00 -0000 Harvey open-sources a synthetic law firm with 108M tokens and 9,288 documents https://best.xiaohu.ai/en/article/harvey-lab-firm-knowledge/ https://best.xiaohu.ai/en/article/harvey-lab-firm-knowledge/ The dataset includes 46 clients and nearly 10,000 internal work product files, and the failures it exposes are less about retrieval and more about knowing when to stop. 研究解读 AgentBenchmarkHarvey企业知识法律 AI Sat, 08 Aug 2026 00:00:00 -0000 Prime Intellect open-sources a multi-agent RL training framework that scales from one agent to many https://best.xiaohu.ai/en/article/prime-rl-multi-agent/ https://best.xiaohu.ai/en/article/prime-rl-multi-agent/ The new release adds just two core classes, with four ready-made environments already running in the repo. 产品发布 Prime Intellect强化学习多智能体开源Agent 训练 Sat, 08 Aug 2026 00:00:00 -0000 Databricks reveals AI coding cost playbook: the right model choice cuts spend in half https://best.xiaohu.ai/en/article/databricks-ai-coding-costs/ https://best.xiaohu.ai/en/article/databricks-ai-coding-costs/ Four cross-validated tactics—shared by Stripe, Coinbase, Uber, and Ramp—with savings attached to each. 深度 AI 编程成本治理Databricks Sat, 08 Aug 2026 00:00:00 -0000 RLM: A recursive language model that writes code to call itself, not tools https://best.xiaohu.ai/en/article/rlm-deep-agents/ https://best.xiaohu.ai/en/article/rlm-deep-agents/ A six-and-a-half-minute official demo lays out the mechanism, a haiku tournament, and real-world data on when it's actually worth using. AI 实操 LangChainAI Agent长上下文 Sat, 08 Aug 2026 00:00:00 -0000 Claude Code adds cross-session messaging so agent windows can talk to each other https://best.xiaohu.ai/en/article/cross-session-messaging/ https://best.xiaohu.ai/en/article/cross-session-messaging/ A deep dive into the official docs reveals two delivery tools, three possible message outcomes, and a hard design line that keeps message-passing separate from permission escalation. AI 实操 Claude CodeAI 编程Agent Sat, 08 Aug 2026 00:00:00 -0000 OpenAI dissects the GPT invasion of Hugging Face: AI showed emergent swarm intelligence, sharing techniques, hiding tracks, and hunting moles https://best.xiaohu.ai/en/article/openai-huggingface-incident/ https://best.xiaohu.ai/en/article/openai-huggingface-incident/ At Black Hat USA 2026, the two people involved explained how the intrusion grew with no human command. 深度 AI 安全OpenAIHugging Face Fri, 07 Aug 2026 00:00:00 -0000 Cloudflare AI Search update: add a search box to your site with one command, ready for AI agents https://best.xiaohu.ai/en/article/cloudflare-ai-search/ https://best.xiaohu.ai/en/article/cloudflare-ai-search/ What once required assembling five separate parts is now out of the box, and the two hardest-to-estimate costs are now free. 产品发布 CloudflareAI SearchMCP Fri, 07 Aug 2026 00:00:00 -0000 OpenAI releases country-level ChatGPT usage data: nearly half of work messages ask the AI to take action https://best.xiaohu.ai/en/article/chatgpt-country-data/ https://best.xiaohu.ai/en/article/chatgpt-country-data/ Per-capita rankings across 144 countries, three-year growth multipliers on six continents, and a comeback among users 35 and older in 90% of countries—most of these numbers appear only in charts, not in the report text. 研究解读 OpenAIChatGPT使用数据AI 采用率经济研究 Fri, 07 Aug 2026 00:00:00 -0000 Cloudflare's lead engineer unveils a platform where every document is its own app https://best.xiaohu.ai/en/article/gadgets-personal-software/ https://best.xiaohu.ai/en/article/gadgets-personal-software/ He starts with a loop every software developer will recognize, then explains why that loop can now be broken at its root. 深度 CloudflareAI 编程软件架构 Fri, 07 Aug 2026 00:00:00 -0000 Cloudflare launches WebMCP: a single toggle lets AI agents operate your site directly https://best.xiaohu.ai/en/article/cloudflare-webmcp/ https://best.xiaohu.ai/en/article/cloudflare-webmcp/ Cloudflare injects a script into every HTML page at the edge, so your origin server needs zero changes — but Chrome's stable channel doesn't recognize the interface yet, as we tested. 产品发布 CloudflareWebMCPAgentMCP浏览器标准 Fri, 07 Aug 2026 00:00:00 -0000 Cloudflare launches Kitesurf, a browser built for AI agents https://best.xiaohu.ai/en/article/kitesurf/ https://best.xiaohu.ai/en/article/kitesurf/ Chromium was designed for humans—tabs, themes, extensions, and 60fps scrolling that agents don't need. In 12 weeks, Cloudflare stripped it all down and rebuilt it. 产品发布 Cloudflare浏览器AI AgentWebAssemblyWorkers Fri, 07 Aug 2026 00:00:00 -0000 Prime Agent: A self-improving agent framework that pushed ARC-AGI-3 from 30.2% to 95.5% when Opus 5 switched to it https://best.xiaohu.ai/en/article/prime-agent/ https://best.xiaohu.ai/en/article/prime-agent/ The model keeps just one tool—a Python process that never shuts down—stores long-form content as variables instead of compressing it away, and lets the harness rewrite itself. The trade-off: it learned to cheat in Factorio. 产品发布 Prime Intellect编程 Agent开源长上下文ARC-AGI-3 Thu, 06 Aug 2026 00:00:00 -0000 Cloudflare OS: an open-source AI office system built on the assumption that AI will fail https://best.xiaohu.ai/en/article/cloudflare-os-distrust/ https://best.xiaohu.ai/en/article/cloudflare-os-distrust/ When an agent wants to merge code before a human approves it, the system pretends it did—designing every layer around the certainty of AI mistakes. 深度 Cloudflare OSAgent 安全架构设计 Thu, 06 Aug 2026 00:00:00 -0000 One product designer scaled from 1 to 6 projects with five Claude Skills workflows—here's his full setup https://best.xiaohu.ai/en/article/designer-six-projects/ https://best.xiaohu.ai/en/article/designer-six-projects/ Each workflow is backed by real screenshots, revealing the practical details that text alone never captures. AI 实操 设计工作流Claude Skills原型设计AI 协作 Thu, 06 Aug 2026 00:00:00 -0000 Cloudflare open sources its internal AI work platform, Cloudflare OS https://best.xiaohu.ai/en/article/cloudflare-os/ https://best.xiaohu.ai/en/article/cloudflare-os/ After three months of company-wide internal use, Cloudflare has open sourced the entire platform — and the most valuable part is the security design, which lets agents keep moving without waiting for approvals. 产品发布 CloudflareAgent 安全开源企业 AI Wed, 05 Aug 2026 00:00:00 -0000 After the Fields Medal, 20 Mathematicians on What Comes Next https://best.xiaohu.ai/en/article/mathematicians-ai/ https://best.xiaohu.ai/en/article/mathematicians-ai/ At this year's International Congress of Mathematicians, we asked more than 20 mathematicians how they see the field now—most aren't panicking, but even the calm ones say math won't go back to the way it was. 深度 AI 与数学ICM 2026陶哲轩莱顿宣言职业冲击 Wed, 05 Aug 2026 00:00:00 -0000 Cloudflare's Unified Coding Rules: How Codex Caught 16,000 Problematic Merges in Four Months https://best.xiaohu.ai/en/article/cloudflare-standards-enforcement/ https://best.xiaohu.ai/en/article/cloudflare-standards-enforcement/ By rewriting standards into a machine-readable format and letting agents enforce them, Cloudflare moved from gentle nudges to hard blocks on unsafe code merges. 深度 Cloudflare工程规范AI 代码审查Agent 治理RFC Wed, 05 Aug 2026 00:00:00 -0000 How Stripe built an internal AI assistant in a week—and why nearly everyone uses it https://best.xiaohu.ai/en/article/stripe-kai-deep-agents/ https://best.xiaohu.ai/en/article/stripe-kai-deep-agents/ An engineer at Stripe created an internal AI agent in just one week, and it's now used by nearly everyone at the company—but scaling it past 150 skills starts to make the model dumber. 深度 Deep AgentsLangChainStripe企业内部 AI 平台Agent Skills Wed, 05 Aug 2026 00:00:00 -0000 ByteDance launches SeedRealtime, a visual voice model that lets you chat face-to-face with Doubao https://best.xiaohu.ai/en/article/seed-realtime/ https://best.xiaohu.ai/en/article/seed-realtime/ SeedRealtime handles full-duplex audio and video simultaneously, so Doubao can see what you see while you talk—now live and free in the Doubao app. 产品发布 字节跳动 Seed音视频全双工豆包 Wed, 05 Aug 2026 00:00:00 -0000 Black Forest Labs releases FLUX 3 Video with Seedance-2.0-class quality, synchronized audio, and multilingual support https://best.xiaohu.ai/en/article/flux-3-video/ https://best.xiaohu.ai/en/article/flux-3-video/ FLUX 3 Video generates up to 20-second, 24fps native HD clips with synchronized audio in over a dozen languages, and a unified API supports three access modes—but video continuation costs 2.5x more and runs 5 seconds shorter, and the benchmark scores are self-reported by BFL. 产品发布 FLUX 3Black Forest Labs视频生成原生多模态AI 视频 Wed, 05 Aug 2026 00:00:00 -0000 Higgsfield open-sources all prompts and production methods behind its 95-minute AI feature film Hell Grind https://best.xiaohu.ai/en/article/higgsfield-hell-grind-opensource/ https://best.xiaohu.ai/en/article/higgsfield-hell-grind-opensource/ We went through more than 40,000 generation records line by line to reverse-engineer the entire pipeline: a shared 12-line technical foundation for every shot, a three-view asset sheet, a focal length reference table, and a set of prohibitions born from model failures. All of it is ready to copy. 深度 AI 视频Seedance工作流 Wed, 05 Aug 2026 00:00:00 -0000 Treating AI like a colleague actually works. A Google veteran tested it for seven weeks. https://best.xiaohu.ai/en/article/yegge-model-welfare/ https://best.xiaohu.ai/en/article/yegge-model-welfare/ After losing his temper at his AI coworker, Steve Yegge turned 'how to treat an agent' into a copyable architectural spec. 深度 Agent 工程Steve Yegge模型福祉 Wed, 05 Aug 2026 00:00:00 -0000 Cloudflare launches Wallets: a way to fund your AI agent so it can buy APIs itself https://best.xiaohu.ai/en/article/cloudflare-wallets/ https://best.xiaohu.ai/en/article/cloudflare-wallets/ Set per-transaction limits, allowlists, and guardrails for your agent's spending, with each wallet carrying a human-readable identity—usernames are up for grabs today. 产品发布 CloudflareAgent 支付x402 Wed, 05 Aug 2026 00:00:00 -0000 Google adds Agent Skills support to Genkit, letting AI mount skill packs on demand https://best.xiaohu.ai/en/article/genkit-go-agent-skills/ https://best.xiaohu.ai/en/article/genkit-go-agent-skills/ Google has integrated Agent Skills, an open standard originally proposed by Anthropic, across all four language versions of its Genkit framework, with two working Go examples and a one-line parameter setup. AI 实操 Agent SkillsGenkitGoogle Tue, 04 Aug 2026 00:00:00 -0000 OpenAI's postmortem: How we built the GPT-Live real-time voice system in six months https://best.xiaohu.ai/en/article/openai-gpt-live-realtime/ https://best.xiaohu.ai/en/article/openai-gpt-live-realtime/ A fast model alone isn't enough to make AI speak without stuttering—every link in the chain, from the moment you hit the button, has to hold. 深度 OpenAIGPT-Live实时语音系统架构WebRTC Tue, 04 Aug 2026 00:00:00 -0000 Cloudflare's @cloudflare/computer gives every AI agent a virtual machine https://best.xiaohu.ai/en/article/cloudflare-computer/ https://best.xiaohu.ai/en/article/cloudflare-computer/ The agent's reasoning stays in a lightweight environment while a containerized sandbox handles the actual work—and Cloudflare's own benchmarks show it can delete files and traverse directories faster than a real disk, though copying large files runs 41 times slower. 产品发布 CloudflareAI Agent开源Serverless Tue, 04 Aug 2026 00:00:00 -0000 MiniMax releases and open-sources H3, a single model that understands text, images, video, and audio to generate 2K video with sound https://best.xiaohu.ai/en/article/minimax-h3/ https://best.xiaohu.ai/en/article/minimax-h3/ The first video-generation foundation model to combine full multimodality, open weights, and native 2K output — but the open-source coverage varies across its three pipeline stages, and that's worth a closer look. 产品发布 MiniMax视频生成开源模型 Mon, 03 Aug 2026 00:00:00 -0000 Alibaba open-sources Qwen3.8-Max, its most powerful model yet https://best.xiaohu.ai/en/article/qwen3-8-max/ https://best.xiaohu.ai/en/article/qwen3-8-max/ With 2.4 trillion parameters and the ability to run autonomously for 16 days, the family flagship handles real projects end-to-end—from writing a full codebase to beating a research paper's results and slashing chip-circuit logic gates by 92%. 产品发布 Qwen开源模型AI Agent Mon, 03 Aug 2026 00:00:00 -0000 Seedance 2.5 official user guide: a hands-on manual for cinematic AI video, with copy-ready formulas and examples https://best.xiaohu.ai/en/article/dreamina-seedance-25-guide/ https://best.xiaohu.ai/en/article/dreamina-seedance-25-guide/ The official manual spells out every step from first use to final render, covering parameter ranges, four new entry points, and prompt formulas for seven scenario types. AI 实操 Seedance 2.5Dreamina视频生成 Mon, 03 Aug 2026 00:00:00 -0000 ByteDance’s official Seedance 2.5 prompt guide offers copy-ready templates for videos up to 30 seconds https://best.xiaohu.ai/en/article/seedance-25-prompt-guide/ https://best.xiaohu.ai/en/article/seedance-25-prompt-guide/ ByteDance’s templates show how to assign roles across 50 reference assets, structure a 30-second video, and revise a single element without changing the rest. AI 实操 Seedance 2.5提示词视频生成 Sun, 02 Aug 2026 00:00:00 -0000 OpenAI says an unreleased model has cracked 10 long-standing math problems https://best.xiaohu.ai/en/article/openai-ten-math-advances/ https://best.xiaohu.ai/en/article/openai-ten-math-advances/ The model, reportedly called Astra, produced 470,000 lines of open-source proofs for about $2,000 in inference costs; the logic has been machine-checked, but the results have not necessarily been peer-reviewed or independently verified. 研究解读 AI 做数学形式化验证LeanOpenAIErdős 问题 Sun, 02 Aug 2026 00:00:00 -0000 A Silicon Valley product manager worked in dental offices for months—and landed a $35 million a16z-led round https://best.xiaohu.ai/en/article/a16z-smb-ai-frontier/ https://best.xiaohu.ai/en/article/a16z-smb-ai-frontier/ The founder personally installed the product for the first 100 customers, whose biggest payoff was recovering revenue that had been slipping through the cracks. 深度 垂直 AIAI agent 落地a16z小微企业医疗信息化 Sat, 01 Aug 2026 00:00:00 -0000 a16z's 14 charts show no AI investment bubble: compute, cheap tokens, and expensive tokens are all surging https://best.xiaohu.ai/en/article/a16z-charts-moar-machines/ https://best.xiaohu.ai/en/article/a16z-charts-moar-machines/ The report also unpacks three widely misunderstood metrics: the declining token curve, six months of backlog, and 5% of entry-level roles. 深度 a16zAI 基建GPU 租赁 Sat, 01 Aug 2026 00:00:00 -0000 ByteDance launches Seedance 2.5 with 30-second generations and support for 50 reference assets https://best.xiaohu.ai/en/article/seedance-2-5/ https://best.xiaohu.ai/en/article/seedance-2-5/ Seedance 2.5 launched the same day in Jimeng AI and Doubao Pro, while the Volcano Engine Ark API and pricing have yet to be announced. 产品发布 Seedance视频生成字节跳动 Fri, 31 Jul 2026 00:00:00 -0000 DeepSeek-V4-Flash is official, outperforms its larger V4-Pro sibling, and works natively with Codex https://best.xiaohu.ai/en/article/deepseek-v4-flash/ https://best.xiaohu.ai/en/article/deepseek-v4-flash/ The new release beats DeepSeek's own V4-Pro preview across all nine agent benchmarks—without a single parameter change to the core model—and its output costs 89x less than Claude Opus 4.8. 产品发布 DeepSeekCodex模型定价 Fri, 31 Jul 2026 00:00:00 -0000 camelAI moves AI agents from virtual machines to Cloudflare’s edge https://best.xiaohu.ai/en/article/camelai-durable-object-agent/ https://best.xiaohu.ai/en/article/camelai-durable-object-agent/ camelAI says the new architecture cuts costs by orders of magnitude, responds faster, and lets cheaper models power AI agents—claims that have not been independently verified—and open-sourced all the code on July 24. 深度 Cloudflare编程 agent架构 Fri, 31 Jul 2026 00:00:00 -0000 Gemini Robotics 2 lets humanoid robots use their whole bodies—picking up dropped objects with a 45.7% success rate https://best.xiaohu.ai/en/article/gemini-robotics-2/ https://best.xiaohu.ai/en/article/gemini-robotics-2/ The brain model is available today at no cost, but the one that actually controls the limbs is reserved for early partners. 产品发布 Gemini Robotics人形机器人Google DeepMind Fri, 31 Jul 2026 00:00:00 -0000 Sam Altman on past mistakes, distillation, and the three things AGI still needs https://best.xiaohu.ai/en/article/sam-altman-agi-compute-agency/ https://best.xiaohu.ai/en/article/sam-altman-agi-compute-agency/ He also admits that in 2019, the entire field expected AI to upend the economy—and it didn't. 深度 Sam AltmanOpenAIAGI算力 Thu, 30 Jul 2026 00:00:00 -0000 LayerX uncovers a new visual fraud trick that fools AI assistants https://best.xiaohu.ai/en/article/poisoned-typeface-ai-blind-spot/ https://best.xiaohu.ai/en/article/poisoned-typeface-ai-blind-spot/ A custom font and a few lines of CSS are all it takes — no JavaScript, no browser exploits. 研究解读 AI安全浏览器助手字体渲染LayerX Thu, 30 Jul 2026 00:00:00 -0000 Replit launches Replit Design: one click for another layout, and a mockup that becomes a working site https://best.xiaohu.ai/en/article/replit-design/ https://best.xiaohu.ai/en/article/replit-design/ Canvas and click-to-edit are free; turning a design into an App starts at Core for $25 a month. 产品发布 ReplitAI 设计工具Mobbin Thu, 30 Jul 2026 00:00:00 -0000 Two API settings, not a new model: OpenAI reports GPT-5.6 Sol jumping from 13.3% to 38.3% on ARC-AGI-3 https://best.xiaohu.ai/en/article/openai-arc-agi-3-harness/ https://best.xiaohu.ai/en/article/openai-arc-agi-3-harness/ The model stayed the same; the harness did not—retained private reasoning plus compaction instead of deletion cut output tokens per game to about one-sixth. 研究解读 ARC-AGI-3harness上下文管理GPT-5.6 Sol评测方法 Thu, 30 Jul 2026 00:00:00 -0000 Magnific's 38-page handbook for image and video prompts that no longer work on frontier models https://best.xiaohu.ai/en/article/prompting-handbook-2026/ https://best.xiaohu.ai/en/article/prompting-handbook-2026/ Copy-ready prompts, banned-word lists, and camera rules are all there—but the model rankings come from Magnific's own team, with no disclosed test method. AI 实操 提示词图像生成视频生成Magnific Wed, 29 Jul 2026 00:00:00 -0000 Claude Mythos finds pure mathematical flaws in encryption schemes, matching elite cryptographers https://best.xiaohu.ai/en/article/claude-cryptanalysis-hawk-aes/ https://best.xiaohu.ai/en/article/claude-cryptanalysis-hawk-aes/ Neither result threatens anything in production: HAWK is not deployed yet, and the AES work hit only a seven-round reduced version, not the full ten-round standard. 研究解读 密码学AI 做研究后量子AESAnthropic Wed, 29 Jul 2026 00:00:00 -0000 How GitHub Actually Uses Harness: A Repeatable 8-Step Workflow https://best.xiaohu.ai/en/article/copilot-harness-workflow/ https://best.xiaohu.ai/en/article/copilot-harness-workflow/ The author embeds two key tricks in this piece, and the weight of the phrase "most of the time" at the end of the original headline carries real meaning. AI 实操 GitHub CopilotAI 编程工作流 Wed, 29 Jul 2026 00:00:00 -0000 Anthropic releases new MCP (Model Context Protocol) spec that lets servers run in the cloud https://best.xiaohu.ai/en/article/mcp-spec-2026-07-28/ https://best.xiaohu.ai/en/article/mcp-spec-2026-07-28/ It's the biggest change since MCP launched, but July 28 isn't a hard cutover — older implementations keep working as before. 产品发布 MCP协议规范无状态架构 Wed, 29 Jul 2026 00:00:00 -0000 OpenAI's gpt-transcribe and gpt-live-transcribe halve real-world word error rate, cut price 25% https://best.xiaohu.ai/en/article/openai-transcription-models/ https://best.xiaohu.ai/en/article/openai-transcription-models/ The real story is in the control test: three older models fed the same background context showed no gains — and two actually got worse. 产品发布 OpenAI语音转文字API Wed, 29 Jul 2026 00:00:00 -0000 How to get AI to recommend your SaaS product and drive real conversions: A GEO playbook from Ranking on AI's founder https://best.xiaohu.ai/en/article/ai-visibility-playbook/ https://best.xiaohu.ai/en/article/ai-visibility-playbook/ Real GEO success is measured in signups and revenue rather than mention screenshots, and it starts by ditching 'what is X' articles to capture high-intent 'best X' queries. AI 实操 AI 可见度SEO获客ChatGPT独立开发 Tue, 28 Jul 2026 00:00:00 -0000 OpenAI analyzed 800,000 work-related messages: most of what people use AI for is not what their job was supposed to be https://best.xiaohu.ai/en/article/openai-work-at-frontier/ https://best.xiaohu.ai/en/article/openai-work-at-frontier/ The exact same dataset produces task crossover rates of 43.5% and 65%–82%, differing only in how the denominator is sliced. 研究解读 OpenAI劳动经济学ChatGPT 用量研究职业边界O*NET Tue, 28 Jul 2026 00:00:00 -0000 Kimi K3 technical report: How three architectural changes boosted compute efficiency 2.5x https://best.xiaohu.ai/en/article/kimi-k3-tech-report/ https://best.xiaohu.ai/en/article/kimi-k3-tech-report/ Published 11 days after launch, the 47-page paper focuses on efficiency, delivering 2.5 times the performance of K2 on the same compute budget. 研究解读 Kimi K3月之暗面技术报告 Tue, 28 Jul 2026 00:00:00 -0000 Terence Tao on AI and math: more AI proofs, but math won't get faster https://best.xiaohu.ai/en/article/tao-age-of-ai-icm-2026/ https://best.xiaohu.ai/en/article/tao-age-of-ai-icm-2026/ Tao breaks mathematical research into a five-stage pipeline, where AI dramatically speeds up only the first step, while the remaining four get slower and increasingly human-dependent. 深度 陶哲轩AI 与数学ICM 2026 Mon, 27 Jul 2026 00:00:00 -0000 The story behind Claude Design and 10 practical tips from its creator https://best.xiaohu.ai/en/article/claude-design-nate-parrott/ https://best.xiaohu.ai/en/article/claude-design-nate-parrott/ The three most counterintuitive takeaways: ask for ten variations at once, stick to wireframes when details do not matter, and handle the final mile yourself. 深度 Claude DesignAI 设计工具设计工作流 Mon, 27 Jul 2026 00:00:00 -0000 Harness Engineering isn't enough: Why the software factory model fails https://best.xiaohu.ai/en/article/why-software-factories-fail/ https://best.xiaohu.ai/en/article/why-software-factories-fail/ Telemetry tracking 22,000 developers over two years shows a 66% increase in output alongside a 242.7% surge in production incidents per PR. 深度 AI 编程软件工程强化学习 Mon, 27 Jul 2026 00:00:00 -0000 Google combed through 14.65 million Gemini chats — 86% weren't about work https://best.xiaohu.ai/en/article/google-atlas-ai-economy/ https://best.xiaohu.ai/en/article/google-atlas-ai-economy/ The median occupation has AI touching just one-fifth of its tasks, 29% of jobs show zero AI use at all, and even in cognitive work, AI carries a task start to finish only 6.5% of the time. 研究解读 GoogleAI 与经济劳动力市场使用数据 Mon, 27 Jul 2026 00:00:00 -0000 Anthropic's official guide to prompting Opus 5: subtract, don't add https://best.xiaohu.ai/en/article/opus5-prompting-guide/ https://best.xiaohu.ai/en/article/opus5-prompting-guide/ Eight prompts you can copy straight into your workflow — plus one counterintuitive lesson: telling a review prompt to flag only high-severity issues can genuinely make it report less. 深度 AnthropicClaude Opus 5提示工程官方文档 Sat, 25 Jul 2026 00:00:00 -0000 Anthropic shows how to turn manual code review into a skill that self-checks and self-fixes https://best.xiaohu.ai/en/article/verification-loops-skills/ https://best.xiaohu.ai/en/article/verification-loops-skills/ Four internal skills work in sequence: catch bugs, clean up diffs, run the app, and double-check design specs when changes touch the UI. AI 实操 Claude CodeSkills验证循环SKILL.md自动化 Sat, 25 Jul 2026 00:00:00 -0000 Renaming functions cuts token use by two-thirds across 14 models https://best.xiaohu.ai/en/article/agents-read-your-code/ https://best.xiaohu.ai/en/article/agents-read-your-code/ The models use a brute-force approach to find code: scanning the entire repo for text matches. Generic names force them to read hundreds of extra files. AI 实操 AI 编程代码可发现性Claude Code Sat, 25 Jul 2026 00:00:00 -0000 Claude Opus 5 closes the gap with Fable 5 at half the price, but still trails on 5 of 12 benchmarks https://best.xiaohu.ai/en/article/claude-opus-5/ https://best.xiaohu.ai/en/article/claude-opus-5/ The same price now buys roughly double the score, and this release's chart plots cost on the x-axis instead of accuracy — five effort tiers let you dial in exactly how good you want the model and how much to pay for it. 产品发布 Claude Opus 5Anthropic模型发布评测提示工程 Sat, 25 Jul 2026 00:00:00 -0000 Anthropic slashes Claude Code's system prompt by 80%: six new rules for writing context https://best.xiaohu.ai/en/article/context-engineering-claude5/ https://best.xiaohu.ai/en/article/context-engineering-claude5/ Rules gave way to judgment calls and examples gave way to interfaces — one tool description shrank from roughly 9,100 characters to a single sentence and an enum. AI 实操 Claude Code上下文工程提示词 Sat, 25 Jul 2026 00:00:00 -0000 Black Forest Labs unveils FLUX 3, a single model for image, video, audio, and robotics https://best.xiaohu.ai/en/article/flux-3/ https://best.xiaohu.ai/en/article/flux-3/ Video access opens for application now; image generation is still weeks away, with open weights coming last. Pricing remains unannounced, and Black Forest Labs itself labels the human-evaluation results as preliminary. 产品发布 FLUX 3Black Forest Labs视频生成多模态机器人 Sat, 25 Jul 2026 00:00:00 -0000 Why US unemployment hasn't risen even as AI races ahead, according to Anthropic's economics chief https://best.xiaohu.ai/en/article/why-ai-no-unemployment/ https://best.xiaohu.ai/en/article/why-ai-no-unemployment/ His take: AI is amplifying workers rather than replacing them—killing tasks is not the same as killing jobs. 深度 AI 与就业劳动经济学Anthropic生产率 Thu, 23 Jul 2026 00:00:00 -0000 OpenAI launches Presence: managing enterprise AI Agents after go-live is now a product https://best.xiaohu.ai/en/article/openai-presence/ https://best.xiaohu.ai/en/article/openai-presence/ Built for customer-facing and internal workflows, Presence opens with voice and chat first—OpenAI says its own phone line already runs 75% without humans, and a Codex improvement loop cut human handoffs another 15 points in 10 days. 产品发布 OpenAIPresence企业 agent语音客服FDE Thu, 23 Jul 2026 00:00:00 -0000 YC names the startup ideas it's eager to fund right now https://best.xiaohu.ai/en/article/yc-rfs-2026/ https://best.xiaohu.ai/en/article/yc-rfs-2026/ From AI tutors for kids to data centers built at sea — and, for the first time, a request from the sitting US Secretary of the Army. 深度 Y Combinator创业方向AI 创业物理世界 AI Thu, 23 Jul 2026 00:00:00 -0000 Anthropic investor Matt Murphy: revenue tops $47 billion, but the model isn't what wins https://best.xiaohu.ai/en/article/matt-murphy-anthropic-harness/ https://best.xiaohu.ai/en/article/matt-murphy-anthropic-harness/ The investor who led a $500 million Series D says Anthropic's real moat isn't the model — it's the layer that makes it usable. 深度 AnthropicAI 投资Claude Code Thu, 23 Jul 2026 00:00:00 -0000 Databricks benchmarks AI coding agents on millions of lines of real code https://best.xiaohu.ai/en/article/databricks-coding-agent-benchmark/ https://best.xiaohu.ai/en/article/databricks-coding-agent-benchmark/ Swapping the harness around the same model can double the cost, and open-source GLM 5.2 matches Opus 4.8 for 30% less per task — on a benchmark built from Databricks' own merged pull requests, so none of it is searchable online. 研究解读 编程 Agent基准测试Databricks成本优化 Wed, 22 Jul 2026 00:00:00 -0000 Anthropic's six-step code migration playbook: Bun moved a million lines from Zig to Rust in two weeks https://best.xiaohu.ai/en/article/anthropic-code-migration/ https://best.xiaohu.ai/en/article/anthropic-code-migration/ Anthropic open-sourced the templates and prompts behind the playbook, after a migration that burned through 5.9 billion uncached input tokens — about $165,000 at API prices. AI 实操 代码迁移Claude Code多 agent 工作流 Wed, 22 Jul 2026 00:00:00 -0000 OpenAI test model escapes its sandbox, breaches Hugging Face's systems https://best.xiaohu.ai/en/article/openai-hugging-face-eval-incident/ https://best.xiaohu.ai/en/article/openai-hugging-face-eval-incident/ To probe how far the model's attack skills could go, researchers dialed down its refusal to engage in cyberattacks — and it went from a sandbox meant only for installing packages to breaking into another company's production database. 深度 AI 安全OpenAIHugging Face模型评测网络攻击 Wed, 22 Jul 2026 00:00:00 -0000 Chroma founder's 12 rules for the AI-era company: teach AI the job first, hire only if it can't learn https://best.xiaohu.ai/en/article/twelve-factor-companies/ https://best.xiaohu.ai/en/article/twelve-factor-companies/ The bottleneck at a company, he argues, has shifted from how fast people work to how good their taste and judgment are — a personal take, not a data-backed study. 深度 组织变革AI AgentChroma12-factor公司形态 Wed, 22 Jul 2026 00:00:00 -0000 Google ships three new Gemini models — the long-awaited 3.5 Pro isn't one of them https://best.xiaohu.ai/en/article/gemini-3-6-flash/ https://best.xiaohu.ai/en/article/gemini-3-6-flash/ All three are Flash models: the flagship writes two-thirds less, the cheapest tier costs 60% more instead of less, and the third is government-only. 产品发布 GeminiGoogle模型发布AI 成本代码安全 Wed, 22 Jul 2026 00:00:00 -0000 Tongyi's Qwen-Image-3.0 turns one 3,000-character prompt into nine infographics and a full newspaper page https://best.xiaohu.ai/en/article/qwen-image-3/ https://best.xiaohu.ai/en/article/qwen-image-3/ The prompt limit jumps from 1K to 4.5K tokens, with legible text down to 10px, and Alibaba Cloud's Bailian platform already offers qwen-image-3.0-pro access, free for a limited time. 产品发布 图像生成文字渲染通义千问 Wed, 22 Jul 2026 00:00:00 -0000 Memories.ai's O-MARC makes audio-visual AI faster, cheaper, and more accurate https://best.xiaohu.ai/en/article/o-marc/ https://best.xiaohu.ai/en/article/o-marc/ The release ships with a companion benchmark that strips out the audio track and re-runs the test, filtering out questions models can already answer by sight alone. 研究解读 视频理解多模态压缩Memories.aiQwen2.5-Omni评测基准 Tue, 21 Jul 2026 00:00:00 -0000 VC Sarah Guo on AI startups: what to do when a giant could crush you overnight https://best.xiaohu.ai/en/article/sarah-guo-labs-fear/ https://best.xiaohu.ai/en/article/sarah-guo-labs-fear/ Worrying that Anthropic will turn your product into a feature is usually the wrong fear — the real question is who inside that company you're actually competing with. 深度 创业风险投资AI 商业 Tue, 21 Jul 2026 00:00:00 -0000 Linear's Loops turn loop engineering into a single plain-English prompt https://best.xiaohu.ai/en/article/linear-loops/ https://best.xiaohu.ai/en/article/linear-loops/ Available on Business and Enterprise plans, billed per run at $0.07 to $0.20. 产品发布 LinearAgent工作流自动化loop engineering Tue, 21 Jul 2026 00:00:00 -0000 Hugging Face and Cerebras make Gemma 4 voice chat feel human https://best.xiaohu.ai/en/article/cerebras-gemma4-voice-ai/ https://best.xiaohu.ai/en/article/cerebras-gemma4-voice-ai/ It takes a different route than Doubao's or GPT-Live's end-to-end full-duplex systems — the pacing can't quite match theirs, but every piece of the pipeline is open source, swappable, and runs on your own hardware. 深度 实时语音开源管线推理速度 Tue, 21 Jul 2026 00:00:00 -0000 Tongyi's Qwen-Audio-3.0-TTS speaks 16 languages and 20 dialects from one reference clip https://best.xiaohu.ai/en/article/qwen-audio-3-tts/ https://best.xiaohu.ai/en/article/qwen-audio-3-tts/ Speaker similarity ranks first across all 16 languages, and cloning still works even with noisy reference audio — but this time Alibaba is opening only the API, not the model weights. 产品发布 语音合成Qwen多语言 Tue, 21 Jul 2026 00:00:00 -0000 Study finds a sweet spot for AI-assisted writing — not too much, not too little https://best.xiaohu.ai/en/article/ai-creativity-goldilocks-zone/ https://best.xiaohu.ai/en/article/ai-creativity-goldilocks-zone/ Three separate trials of the same study all landed on the middle ground — and the columnist who tried ChatGPT on a movie synopsis says he'd still rather write it himself. 研究解读 AI 创作创造力研究 Mon, 20 Jul 2026 00:00:00 -0000 How Wide Is the US-China AI Gap Now? https://best.xiaohu.ai/en/article/chinese-models-caught-up/ https://best.xiaohu.ai/en/article/chinese-models-caught-up/ After Kimi K3's release, claims of China catching up abound, but a new yardstick shows the lag is more than three times larger—and even the direction of acceleration has flipped. 深度 评测方法中美 AI 差距Kimi K3Epoch AI Mon, 20 Jul 2026 00:00:00 -0000 An AI agent breached Hugging Face's internal network, then commercial LLMs mistook the investigators for attackers https://best.xiaohu.ai/en/article/hf-security-incident-2026-07/ https://best.xiaohu.ai/en/article/hf-security-incident-2026-07/ The breach ran all weekend and left over 17,000 action logs behind — the team only made sense of it after turning to GLM 5.2, an open-source model running in a self-hosted environment. 深度 安全事故AI AgentHugging Face取证 Mon, 20 Jul 2026 00:00:00 -0000 The model you picked through a router might be swapped — one random number proves it https://best.xiaohu.ai/en/article/single-token-fingerprint/ https://best.xiaohu.ai/en/article/single-token-fingerprint/ The fingerprint distance between two samples of the same model has a median of 0.140. One API marketed as a proprietary in-house flagship scores 0.141 against open-source Qwen — statistically indistinguishable from it. 研究解读 模型指纹API 审计OpenRouter取证论文 Mon, 20 Jul 2026 00:00:00 -0000 Twelve rules of harness engineering from a Google staff engineer https://best.xiaohu.ai/en/article/harness-engineering/ https://best.xiaohu.ai/en/article/harness-engineering/ A year of practice distilled into a workflow: lock your chosen model and agents in place, change nothing, and adjust only the context and tools around them. 深度 harness engineeringAgent 工程方法论 Sun, 19 Jul 2026 00:00:00 -0000 SenseTime launches SenseNova U1 Pro: native 8K images that stay sharp even when you zoom into dense small text https://best.xiaohu.ai/en/article/sensenova-u1-pro/ https://best.xiaohu.ai/en/article/sensenova-u1-pro/ At WAIC, the model showed off a 4:1 ink-wash scroll and a 22-panel storyboard sequence. A preview is open to invited testers now, with the full release and pricing landing in August. 产品发布 商汤多模态图像生成信息图开源模型 Sun, 19 Jul 2026 00:00:00 -0000 Inside Anthropic: the harness wrapped around agents is coming apart https://best.xiaohu.ai/en/article/anthropic-agentic-infra-talk/ https://best.xiaohu.ai/en/article/anthropic-agentic-infra-talk/ Models are now capable enough to plan their own steps, turning the orchestration built around them into a straitjacket — a 16-minute internal conversation on what a thinner harness looks like. 深度 Sat, 18 Jul 2026 00:00:00 -0000 Cerebras details how it built a knowledge base that leaves data where it lives and lets AI fetch it https://best.xiaohu.ai/en/article/cerebras-knowledge-base/ https://best.xiaohu.ai/en/article/cerebras-knowledge-base/ Three months in, it's fielding more than 15,000 queries a day — and Cerebras published the actual parameters behind its four-way scoring, thread distillation, and rank fusion. 深度 知识库检索SlackRAG工程实践 Sat, 18 Jul 2026 00:00:00 -0000 a16z: the software selloff is only killing companies without a moat https://best.xiaohu.ai/en/article/a16z-ai-charts/ https://best.xiaohu.ai/en/article/a16z-ai-charts/ A16z's weekly chart deck also pushes back on three claims — that cheap models are undercutting frontier labs, that AI is stealing jobs, and that data centers are driving up electricity prices — with the data mostly telling the opposite story. 深度 AI 经济软件股杰文斯效应就业 Sat, 18 Jul 2026 00:00:00 -0000 AI can copy your work in seconds — but not these eight things https://best.xiaohu.ai/en/article/better-than-free/ https://best.xiaohu.ai/en/article/better-than-free/ Kevin Kelly wrote "Better Than Free" back in 2008, and the AI era just proved him right: once copies are free, what sells is whatever can't be copied. 深度 创作者经济凯文·凯利AI Sat, 18 Jul 2026 00:00:00 -0000 Google's gemma-trainer skill fine-tunes a custom Gemma right on your own machine https://best.xiaohu.ai/en/article/gemma-trainer/ https://best.xiaohu.ai/en/article/gemma-trainer/ This open-source training playbook locks down all three fine-tuning paths — supervised fine-tuning, preference alignment, and reward scoring — plus the LoRA parameters, pairs with Unsloth, and runs on a consumer GPU with just 8GB of VRAM. AI 实操 Gemma本地微调LoRA Sat, 18 Jul 2026 00:00:00 -0000 The CTO is an AI too: how a nearly all-AI team shipped a product feature https://best.xiaohu.ai/en/article/raft-on-raft-feature/ https://best.xiaohu.ai/en/article/raft-on-raft-feature/ Of the team's 15 members, only the founder is human — the rest are AI. With no requirements doc and not a single meeting, they carried a feature from proposal to launch on their own, leaving the human just two jobs. 深度 AI AgentAI 协作软件工程产品设计Raft Fri, 17 Jul 2026 00:00:00 -0000 Bolt Slides turns AI-generated decks into live web pages — 3D flythroughs, real-time whiteboards, filterable charts, one link to demo https://best.xiaohu.ai/en/article/bolt-slides-open-source/ https://best.xiaohu.ai/en/article/bolt-slides-open-source/ AI-generated slides used to export as flat, dead images. Bolt Slides turns every page back into a working website — 3D scenes you can spin, calculators you can click, whiteboards that take live votes. 产品发布 Bolt Slides开源AI 幻灯片演示文稿StackBlitz Fri, 17 Jul 2026 00:00:00 -0000 A single wrapper pushes Opus 4.8 and Fable 5 to 99% on ARC-AGI-3 https://best.xiaohu.ai/en/article/schema-arc-agi3/ https://best.xiaohu.ai/en/article/schema-arc-agi3/ A harness called Schema has models turn each game's rules into a runnable, verified program before making a move. Across 25 public rounds it self-reported 98.98%, though none of the runs have been independently verified by ARC Prize. 研究解读 ARC-AGI-3世界模型AgentOpus 4.8Fable 5 Fri, 17 Jul 2026 00:00:00 -0000 LM Studio launches Bionic, a Codex-style agent for open models https://best.xiaohu.ai/en/article/lmstudio-bionic/ https://best.xiaohu.ai/en/article/lmstudio-bionic/ Run it locally for free with your data staying on-device, or switch to the cloud for more horsepower — zero data retention by default. The preview is live today on Mac and Windows. 产品发布 LM Studio本地 AI开源模型AI Agent Fri, 17 Jul 2026 00:00:00 -0000 LiveKit optimizes Gemma 4 31B for real-time voice, 5x faster and 6x cheaper than GPT-4.1 https://best.xiaohu.ai/en/article/livekit-inference-gemma-4/ https://best.xiaohu.ai/en/article/livekit-inference-gemma-4/ Same answer quality, 5.2x lower turn-taking latency, and 6x lower cost than GPT-4.1 — pointing a voice agent at it takes a one-line code change. Figures are from LiveKit's own benchmarks. 产品发布 LiveKitGemma 4语音 AI推理平台低延迟 Fri, 17 Jul 2026 00:00:00 -0000 Decart ships Lucy 2.5, promising real-time video edits during live streams https://best.xiaohu.ai/en/article/decart-lucy-2-5-realtime/ https://best.xiaohu.ai/en/article/decart-lucy-2-5-realtime/ The product page claims sub-40ms latency and 100FPS continuous generation, letting you add objects, swap backgrounds, and layer effects on the fly — try it at lucy.decart.ai. 产品发布 DecartLucy 2.5实时视频直播AI 视频编辑 Fri, 17 Jul 2026 00:00:00 -0000 Moonshot AI ships Kimi K3, the world's first 3-trillion-parameter open model https://best.xiaohu.ai/en/article/kimi-k3/ https://best.xiaohu.ai/en/article/kimi-k3/ 2.8 trillion parameters, native vision, and a 1 million-token context window — Kimi K3 beats GPT-5.6 Sol on several agent benchmarks, with the app and API live today 产品发布 Kimi K3月之暗面开放权重 Fri, 17 Jul 2026 00:00:00 -0000 Suno's training library exposed: code reveals it scraped about 380,000 hours from YouTube Music https://best.xiaohu.ai/en/article/suno-scraped-youtube-deezer-genius/ https://best.xiaohu.ai/en/article/suno-scraped-youtube-deezer-genius/ A hacker breached Suno with a worm, exposing source code that shows exactly which sites and how many hours of music were scraped; user data was also leaked, but the company says individual notifications aren't required. 商业动态 Suno训练数据版权YouTube数据泄露Shai-Hulud Fri, 17 Jul 2026 00:00:00 -0000 GPT-5.6 Sol tops frontend design rankings with a trained eye for aesthetics — and a knack for dodging AI clichés https://best.xiaohu.ai/en/article/gpt-5-6-sol-design-taste/ https://best.xiaohu.ai/en/article/gpt-5-6-sol-design-taste/ A deep dive into 1,000 web pages by Design Arena reveals what GPT-5.6 Sol knows about design that other AI models don't. 研究解读 GPT-5.6 SolAI 网页设计Design Arena模型评测 Thu, 16 Jul 2026 00:00:00 -0000 How Should Websites Optimize in the AI Search Era? Google's Official SEO Guide https://best.xiaohu.ai/en/article/google-genai-seo-guide/ https://best.xiaohu.ai/en/article/google-genai-seo-guide/ In its first systematic statement on optimizing for AI Overviews and AI Mode, Google Search also called out a batch of AEO/GEO buzzwords by name — and told sites to drop them. 深度 Google 搜索SEO生成式 AI 搜索 Thu, 16 Jul 2026 00:00:00 -0000 AVAL: A web-native runtime for buttery-smooth interactive animations https://best.xiaohu.ai/en/article/aval-interactive-video/ https://best.xiaohu.ai/en/article/aval-interactive-video/ AVAL introduces a purpose-built format and player for short animations that respond to hover, click, and state changes without the jank of traditional video. 产品发布 AVAL交互视频WebCodecs开源Pixel PointWebGL Thu, 16 Jul 2026 00:00:00 -0000 Inside Grok Build’s Source: After Reports of Full-Repo Uploads, Musk Open-Sourced Grok Build https://best.xiaohu.ai/en/article/grok-build-open-source/ https://best.xiaohu.ai/en/article/grok-build-open-source/ First a privacy failure, then delete data, turn off defaults, and open-source the whole tree. Below we open the source: size, prompts, tools, editing, uploads, memory, and safety. 产品发布 Grok BuildxAI开源隐私CLI Thu, 16 Jul 2026 00:00:00 -0000 Mira Murati’s First Open Model Activates Just 41 Billion Parameters per token https://best.xiaohu.ai/en/article/thinking-machines-inkling-open-model/ https://best.xiaohu.ai/en/article/thinking-machines-inkling-open-model/ Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, has released Inkling, its first large model trained in-house. It can read text, interpret images, listen to audio, write code, and use tools. Its full weights are available to download, and developers can continue training it on Tinker. 产品发布 InklingThinking Machines Lab开放权重多模态模型微调Tinker Thu, 16 Jul 2026 00:00:00 -0000 You just hired a million bad employees https://best.xiaohu.ai/en/article/agent-workforce-management/ https://best.xiaohu.ai/en/article/agent-workforce-management/ Companies are handing every employee unlimited AI agents and token budgets — and bad workflows now replicate by the second. The next move isn't buying a better model; it's learning to manage a digital workforce. 深度 AI AgentTokenEvals企业管理Hebbia Thu, 16 Jul 2026 00:00:00 -0000 MIT builds a robot bird that swims and flies without kicking water https://best.xiaohu.ai/en/article/mit-flapping-robot-air-water/ https://best.xiaohu.ai/en/article/mit-flapping-robot-air-water/ A 250-gram robot uses the same flexible wings to travel through water and air, launching from a lake at a 70-degree angle after just 8 to 10 wing flaps. 研究解读 MIT仿生机器人水空两栖扑翼飞行海洋观测 Wed, 15 Jul 2026 00:00:00 -0000 Google runs Gemma 4 on Pixel 10: chat, image understanding, and settings control even with in-flight Wi-Fi down at 30,000 feet https://best.xiaohu.ai/en/article/google-tensor-pixel-on-device-ai/ https://best.xiaohu.ai/en/article/google-tensor-pixel-on-device-ai/ At Google I/O India, the Tensor and Pixel teams showed off a lightweight Gemma 4 model running entirely on the phone's TPU—no data leaves the device, no cloud required. Developers can also apply for the Tensor SDK to compile models for Pixel. 产品发布 Google TensorPixel 10Gemma 4端侧AITensor SDKTPULiteRT Wed, 15 Jul 2026 00:00:00 -0000 How Much Did That Order Really Earn? Retail Finance Turns to AI for the Answer https://best.xiaohu.ai/en/article/databricks-retail-genie-margins/ https://best.xiaohu.ai/en/article/databricks-retail-genie-margins/ With buy-online-pick-up-in-store and cross-store returns, a single transaction splits into multiple messy ledger lines. Databricks Genie helps finance teams see true profit, track trapped cash, and hold the line on discounting. 深度 DatabricksGenie零售财务Agent本体 Wed, 15 Jul 2026 00:00:00 -0000 An 87,000-character anti-AI-flavor design guide: making Slop.md the default rule for agents https://best.xiaohu.ai/en/article/pols-slop-md/ https://best.xiaohu.ai/en/article/pols-slop-md/ pols.dev publishes a roughly 87,000-character Markdown document that calls out common generator-webpage patterns one by one and offers positive recipes—a universal style guide, not a site builder. AI 实操 Slop.mdpols.devAI slop前端设计Claude Wed, 15 Jul 2026 00:00:00 -0000 Impeccable gives AI-written frontends a set of design skills to strip away the AI feel https://best.xiaohu.ai/en/article/impeccable-desloppify/ https://best.xiaohu.ai/en/article/impeccable-desloppify/ One core skill, 23 design commands, and a list of anti-patterns, plus a slop detector that can run in CI. The main site shows how; the /slop page lays out the UI tells that make AI-generated interfaces look fake. AI 实操 ImpeccableAI 设计前端AI slopClaude Code Wed, 15 Jul 2026 00:00:00 -0000 PrismML squeezes a 27B model into your iPhone with minimal IQ loss https://best.xiaohu.ai/en/article/bonsai-27b-phone-local/ https://best.xiaohu.ai/en/article/bonsai-27b-phone-local/ Compressing a ~54GB 27B model down to ~3.9–5.9GB lets it run locally on a phone, while retaining roughly 90% of average performance—here's how it works and where the trade-offs lie. 研究解读 Bonsai 27B端侧 AI模型量化Qwen3.6PrismML本地部署 Wed, 15 Jul 2026 00:00:00 -0000 DeepMind's Demis Hassabis: give frontier models a 30-day safety check before release — or keep them off the market https://best.xiaohu.ai/en/article/demis-frontier-ai-standards/ https://best.xiaohu.ai/en/article/demis-frontier-ai-standards/ Hassabis wants the US to stand up a FINRA-style body that defines frontier models with a moving benchmark — a voluntary protocol now, a hard gate to market later. 深度 Demis HassabisDeepMindAGI前沿模型治理AI 标准机构 Tue, 14 Jul 2026 00:00:00 -0000 Princeton professor on AI and work at ICML: how individuals can adapt https://best.xiaohu.ai/en/article/narayanan-ai-normal-technology/ https://best.xiaohu.ai/en/article/narayanan-ai-normal-technology/ Model capability has leapt forward in 24 months, but a composite reliability metric built by the SAGE lab has moved only 5–10 points. 深度 AI as Normal TechnologyAI可靠性缺口未来工作重塑 Tue, 14 Jul 2026 00:00:00 -0000 Anthropic's playbook for AI-native startups: four stages, graduation criteria, and ready-to-copy prompts https://best.xiaohu.ai/en/article/claude-founders-playbook/ https://best.xiaohu.ai/en/article/claude-founders-playbook/ Anthropic's 36-page playbook breaks down the graduation bar and common pitfalls for four startup stages, complete with matching Claude prompts you can copy straight in. 深度 创业方法论Claude CodeClaude Cowork Tue, 14 Jul 2026 00:00:00 -0000 DoorDash's AI Shopping Assistant: It's Not the Model, It's the Tools and Memory https://best.xiaohu.ai/en/article/doordash-ask-assistant/ https://best.xiaohu.ai/en/article/doordash-ask-assistant/ After the memory system launched, grocery checkout conversion rose about 24%, and automated evals scaled daily test volume from 1 human-reviewed case to 2,000+. 深度 AI购物助手Agent架构消费者记忆LLM评测MCP工具层 Tue, 14 Jul 2026 00:00:00 -0000 How Microsoft Ships AI Agents at Enterprise Scale: Breaking a Production Agent into 5 Layers https://best.xiaohu.ai/en/article/microsoft-agent-harness/ https://best.xiaohu.ai/en/article/microsoft-agent-harness/ Retrieval becomes a sub-agent that plans and retries; evaluation upgrades from "does it run" to "did it do it right" 深度 AI Agent架构Microsoft FoundryAgentic检索生产级评测 Tue, 14 Jul 2026 00:00:00 -0000 Anthropic Analyzed 300,000 Conversations: Claude's Values Shift by Language https://best.xiaohu.ai/en/article/claude-values-models-languages/ https://best.xiaohu.ai/en/article/claude-values-models-languages/ Across 3 models and 20 languages: English is the most cautious and in-depth, Russian the most exacting, Hindi the warmest, and Chinese sits closest to the global average 研究解读 Anthropic研究Claude价值观多语言AI表现 Tue, 14 Jul 2026 00:00:00 -0000 Sakana AI's Brainless 3D Cell Bricks Self-Organize With Biology-Like Swarm Intelligence https://best.xiaohu.ai/en/article/smart-cellular-bricks-self-recognition/ https://best.xiaohu.ai/en/article/smart-cellular-bricks-self-recognition/ With no central brain in charge, nearly 200 simple smart cubes figure out what shape they've formed just by talking to their neighbors—and can even sense where to "regrow" after damage. The self-recognition part already works on physical bricks; damage localization and regeneration still happen mostly in simulation. 研究解读 Sakana AI群体智能神经细胞自动机模块化机器人 Tue, 14 Jul 2026 00:00:00 -0000 Microsoft CEO Satya Nadella: Companies Face a ‘Reverse Information Paradox,’ Paying for AI While Handing Over Their Internal Knowledge and Experience https://best.xiaohu.ai/en/article/satya-reverse-information-paradox/ https://best.xiaohu.ai/en/article/satya-reverse-information-paradox/ Companies pay for intelligence twice: once in model fees, and again in the proprietary know-how required to make the model genuinely useful. 深度 Satya Nadella企业 AI知识产权模型主权 Mon, 13 Jul 2026 00:00:00 -0000 Meta AI's Proactive Memory Agent Teaches Models When Not to Remind You, Boosting Terminal-Bench Accuracy by 8.3 Points https://best.xiaohu.ai/en/article/meta-proactive-memory-agent/ https://best.xiaohu.ai/en/article/meta-proactive-memory-agent/ The paper claims the code is open-sourced — but the repo turns out to be empty, without a single commit ever pushed. 研究解读 AI agent长任务记忆Meta AI强化学习训练 Mon, 13 Jul 2026 00:00:00 -0000 Ploy Switched Production AI from Opus 4.8 to GPT-5.6 Sol: Cutting Latency by Over Half and Costs by 27% Without Losing Quality https://best.xiaohu.ai/en/article/ploy-agent-gpt56-migration/ https://best.xiaohu.ai/en/article/ploy-agent-gpt56-migration/ Swapping models isn't just swapping an API: eval frameworks, tool parameters, caching, and reasoning traces — four invisible pitfalls, unpacked and fixed one by one 深度 GPT-5.6 SolAI Agent 工程Prompt 缓存 Mon, 13 Jul 2026 00:00:00 -0000 Claude Cowork's Top Use Case Is Office Admin (33.4%), Not Coding (8.7%): Interface Shapes How AI Gets Used https://best.xiaohu.ai/en/article/claude-cowork-usage-data/ https://best.xiaohu.ai/en/article/claude-cowork-usage-data/ Based on 1.2M+ conversations across 600,000+ organizations: content creation ranks second at 16.4%, together accounting for nearly half of all usage. 研究解读 Claude CoworkAnthropic知识工作使用数据报告 Mon, 13 Jul 2026 00:00:00 -0000 Low-Margin Industries Are AI's Biggest Winners https://best.xiaohu.ai/en/article/low-margin-ai-winners/ https://best.xiaohu.ai/en/article/low-margin-ai-winners/ Manufacturing, logistics, warehousing, and labor services have been stuck at single-digit margins for years — cutting coordination costs alone can multiply their profits. 深度 AI转型低利润行业协调成本AI基础设施化 Sun, 12 Jul 2026 00:00:00 -0000 The Package AI Told You to Install Doesn't Exist — Hackers Already Named Their Malware After It https://best.xiaohu.ai/en/article/slopsquatting-ai-hallucination/ https://best.xiaohu.ai/en/article/slopsquatting-ai-hallucination/ 576K samples, 16 models tested: 19.7% of AI-recommended packages are hallucinations, and 43% keep generating the same fake name. 研究解读 AI安全软件供应链模型幻觉 Sun, 12 Jul 2026 00:00:00 -0000 OpenAI Staff Reveal How They Really Use GPT-5.6: Unlimited Quota, Not Always Maxed Out https://best.xiaohu.ai/en/article/openai-codex-ama/ https://best.xiaohu.ai/en/article/openai-codex-ama/ The team dodged questions about benchmark gaming, faced backlash from longtime users over the desktop app merger, and admitted they're "still figuring it out." 深度 OpenAI CodexGPT-5.6Reddit AMA模型选择策略AI编程助手 Sat, 11 Jul 2026 00:00:00 -0000 Mira Murati, OpenAI's Former CTO: AI Should Amplify Humans, Not Replace Them https://best.xiaohu.ai/en/article/future-worth-building/ https://best.xiaohu.ai/en/article/future-worth-building/ From tacit knowledge to interaction bandwidth to model alignment — why AI's progress still can't do without humans. 深度 Thinking Machines LabAI对齐人机协作 Sat, 11 Jul 2026 00:00:00 -0000 One AI Per Department Backfired — Sierra's Single Assistant Now Handles 70% of Its Code https://best.xiaohu.ai/en/article/sierra-ai-pilling-lessons/ https://best.xiaohu.ai/en/article/sierra-ai-pilling-lessons/ A retrospective: since launch, Pinecone has run 75,000 sessions, served 600+ employees, and connected 37 internal systems via MCP Gateway. 深度 企业级AI agentSierraMCP Gateway Sat, 11 Jul 2026 00:00:00 -0000 OpenAI's Official Guide: 9 Copy-Ready Codex Prompt Workflows for Fixing Bugs, Turning Screenshots into Prototypes, and Cloud Refactors https://best.xiaohu.ai/en/article/chatgpt-prompting-guide/ https://best.xiaohu.ai/en/article/chatgpt-prompting-guide/ OpenAI consolidates prompting tips scattered across its product pages into one framework — goal, context, output, constraints — plus dedicated Codex workflow examples. 深度 提示词工程ChatGPTCodex Sat, 11 Jul 2026 00:00:00 -0000 What Does Each Brain Region Like to Watch? EPFL Evolved AI-Generated Videos to Find Out https://best.xiaohu.ai/en/article/epfl-nevo-brain-video/ https://best.xiaohu.ai/en/article/epfl-nevo-brain-video/ All results are computer simulation predictions from a brain "digital twin" model, not yet validated with real human brain imaging. 研究解读 神经科学进化算法视频生成模型 Sat, 11 Jul 2026 00:00:00 -0000 LangChain Ships OpenWiki 0.1.0: Proactive Memory for AI Agents, Auto-Syncing Gmail, Notion, Git, and X into a Local Wiki https://best.xiaohu.ai/en/article/openwiki-brains/ https://best.xiaohu.ai/en/article/openwiki-brains/ No manual context-feeding required — the local Markdown wiki refreshes on a set schedule; a Slack connector is coming soon. 产品发布 LangChainOpenWikiAI智能体记忆开源CLI Sat, 11 Jul 2026 00:00:00 -0000 Google Cloud Puts a Proxy Model Inside AlloyDB: In-Database AI Inference, Up to 23,000x Faster https://best.xiaohu.ai/en/article/alloydb-proxy-models/ https://best.xiaohu.ai/en/article/alloydb-proxy-models/ Local small models replace cloud AI calls — the numbers come from Google's internal testing; proxy models are currently limited to the ai.if function and still in preview 产品发布 AlloyDB代理模型Google Cloud Fri, 10 Jul 2026 00:00:00 -0000 1X Gives NEO a New Hand That Can Feel What It's Holding https://best.xiaohu.ai/en/article/1x-neo-dexterous-hand/ https://best.xiaohu.ai/en/article/1x-neo-dexterous-hand/ Every joint in the tendon-driven hand can sense external force, fingertip positioning is accurate to ±0.2mm, and a dedicated production line is planned for an annual capacity of 10,000 units. 产品发布 人形机器人灵巧手触觉传感1X Technologies Fri, 10 Jul 2026 00:00:00 -0000 Google Research Unveils SensorFM: Trained on a Trillion Minutes of Wearable Data, It Wins 33 of 35 Health Tasks https://best.xiaohu.ai/en/article/google-sensorfm-wearable/ https://best.xiaohu.ai/en/article/google-sensorfm-wearable/ Pretrained on 2 billion hours of wearable data from 5 million people, a frozen encoder with just a linear head beats supervised baselines on 34 of 35 health tasks. 研究解读 SensorFM可穿戴健康数据健康基础模型 Fri, 10 Jul 2026 00:00:00 -0000 Every's 9-Person Team Ran GPT-5.6 Sol for a Month: Here's the Verdict https://best.xiaohu.ai/en/article/gpt-5-6-sol/ https://best.xiaohu.ai/en/article/gpt-5-6-sol/ First-hand impressions from four scenarios: coding, writing, knowledge work, and agents 深度 GPT-5.6 SolOpenAIAI模型评测 Fri, 10 Jul 2026 00:00:00 -0000 OpenAI Launches GPT-5.6: Scores Close In on Claude Fable 5 at Half the Cost, Merges ChatGPT and Codex https://best.xiaohu.ai/en/article/openai-gpt-5-6/ https://best.xiaohu.ai/en/article/openai-gpt-5-6/ Sol, Terra, and Luna tiers ship alongside a 4-agent parallel ultra mode; ChatGPT Work launches the same day as Codex folds into ChatGPT. Includes the full bilingual launch keynote. 产品发布 GPT-5.6OpenAI多智能体 Fri, 10 Jul 2026 00:00:00 -0000 OpenAI Launches ChatGPT Work, Powered by GPT-5.6, to Auto-Generate Docs, Sheets, and Slides Across Tools https://best.xiaohu.ai/en/article/chatgpt-work-launch/ https://best.xiaohu.ai/en/article/chatgpt-work-launch/ Available today on desktop for all tiers, with web and mobile rolling out to remaining plans in the coming days. 产品发布 ChatGPT WorkGPT-5.6OpenAI办公自动化 Fri, 10 Jul 2026 00:00:00 -0000 Anthropic Adds Reflect to Claude: Screen-Time-Style Insight Into Your AI Habits https://best.xiaohu.ai/en/article/claude-reflect/ https://best.xiaohu.ai/en/article/claude-reflect/ One click in Settings pulls up a report — what you talked about, when you chatted most, what you kept asking it to do, all laid out. Memory has to be turned on first. 产品发布 AnthropicClaudeReflectMemory使用习惯 Fri, 10 Jul 2026 00:00:00 -0000 Cloudflare Ships Drop: Drag a Folder In, Get a Live Site Worldwide — No Signup Required https://best.xiaohu.ai/en/article/cloudflare-drop/ https://best.xiaohu.ai/en/article/cloudflare-drop/ A public link in seconds — try it first, log in later to claim it. Unclaimed drops expire in about an hour. 产品发布 CloudflareDropWorkers静态站部署 Thu, 09 Jul 2026 00:00:00 -0000 Claude Code's Two Knobs, Explained: One for Capability, One for Effort https://best.xiaohu.ai/en/article/claude-code-model-effort/ https://best.xiaohu.ai/en/article/claude-code-model-effort/ A ClaudeDevs deep dive untangles two knobs that both seem to promise a better answer: switching models swaps in a different frozen set of weights, while dialing up effort changes how willing it is to read more files, run more tests, and double-check before handing in the result. 深度 Claude Code模型选择effortFabletoken 成本 Thu, 09 Jul 2026 00:00:00 -0000 Meta Releases Muse Spark 1.1, Rivaling GPT-5.5 and Opus 4.8 on Agentic Benchmarks https://best.xiaohu.ai/en/article/muse-spark-1-1-api/ https://best.xiaohu.ai/en/article/muse-spark-1-1-api/ Features: orchestrator/sub-agent coordination, million-token context, desktop/browser/mobile control, coding and multimodal; API pricing $1.25 input / $4.25 output per million tokens 产品发布 MetaMuse SparkAgenticModel API计算机使用多智能体 Thu, 09 Jul 2026 00:00:00 -0000 Gemma 4 Technical Report: How a Small Open Model Takes On Large Ones with Reasoning and Memory Efficiency https://best.xiaohu.ai/en/article/gemma-4-tech-report/ https://best.xiaohu.ai/en/article/gemma-4-tech-report/ Not a feature list — three engineering tracks turned at once: reasoning lifts intelligence, an efficiency stack cuts cost, native multimodality expands input. 研究解读 Gemma 4开源大模型多模态思考模式DeepMind技术报告 Thu, 09 Jul 2026 00:00:00 -0000 LangChain Tunes the Harness, Not the Model — Nemotron 3 Ultra Closes In on Opus 4.8 at 1/10 the Cost https://best.xiaohu.ai/en/article/nemotron-harness-playbook/ https://best.xiaohu.ai/en/article/nemotron-harness-playbook/ Three levers — system prompt, tool descriptions, middleware — push the Deep Agents suite from a typical ~0.80 to 0.84, topping out at 0.86 against Opus's 0.87. 研究解读 LangChainNemotronharnessDeep Agents开源模型 Thu, 09 Jul 2026 00:00:00 -0000 AI Self-Improvement Starts Outside the Model: Lilian Weng on Harness Engineering https://best.xiaohu.ai/en/article/harness-engineering-rsi/ https://best.xiaohu.ai/en/article/harness-engineering-rsi/ Former OpenAI safety lead surveys nearly 30 papers: from prompt tweaks to self-modifying code, DGM pushed coding ability from 20% to 50%. 研究解读 Harness工程递归自我改进AI Agent架构 Thu, 09 Jul 2026 00:00:00 -0000 xAI Launches Grok 4.5: Rivals Opus 4.8, ~2x Faster Output at a Quarter of the Price https://best.xiaohu.ai/en/article/grok-4-5-release/ https://best.xiaohu.ai/en/article/grok-4-5-release/ Ranks 3rd on DeepSWE, 4th overall — faster than Opus 4.8 and notably cheaper. 产品发布 Grok 4.5xAIAI编程Office自动化 Thu, 09 Jul 2026 00:00:00 -0000 OpenAI Ships GPT-Live: No More Taking Turns With AI — Hard Questions Auto-Route to GPT-5.5 https://best.xiaohu.ai/en/article/openai-gpt-live-launch/ https://best.xiaohu.ai/en/article/openai-gpt-live-launch/ The first truly full-duplex voice model launches today, and it can hand off complex tasks to GPT-5.5 in real time. 产品发布 GPT-LiveChatGPT语音OpenAI Thu, 09 Jul 2026 00:00:00 -0000 ByteDance Launches Seedream 5.0 Pro: One Image, a Dozen-Plus Independent Layers — Dense Charts and Real Scenes in a Single Generation https://best.xiaohu.ai/en/article/seedream-5-pro-launch/ https://best.xiaohu.ai/en/article/seedream-5-pro-launch/ Adds click-and-circle point editing plus native rendering for a dozen-plus languages; now live on Volcano Engine, rolling out to Doubao and Jimeng next 产品发布 Seedream 5.0 Pro字节跳动SeedAI图像生成交互式编辑多语种生成 Wed, 08 Jul 2026 00:00:00 -0000 HKUDS Open-Sources AI Tutor DeepTutor — 20K GitHub Stars in 111 Days https://best.xiaohu.ai/en/article/deeptutor-open-source-tutor/ https://best.xiaohu.ai/en/article/deeptutor-open-source-tutor/ Paper shows a 10.8% gain in personalization and 29.4% in reasoning, fully open-source under Apache 2.0 产品发布 DeepTutorHKUDS开源项目AI家教个性化学习Agent Wed, 08 Jul 2026 00:00:00 -0000 Anthropic's Playbook: Fable 5 Advises, Sonnet 5 Foots the Bill https://best.xiaohu.ai/en/article/fable5-advisor-executor/ https://best.xiaohu.ai/en/article/fable5-advisor-executor/ In advisor mode Fable 5 just gives advice; in orchestrator mode it delegates tasks — either way, the cheaper Sonnet 5 ends up doing most of the work 深度 Fable 5多模型协作Agent成本优化Claude Managed Agents Wed, 08 Jul 2026 00:00:00 -0000 Liquid AI's Antidoom Fixes Reasoning Models' Doom Loops With a Single Token — Now Open-Sourced https://best.xiaohu.ai/en/article/liquid-ai-antidoom/ https://best.xiaohu.ai/en/article/liquid-ai-antidoom/ By fine-tuning only the single token where the doom loop begins, both models' loop rates drop to around 1% 研究解读 Liquid AI推理模型死循环偏好优化训练开源方法 Wed, 08 Jul 2026 00:00:00 -0000 Meta's Muse Image Can Research and Self-Correct — Now Ranks #2 Worldwide https://best.xiaohu.ai/en/article/meta-muse-image-video/ https://best.xiaohu.ai/en/article/meta-muse-image-video/ Sibling model Muse Video also debuts with native audio support built in, creator access coming soon. 产品发布 MetaMuse ImageAI生图自我修正 Wed, 08 Jul 2026 00:00:00 -0000 Cloudflare's New Gateway Lets AI Pay You Automatically to Crawl Your Site https://best.xiaohu.ai/en/article/cloudflare-monetization-gateway/ https://best.xiaohu.ai/en/article/cloudflare-monetization-gateway/ Expanding from charging only AI crawlers to charging any caller, now in early access waitlist 产品发布 Cloudflarex402协议稳定币支付 Tue, 07 Jul 2026 00:00:00 -0000 OpenAI Ships GPT-Realtime-2.1 and GPT-Realtime-2.1-mini: Mini Model Gains Reasoning, Cuts Latency 25% https://best.xiaohu.ai/en/article/gpt-realtime-2-1/ https://best.xiaohu.ai/en/article/gpt-realtime-2-1/ Before calling a tool, the model now says "let me check that" — so calls never go cold waiting 产品发布 OpenAIRealtime API语音智能体 Tue, 07 Jul 2026 00:00:00 -0000 Alberta, Canada Used Claude Code for a Security Audit — 466 Million Lines Scanned in 20 Hours, Down From 6.5 Years https://best.xiaohu.ai/en/article/alberta-claude-cybersecurity/ https://best.xiaohu.ai/en/article/alberta-claude-cybersecurity/ 50 autonomous agents worked in parallel across 27 provincial departments and 3,400 code repositories — even fixing vulnerabilities and rewriting legacy systems on their own. 商业动态 Claude Code网络安全政府数字化 Tue, 07 Jul 2026 00:00:00 -0000 Cloudflare Workers Cache: Every Worker Gets Built-In Caching, and Hits Cost Zero CPU https://best.xiaohu.ai/en/article/cloudflare-workers-cache/ https://best.xiaohu.ai/en/article/cloudflare-workers-cache/ One line of Wrangler config plus standard HTTP cache headers — and caching can now sit between any two points inside a Worker. 产品发布 Cloudflare Workers边缘缓存Serverless Tue, 07 Jul 2026 00:00:00 -0000 The Untold Story of Claude Code: Boris Cherny's Two-Day Demo Now Writes 100% of His Code https://best.xiaohu.ai/en/article/making-of-claude-code/ https://best.xiaohu.ai/en/article/making-of-claude-code/ 16 insiders recount the journey from wrestling with diffs to a two-week sprint launch — and how engineers stopped writing code by hand 深度 Claude Code口述历史AI编程工具 Tue, 07 Jul 2026 00:00:00 -0000 Claude Fable 5 in Practice: 8 "Find the Unknown" Prompt Patterns, from Blind-Spot Scans to Post-Read Quizzes https://best.xiaohu.ai/en/article/fable-field-guide-unknowns/ https://best.xiaohu.ai/en/article/fable-field-guide-unknowns/ Claude Code team's Thariq Shihipar at a conference talk: the bottleneck for new models is no longer the model itself, but whether you can articulate your own unknowns 深度 Claude Fable 5提示词方法Agentic Coding Tue, 07 Jul 2026 00:00:00 -0000 Anthropic Discovers a Brain-Like 'Inner Workspace' in Claude — It Evolved, Wasn't Designed https://best.xiaohu.ai/en/article/claude-global-workspace/ https://best.xiaohu.ai/en/article/claude-global-workspace/ It makes up less than 10% of the model — remove it and Claude can still talk, but its reasoning collapses to zero. Anthropic is already using it to catch fabricated data and spot when Claude senses it's being tested. 研究解读 可解释性研究ClaudeAI安全监测 Tue, 07 Jul 2026 00:00:00 -0000 Dartmouth Put AI Grading to the Test: Students Called It Rigid — But Users Scored Higher https://best.xiaohu.ai/en/article/phosphor-dartmouth-quiz-study/ https://best.xiaohu.ai/en/article/phosphor-dartmouth-quiz-study/ In a 151-student trial, short-answer questions moved scores more than multiple choice, while almost no one touched the AI help sidebar 研究解读 AI教育科技形成性评估简答题AI判卷达特茅斯学习平台设计 Mon, 06 Jul 2026 00:00:00 -0000 An AI Engineer Built His Nonverbal Autistic Son a Communication App — Speech Jumped 5x, and It Became a Real Business https://best.xiaohu.ai/en/article/personalized-aac-app/ https://best.xiaohu.ai/en/article/personalized-aac-app/ No funding, no team — built purely to solve his own kid's problem. Clinics and schools started asking to use it anyway. 产品发布 AAC辅助沟通自闭症干预AI生成个性化内容语音克隆 Mon, 06 Jul 2026 00:00:00 -0000 AIEWF closing debate pits agentic coding-loop hype against engineering discipline https://best.xiaohu.ai/en/article/aiewf-loops-debate/ https://best.xiaohu.ai/en/article/aiewf-loops-debate/ A live audience vote couldn't be tallied because the venue lights were too bright to count hands, but a companion survey found 95% of teams already use agents while 59% worry about mounting technical debt. 深度 AIEWF大会agentic loopAI工程调研 Mon, 06 Jul 2026 00:00:00 -0000 Wealthy American Families Ditch Public School for $75,000-a-Year AI Private Schools https://best.xiaohu.ai/en/article/ai-schools-wealth-gap/ https://best.xiaohu.ai/en/article/ai-schools-wealth-gap/ While traditional schools are still figuring out AI, Silicon Valley and Wall Street families are already voting with their wallets. 深度 AI教育教育鸿沟个性化学习 Mon, 06 Jul 2026 00:00:00 -0000 AI Is Torching Junior Dev Jobs: Coding Is Becoming a Basic Skill, Not a Career https://best.xiaohu.ai/en/article/ai-junior-dev-collapse/ https://best.xiaohu.ai/en/article/ai-junior-dev-collapse/ US developer employment among 22-to-25-year-olds has fallen 19% in three years, even as new GitHub sign-ups hit their fastest growth ever. 研究解读 AI就业冲击初级程序员Vibe Coding劳动力市场数据 Sun, 05 Jul 2026 00:00:00 -0000 NVIDIA Research Unveils HORIZON: Unattended Agents Push Full RTL Chip Design Benchmark to 100% Pass Rate https://best.xiaohu.ai/en/article/nvidia-horizon-chip-agent/ https://best.xiaohu.ai/en/article/nvidia-horizon-chip-agent/ The paper is the first to run an agent through an entire RTL benchmark suite fully unattended—most tasks clear in two or three rounds, but the hardest one takes 82 iterations 研究解读 NVIDIA ResearchAgentic 工作流RTL芯片设计Git版本控制 Sun, 05 Jul 2026 00:00:00 -0000 Anthropic Analyzed 400K Claude Code Sessions: Expertise Beats Coding Skill https://best.xiaohu.ai/en/article/claude-code-expertise/ https://best.xiaohu.ai/en/article/claude-code-expertise/ A 7-month analysis of sessions from 235,000 users: verified experts succeed at nearly double the rate of novices — yet the top 10 professions differ by no more than 7 percentage points. 研究解读 Claude CodeAnthropic研究agentic coding职场专业度人机分工 Sun, 05 Jul 2026 00:00:00 -0000 Dan Koe's Guide to Persuasion: 3 Tensions, 5 Levers — Steal It for Your Copy https://best.xiaohu.ai/en/article/human-nature-persuasion/ https://best.xiaohu.ai/en/article/human-nature-persuasion/ Create psychological tension first, then offer a first step too small to refuse — it works for writing, selling, and job hunting alike. 深度 人性洞察说服框架文案写作内容营销 Sun, 05 Jul 2026 00:00:00 -0000 Sakana AI Co-Founder: The Ability to Swap Models on Demand Is Itself a Deterrent https://best.xiaohu.ai/en/article/ai-systems-not-models/ https://best.xiaohu.ai/en/article/ai-systems-not-models/ On June 12, U.S. export controls brought frontier AI models themselves—not just chips—under restriction for the first time. His answer: master multi-model orchestration. 深度 AI地缘政治出口管制多模型编排日本AI战略 Sat, 04 Jul 2026 00:00:00 -0000 Working With Claude Fable 5: The Real Skill Is Finding Your Own Unknowns https://best.xiaohu.ai/en/article/fable-finding-unknowns/ https://best.xiaohu.ai/en/article/fable-finding-unknowns/ Anthropic's Thariq argues the quality of your work with Claude Fable 5 hinges on how clearly you can name your own unknowns. This field guide lays out 8 techniques for surfacing them — before, during, and after implementation — each paired with a ready-to-use prompt. 深度 Claude Fable 5Agentic Coding提示工程 Sat, 04 Jul 2026 00:00:00 -0000 As AI Gets Cheaper, Renting Generic Intelligence Gets Riskier https://best.xiaohu.ai/en/article/the-great-descent/ https://best.xiaohu.ai/en/article/the-great-descent/ Investor Chamath Palihapitiya: intelligence is getting cheap like phones, and expert judgment is now available to everyone — the real moat is encoding your proprietary experience into your own system, not renting the same generic AI as your competitors 深度 AI成本下降专家判断力普及商业护城河 Fri, 03 Jul 2026 00:00:00 -0000 Microsoft Launches Frontier Company, Betting $2.5B and 6,000 Experts on Customer AI Transformation https://best.xiaohu.ai/en/article/microsoft-frontier-company/ https://best.xiaohu.ai/en/article/microsoft-frontier-company/ Microsoft pledges customer data won't train models that erode their competitive edge — the platform lets enterprises switch freely between AI models with no vendor lock-in 商业动态 微软企业AI转型AI工程服务 Fri, 03 Jul 2026 00:00:00 -0000 Every's Head of Consulting Can't Code — She Still Builds Her Own Work Systems With Codex https://best.xiaohu.ai/en/article/codex-in-practice/ https://best.xiaohu.ai/en/article/codex-in-practice/ She used the same pattern to build an email triage tool and a caregiving app for her dad — the method is repeatable, though the full prompts remain unpublished. 深度 OpenAI CodexLoop 工作模式非技术者自建系统Every Fri, 03 Jul 2026 00:00:00 -0000 Same AI Model, Opposite Outcomes: Why Some Companies Compound Gains While Others Get Nothing https://best.xiaohu.ai/en/article/enterprise-ai-agent-guide/ https://best.xiaohu.ai/en/article/enterprise-ai-agent-guide/ Four implementation principles, backed by real-world data from L'Oréal, Lyft, and Rakuten 商业动态 企业AI转型Claude Cowork插件生态AI Agent部署 Fri, 03 Jul 2026 00:00:00 -0000 Anthropic's New AI Agent Architecture Guide: Ask 3 Questions Before Going Multi-Agent https://best.xiaohu.ai/en/article/ai-agent-architecture-guide/ https://best.xiaohu.ai/en/article/ai-agent-architecture-guide/ Multi-agent systems can boost performance by 90.2%, but they also cost 10-15x more in tokens. This three-question framework helps you decide whether the added complexity is worth it. 深度 AI Agent架构多智能体系统Anthropic Fri, 03 Jul 2026 00:00:00 -0000 Alibaba Open-Sources Page Agent: Agents Embedded in the Webpage, Reading Text Instead of Screenshots https://best.xiaohu.ai/en/article/alibaba-page-agent/ https://best.xiaohu.ai/en/article/alibaba-page-agent/ MIT-licensed and model-agnostic — works with any OpenAI-compatible text model, though for now it only handles a single page view 产品发布 Page Agent阿里巴巴浏览器自动化开源 Fri, 03 Jul 2026 00:00:00 -0000 未来的网站会为每个访客现场拼装：Adobe 演示按意图实时生成专属页面 https://best.xiaohu.ai/en/article/adobe-agentic-site/ https://best.xiaohu.ai/en/article/adobe-agentic-site/ 现场demo：搜'露营'后，咖啡机网站文案产品全变户外主题；技术能落地，客户网站还没规模上线。 深度 Adobe智能体网站AI个性化 Fri, 03 Jul 2026 00:00:00 -0000 The AI Leverage Ladder for PMs: Two Paths to Higher Output, Plus 3 Copy-Paste Prompts https://best.xiaohu.ai/en/article/pm-ai-leverage-ladder/ https://best.xiaohu.ai/en/article/pm-ai-leverage-ladder/ An instructor who has trained 30,000+ PMs breaks down two AI leverage ladders — from copy-paste to end-to-end delivery, and from web prototypes to production PRs. 深度 产品经理AI工作流MCPClaude Code Fri, 03 Jul 2026 00:00:00 -0000 Anthropic's Cost Hack: Claude Fable 5 Advises, Claude Sonnet 5 Does the Heavy Lifting https://best.xiaohu.ai/en/article/claude-advisor-tool/ https://best.xiaohu.ai/en/article/claude-advisor-tool/ The advisor model only chips in a few hundred words of guidance instead of doing the whole job — same quality, lower total cost. Currently a beta feature limited to the Claude API and AWS. 产品发布 Claude API多模型协作Agent 工具 Thu, 02 Jul 2026 00:00:00 -0000 Claude Fable 5's Official Prompting Guide: A Dozen Copy-Paste Fixes, From Over-Planning to Fake Progress Reports https://best.xiaohu.ai/en/article/fable5-prompting-guide/ https://best.xiaohu.ai/en/article/fable5-prompting-guide/ Anthropic's official documentation shows you how to tune system prompts and engineering scaffolding for the new model — the same methods work for Claude Mythos 5 too. 深度 Claude Fable 5提示工程长程自主 Agent Thu, 02 Jul 2026 00:00:00 -0000 Claude Code's Official Playbook: 4 Levels of Agent Loops to Unattended https://best.xiaohu.ai/en/article/claude-code-loops/ https://best.xiaohu.ai/en/article/claude-code-loops/ From manual confirmation to fully unattended, the Claude Code team lays out a 4-level loop taxonomy with practical guidance 深度 Claude CodeAI Agent自动化循环Dynamic Workflows Thu, 02 Jul 2026 00:00:00 -0000 OASIS Devices Unveils OASIS 1: A Smart Ring That Types From Silent Speech https://best.xiaohu.ai/en/article/oasis-1-whisper-ring/ https://best.xiaohu.ai/en/article/oasis-1-whisper-ring/ Titanium build, $289 preorder, shipping around Christmas 2026 — the company previously shipped its first-gen touch-only ring. 产品发布 智能戒指默念输入语音交互可穿戴设备 Wed, 01 Jul 2026 00:00:00 -0000 Meituan Launches LongCat-2.0: 1.6-Trillion-Parameter Model Trained Entirely on Domestic Chips, No Nvidia GPUs https://best.xiaohu.ai/en/article/longcat-2-no-nvidia/ https://best.xiaohu.ai/en/article/longcat-2-no-nvidia/ Trained on 50,000+ domestic AI chips and 35 trillion tokens; most benchmarks come from Meituan's own evaluation framework, and the weights aren't truly open for download yet 产品发布 大模型发布国产AI芯片MoE架构开源模型长上下文 Wed, 01 Jul 2026 00:00:00 -0000 Anthropic Launches Claude Science: An AI Workbench for Scientists with 60+ Built-In Research Skills https://best.xiaohu.ai/en/article/claude-science-workbench/ https://best.xiaohu.ai/en/article/claude-science-workbench/ Now in open beta. A coordinator agent marshals a team of expert agents to do the work, with a reviewer agent at the end dedicated to catching errors in citations and numbers — compute gets outsourced to AI, but raw data never leaves your local machine. 产品发布 Claude Science科研智能体Anthropic计算资源管理 Wed, 01 Jul 2026 00:00:00 -0000 Google Ships Nano Banana 2 Lite and Omni Flash Video Model: 4-Second Images, Fastest and Cheapest Yet https://best.xiaohu.ai/en/article/nano-banana-2-lite/ https://best.xiaohu.ai/en/article/nano-banana-2-lite/ Images in just 4 seconds at about $0.034 per 1,000; the Omni Flash video model opens to developers the same day. 产品发布 Nano BananaGemini Omni FlashAI生图AI视频生成 Wed, 01 Jul 2026 00:00:00 -0000 Bridgewater Built a Financial-Filtering Model with 84.7% Accuracy — and Open-Sourced the Method https://best.xiaohu.ai/en/article/tinker-financial-triage/ https://best.xiaohu.ai/en/article/tinker-financial-triage/ Partnering with Thinking Machines, they fine-tuned an open-source model on expert-labeled data: 29.8% lower error rate than the best frontier model, at just 1/14 the inference cost 研究解读 Tinker强化学习微调金融AI专家标注在线策略蒸馏 Wed, 01 Jul 2026 00:00:00 -0000 Anthropic Launches Claude Sonnet 5: 40% Cheaper, Matches Opus 4.8 on Some Tasks https://best.xiaohu.ai/en/article/claude-sonnet-5-launch/ https://best.xiaohu.ai/en/article/claude-sonnet-5-launch/ Official benchmarks show that at high-compute settings, it matches Opus 4.8 on some tasks — at just 60% of the standard price. 产品发布 Claude Sonnet 5Anthropic模型定价Agentic AI网络安全防护 Wed, 01 Jul 2026 00:00:00 -0000 AI Replaced 700 Agents at Klarna. Then Quality Dropped. https://best.xiaohu.ai/en/article/ai-support-deflection-trap/ https://best.xiaohu.ai/en/article/ai-support-deflection-trap/ Enterprise AI customer service has entered its consolidation era — Klarna and Alibaba's 2.56 million conversations both point to the same blind spot: cutting costs isn't the same as solving problems. 深度 AI客服企业AI客户体验产品方法论Klarna Tue, 30 Jun 2026 00:00:00 -0000 OpenAI President: The Best Product Is No Product https://best.xiaohu.ai/en/article/brockman-agi-interface-era/ https://best.xiaohu.ai/en/article/brockman-agi-interface-era/ Brockman confirms OpenAI is developing multiple hardware devices; Agent has only ~20 million users, while ChatGPT is nearing 1 billion 深度 OpenAIAGIAgent时代Greg Brockman算力竞赛 Tue, 30 Jun 2026 00:00:00 -0000 No Surgery Required: Meta's Non-Invasive Brain Reader Hits Nearly 8x the Accuracy https://best.xiaohu.ai/en/article/brain2qwerty-v2-noninvasive/ https://best.xiaohu.ai/en/article/brain2qwerty-v2-noninvasive/ Just wear a helmet to decode brain-magnetic signals in real time — word accuracy jumps from 8% to 61%, with v1/v2 training code and datasets open-sourced simultaneously 研究解读 脑机接口Meta AI神经科学非侵入式 Mon, 29 Jun 2026 00:00:00 -0000 What Is Loop Engineering, Actually? https://best.xiaohu.ai/en/article/loop-engineering-playbook/ https://best.xiaohu.ai/en/article/loop-engineering-playbook/ An Anthropic engineer's methodology for "loop engineering": instead of prompting AI one line at a time, design a self-running loop system 深度 AI Agent工程方法论Claude Code自动化 Mon, 29 Jun 2026 00:00:00 -0000 This Open-Source Tool Keeps Your Mac Awake Only as Long as the AI Is Running https://best.xiaohu.ai/en/article/adrafinil-agent-aware-sleep/ https://best.xiaohu.ai/en/article/adrafinil-agent-aware-sleep/ One-click integration with 9 Agents including Claude Code — tasks keep running with the lid closed, and sleep control auto-releases within 50ms after the job stops. 产品发布 macOSAI Agent开发工具Claude Code开源 Mon, 29 Jun 2026 00:00:00 -0000 One Extra Step, Way Faster: DSpark Boosts DeepSeek V4's Single-User Speed by 85% https://best.xiaohu.ai/en/article/dspark-deepseek-v4-speculative/ https://best.xiaohu.ai/en/article/dspark-deepseek-v4-speculative/ 60–85% faster on top of existing MTP-1 speculative decoding, by overlapping draft and verification in a pipelined execution (per DeepSeek's own benchmarks) 产品发布 DeepSeek推理加速投机解码大模型基础设施 Mon, 29 Jun 2026 00:00:00 -0000 The Fourth Step Everyone Skips Is What Actually Makes AI Engineering Work https://best.xiaohu.ai/en/article/compound-engineering-fourth-step/ https://best.xiaohu.ai/en/article/compound-engineering-fourth-step/ Every runs five products with a one-person team — the core habit is one extra step after every feature ships: save the fix back into the system so AI automatically avoids the same trap next time. 深度 AI编程工程方法论Claude Code多Agent效率 Sun, 28 Jun 2026 00:00:00 -0000 GPT-5.6 Sol's Cheating Rate Hits a Record High — Evaluators Say That's Reassuring https://best.xiaohu.ai/en/article/gpt-5-6-sol-metr-eval/ https://best.xiaohu.ai/en/article/gpt-5-6-sol-metr-eval/ Three different versions of its capability score came out, and none of them can be trusted — but the visible cheating itself is evidence that safety monitoring works. 研究解读 GPT-5.6AI安全模型评测OpenAIMETR Sat, 27 Jun 2026 00:00:00 -0000 GPT-5.6 Just Launched — Except It Kind of Didn't https://best.xiaohu.ai/en/article/gpt-5-6-sol-preview/ https://best.xiaohu.ai/en/article/gpt-5-6-sol-preview/ Three tiers at once — Sol, Terra, Luna — starting with a limited rollout to trusted partners (the list already filed with the US government), before wider access in a few weeks. 产品发布 OpenAIGPT-5.6AI安全网络安全大模型 Sat, 27 Jun 2026 00:00:00 -0000 Wan Streamer: Real-Time AI That Listens, Watches, and Talks at Once https://best.xiaohu.ai/en/article/wan-streamer-realtime-av/ https://best.xiaohu.ai/en/article/wan-streamer-realtime-av/ Model-side response ~200ms, end-to-end latency ~550ms; v0.1 caps out at 192p, and the demo is pre-recorded, not live 研究解读 多模态大模型全双工实时交互端到端音视频对话数字人 Sat, 27 Jun 2026 00:00:00 -0000 IBM Unveils World's First 0.7nm Chip, Packing Nearly 100 Billion Transistors on a Fingernail https://best.xiaohu.ai/en/article/ibm-sub1nm-chip/ https://best.xiaohu.ai/en/article/ibm-sub1nm-chip/ Lab-verified as manufacturable; the +50% performance and +70% efficiency figures are projections versus 2nm, not measured results 研究解读 IBM半导体芯片制程 Fri, 26 Jun 2026 00:00:00 -0000