Python · ★ 12,374 · 🍴 807 · 📈 3,139 stars today
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 60-95% fewer tokens, same answers. Library, proxy, MCP server.
中文介绍 在内容送达 LLM 之前对其进行压缩的工具，支持处理工具输出、日志、文件和 RAG 数据块，可减少 60–95% 的 token 用量，同时保持回答质量不变。提供 Library、代理和 MCP Server 三种集成方式，适合需要控制 LLM 调用成本的开发者和企业用户。
Python · ★ 180,910 · 🍴 31,033 · 📈 1,951 stars today
The agent that grows with you
中文介绍 NousResearch 推出的自适应 AI Agent 框架，定位为"随用户成长的 Agent"，可根据使用场景持续演进。适合希望构建个性化、可扩展 AI 助手的研究人员和开发者，依托 Hermes 系列模型的工具调用与推理能力提供底层支撑。
JavaScript · ★ 207,172 · 🍴 31,805 · 📈 1,736 stars today
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
中文介绍 面向 AI 编程 Agent 的性能优化与增强框架，覆盖技能扩展、本能反应、记忆管理和安全机制，采用研究优先的开发理念。兼容 Claude Code、Codex、Opencode、Cursor 等主流编程 Agent，帮助开发者提升 Agent 在复杂编程任务中的执行效率和可靠性。
Python · ★ 79,831 · 🍴 10,597 · 📈 105 stars today
Turn any PDF or image document into structured data for your AI. A powerful, lightweight OCR toolkit that bridges the gap between images/PDFs and LLMs. Supports 100+ languages.
中文介绍 基于 PaddlePaddle 的轻量级强大 OCR 工具包，可将 PDF 和图像文档转换为结构化数据，充当图像/PDF 与 LLM 之间的桥梁。支持 100 余种语言，适合文档智能化、数据抽取等 AI 应用场景，广泛用于企业文档处理和多语言内容识别。
Python · ★ 108,549 · 🍴 9,594 · 📈 311 stars today
💫 Toolkit to help you get started with Spec-Driven Development
中文介绍 GitHub 官方推出的规格驱动开发（Spec-Driven Development）工具套件，提供起步模板和工具链，帮助开发团队在编码前先撰写清晰的功能规格文档。适合希望通过规范化需求定义来提升协作效率和代码质量的工程团队。
Jupyter Notebook · ★ 8,973 · 🍴 579 · 📈 244 stars today
NVIDIA Cosmos is an open platform of world models, datasets, and tools that enables developers to build Physical AI for robots, autonomous vehicles, smart infrastructure, and more.
中文介绍 NVIDIA 开源的物理 AI 开发平台，集合了世界模型、数据集和配套工具，专为机器人、自动驾驶车辆、智能基础设施等物理 AI 场景设计。开发者可利用该平台训练和部署具备物理世界理解能力的 AI 模型，加速 Physical AI 产品落地。
TypeScript · ★ 24,962 · 🍴 2,910 · 📈 482 stars today
An Open Source implementation of Notebook LM with more flexibility and features
中文介绍 Google NotebookLM 的开源替代实现，在保留核心笔记与 AI 问答功能的基础上提供更高的灵活性与更多特性。支持接入多种 LLM 后端，适合研究人员、学生及知识工作者管理和探索个人知识库，同时保障数据隐私。
Python · ★ 9,558 · 🍴 1,148 · 📈 583 stars today
Talk to any LLM with hands-free voice interaction, voice interruption, and Live2D taking face running locally across platforms
中文介绍 本地跨平台运行的 LLM VTuber 框架，支持免手动触发的语音交互、语音打断及 Live2D 虚拟形象面部驱动。可接入任意 LLM，适合虚拟主播、AI 伴侣和互动娱乐开发者，在本地完成语音识别、模型推理与实时形象渲染的完整流水线。
★ 349,689 · 🍴 83,224 · 📈 740 stars today
A complete computer science study plan to become a software engineer.
中文介绍 系统性的计算机科学自学计划，涵盖数据结构、算法、操作系统、网络等核心知识体系，目标是帮助非科班或自学者达到大厂软件工程师面试水平。内容经过多年社区验证，是 GitHub 上 star 数最高的学习资源之一，适合求职备考者长期规划使用。
Java · ★ 8,953 · 🍴 1,210 · 📈 107 stars today
Multi-platform SDK for integrating GitHub Copilot Agent into apps and services
中文介绍 GitHub 官方发布的多平台 SDK，用于将 GitHub Copilot Agent 集成到第三方应用和服务中。开发者可通过标准化 API 在 Web、桌面、IDE 插件等不同平台快速嵌入 Copilot 的代码生成与 Agent 能力，适合构建 AI 编程助手产品的团队。
Go · ★ 35,656 · 🍴 430 · 📈 255 stars today
Find vulnerabilities, misconfigurations, secrets, SBOM in containers, Kubernetes, code repositories, clouds and more
中文介绍 Aqua Security 出品的全面安全扫描工具，可检测容器镜像、Kubernetes 集群、代码仓库和云环境中的漏洞、配置错误、敏感信息及软件物料清单（SBOM）。单一命令行工具覆盖 DevSecOps 全链路，广泛用于 CI/CD 流水线和基础设施安全合规审计。
C# · ★ 1,310 · 🍴 167 · 📈 358 stars today
Windows companion suite for OpenClaw - System Tray app, Shared library, Node, and PowerToys Command Palette extension
中文介绍 OpenClaw 的 Windows 配套套件，包含系统托盘应用、共享库、节点服务及 PowerToys Command Palette 扩展，将 OpenClaw 功能深度融入 Windows 桌面生态。适合 Windows 用户通过快捷面板和托盘图标便捷调用 OpenClaw 的各项能力。
TypeScript · ★ 5,286 · 🍴 637 · 📈 308 stars today
A modern platform for visual, flexible, and extensible graph-based investigations. For cybersecurity analysts and investigators.
中文介绍 面向网络安全分析师和调查人员的现代化可视化情报调查平台，采用图谱（Graph）方式呈现关系网络，支持灵活扩展调查节点与数据源。适用于 OSINT 情报分析、威胁溯源和数字调查取证场景，提供直观的可视化操作界面降低复杂关系梳理的门槛。
Python · ★ 27,543 · 🍴 2,345 · 📈 173 stars today
AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary
中文介绍 AI Agent 技能模块，可对任意主题在 Reddit、X、YouTube、Hacker News、Polymarket 及开放网络上进行近 30 天的信息检索，并综合生成有据可查的摘要报告。适合需要快速掌握某一话题近期舆论动态和信息趋势的研究人员与内容分析师。
👍 1
LLMs can appear cautious in risk decision-making tasks, yet cautious-looking outputs do not necessarily indicate alignment with human decision-making mechanisms. We investigate this distinction using the St. Petersburg game as a controlled testbed, a classical paradox in which the expected payoff is
👍 10
Feed-forward 3D Gaussian Splatting methods reconstruct a scene from posed or pose-free images in a single forward pass, yet current approaches predict one Gaussian per input pixel, tying the representation budget to camera resolution rather than scene complexity. A flat wall and a richly textured ob
👍 3
Deontic reasoning is the task of answering questions by applying explicit rules and policies to case-specific facts, for example computing tax liability under a statute or determining the outcome of an immigration appeal. A key technical challenge for LLM-based deontic reasoning is that the relevant
👍 3
Large language models (LLMs) are increasingly proposed as clinical agents, yet static, single-turn benchmarks cannot capture how a model dynamically delivers care across an encounter: gathering information, planning treatment, and adapting longitudinal management across successive patient states. Me
👍 35
Rubric-based reinforcement learning (RL) uses an LLM-as-a-Judge (LaaJ) to score model outputs according to rubrics as rewards. However, policy models may exploit latent biases in the judge, leading to reward hacking and ineffective or unsafe training outcomes. In real-world rubric-based RL, such hac
👍 3
Training Data Attribution (TDA) seeks to trace a model's predictions back to its training data. The gold standard for TDA relies on causal interventions, observing how a model changes when data is added or removed, but repeated retraining is computationally challenging for Large Language Models (LLM
👍 9
Scientific and engineering progress is fundamentally a long-horizon iterative process: proposing changes, running experiments, measuring outcomes, and continuously refining artifacts. Yet existing benchmarks for frontier models primarily evaluate either single-turn responses or short-horizon agent t
👍 24
As multi-modal models advance towards long-form video understanding, memory emerges as a critical capability. Despite substantial efforts in developing video datasets and benchmarks, existing works primarily focus on perception and reasoning, without systematically evaluating memory: what models ret
👍 11
Lane-level maps are critical infrastructure for autonomous driving and lane-level navigation, yet constructing and maintaining standardized lane networks for hundreds of cities remains highly labor-intensive. Recent end-to-end vectorized mapping methods can predict lane geometry and topology directl
👍 10
High-quality pretraining data is a central ingredient in modern language models, but German-language resources remain far less developed than their English counterparts: they are often smaller, less carefully curated, weakly documented, and rarely validated through controlled training experiments. W
👍 24
Large Reasoning Models (LRMs) have achieved remarkable progress thanks to Reinforcement Learning with Verifiable Rewards (RLVR) on Chain-of-Thoughts (CoTs). However, since long CoTs naturally contain trial and errors and mainstream RLVR approaches choose outcome-correct CoT trajectories for memoriza
👍 8
Existing benchmarks for MLLM-generated web artifacts assess interaction through local evidence and miss the requirement-induced states and transitions that determine whether a page works. We introduce WebRISE, which compiles task requirements into Interaction Contract Graphs (ICGs) of observable sta
👍 1
Large language models improve final-answer accuracy through extended chain-of-thought reasoning, but often spend tokens inefficiently and offer little inference-time control. Existing efficient reasoning methods control thinking length by shortening, early-stopping, or compressing traces, leaving ho
👍 1
Reinforcement learning (RL) has become a dominant post-training paradigm, enabling large language models (LLMs) to learn from rewards. We observe that societal regulations are structurally similar to reward functions. They define measurable outcomes, thresholds, and exceptions, while often leaving i
👍 0
LLM-agent budget overruns are a documented production failure class: a single retry loop can spend thousands of dollars before an operator notices, and the in-process integrity properties that would prevent it (no aliasing, no double-spend, no use-after-delegation of a cost-bearing value) are enforc
👍 1
Graph Language Models (GLMs) have become a promising direction for adapting Large Language Models (LLMs) to graph learning tasks. By transforming graph topology and node information into graph tokens, GLMs allow LLMs to jointly process structured graph inputs and textual instructions. Yet, it remain
👍 2
Large language model (LLM) agents are evolving from request-response assistants into long-running software actors: they maintain state across model calls, fork subtasks, wait for external events, request human authority, generate tools, and perform side effects that must be resumed and audited. This
👍 0
Low-Rank Adaptation (LoRA) successfully enables personalization in text-to-image generation by adapting pre-trained diffusion models to specific visual concepts and styles. However, extending such models to multi-concept customization remains challenging. Naively combining multiple LoRA weights or t
👍 28
Multimodal agents in robotics, AR, and autonomous driving must reason about places and layouts from continuous egocentric streams, often using evidence outside the current view. Existing benchmarks either evaluate offline over full videos or target events rather than spatial structure. We introduce
👍 6
Structured financial audit verification is difficult for language-model agents because correctness depends on structured evidence rather than text alone. A model must link reported facts to taxonomy concepts, traverse calculation or dimensional relations, and recompute expected values before applyin
👍 5
How can a population of agents self-orchestrate and self-adapt into stronger collective intelligence without centralized control? Inspired by Friedrich Hayek's economic theory of decentralized coordination in markets, we study this question through an agent economy in which agents compete via auctio
👍 11
On-Policy distillation (OPD) in large language models is shifting from full-trace KL supervision toward more selective training paradigms. Recent OPD methods increasingly focus on selecting which trajectories to learn from, which tokens are most informative, and which supervision signals are most re
👍 1
Equivariance theory predicts that an architectural symmetry prior reduces sample complexity by a factor of |G|; this is widely cited but rarely measured as a scaling law with controls that separate the prior from its confounds. On a controlled C_n-symmetric task, we report three findings. First, a w
👍 1
AI glasses present a compelling platform for AI agents to serve as personalized memory assistants. To be genuinely useful, such systems must move beyond short-term video comprehension and address memory gaps that humans experience for practical, personal, or social purposes over longitudinal egocent
👍 1
Transfer learning aims to facilitate the learning of a target domain by transferring knowledge from a source domain. The source domain typically contains semantically meaningful samples (*e.g.*, images) to facilitate effective knowledge transfer. However, a recent study observes that the noise domai
👍 0
Unified multimodal models (UMMs) have emerged as a promising paradigm for general-purpose multimodal intelligence. As they are deployed in real-world applications, effectively updating internal knowledge becomes critical. While knowledge editing has matured for text-only models, it remains unclear w
👍 2
While current multimodal models are proficient at open-ended visual editing, executing precise single-answer edits remains an important obstacle. To probe this challenge, we introduce PaintBench, a dynamically scalable benchmark targeting 20 fundamental precise visual editing operations across four
👍 2
Humans can effortlessly perceive spatial layouts, form cognitive representations, reason about spatial relations, and translate such reasoning into actions in everyday 3D environments. Although recent vision-language models (VLMs) have shown promising performance on observation-conditioned spatial p
👍 4
Weight-space model merging is usually formulated as an algebraic operation on checkpoints, yet at LLM scale the limiting resource is often the set of expert weights that must be read. We introduce MergePipe, a budget-aware execution layer that casts LLM merging as an expert access-set problem: given
👍 20
LLM agents are rapidly evolving from coding assistants into autonomous software engineering systems. However, existing evaluation methodologies remain largely centered on static, isolated, and short-horizon benchmarks that fail to capture the dynamic complexity of real-world production workflows. As
@elpresidank · 116 粉丝 · 2.9M 阅 · 543 赞 · 35 转
Most AI agent memory is built on embeddings. And there's now a proof that this entire class of system is going to forget what you stored in it — and confidently make up things you never stored at all.
中文介绍 基于嵌入向量的 AI agent 记忆系统存在根本性缺陷——有数学证明表明此类系统必然会遗忘已存储内容，并自信地「捏造」从未存储过的信息。作者从拓扑学角度分析上下文结构，指出平坦的向量空间无法表达层级关系，提出用结构化上下文替代纯嵌入方案以规避遗忘问题。
@0xCodez · 3.3K 粉丝 · 637.2K 阅 · 510 赞 · 59 转
Most Claude Code users still write their workflows by hand. They chain prompts, copy outputs, paste them into the next prompt, fix what went wrong, repeat. 9 out of 10 builders haven’t tried Dynamic
中文介绍 针对 Claude Code 动态工作流的系统性教程，梳理了 Anthropic 工程师实际采用的 6 种模式与 14 个步骤。大多数用户仍在手动串联提示词，而动态工作流可让 Claude 自动编排任务链。作者指出 90% 的开发者尚未尝试此特性，文章提供可落地的上手路径。
@Saboo_Shubham_ · 116.2K 粉丝 · 263.3K 阅 · 517 赞 · 74 转
The frontend used to be a fixed thing. Designers drew it. Engineers built it. Users got what shipped. That's over. The interfaces shipping in 2026 are drawn partly by the agent itself, in real time,
中文介绍 探讨「生成式 UI」正在取代传统前端开发范式：2026 年的界面不再由设计师固定交付，而是由 agent 在运行时根据上下文实时生成。文章分析这一转变对前端工程师角色的冲击，以及 AI 驱动界面在产品设计流程中的新位置。
@mem0ai · 17.6K 粉丝 · 82.8K 阅 · 520 赞 · 60 转
Agent harnesses are where AI software actually runs. Cursor, Devin, Claude Code, Codex: these environments handle context, orchestrate tools, coordinate agents, and increasingly, manage memory. The
中文介绍 Mem0 团队梳理主流 agent harness（Cursor、Devin、Claude Code、Codex）的记忆管理现状：这些环境负责上下文维护、工具调度与多 agent 协作，但记忆机制差异显著。文章对比各平台的记忆架构，指出当前痛点并展望标准化方向。
@trq212 · 263.1K 粉丝 · 75.7K 阅 · 542 赞 · 36 转
Last week, we released dynamic workflows in Claude Code. Claude can now write its own harness on the fly, custom-built for the task at hand. While the default Claude Code harness is built for coding,
中文介绍 Anthropic 工程师介绍 Claude Code 最新发布的动态工作流功能：Claude 现在可以在运行时为特定任务即时生成自定义 harness，不再局限于内置的编码流程。文章展示该能力如何扩展至非编码场景，并说明动态 harness 的设计思路与适用边界。
@drfeifei · 738.0K 粉丝 · 72.2K 阅 · 699 赞 · 144 转
“The world is everything that is the case.” — Ludwig Wittgenstein, Tractatus Logico-Philosophicus, 1921 The world is not made of words. In an earlier essay, we argued that spatial intelligence is AI’s
中文介绍 李飞飞发表关于「世界模型功能分类」的长文，延续其空间智能论述。文章以维特根斯坦开篇，论证世界由非语言结构构成，进而提出对 AI 世界模型的功能性分类框架，探讨空间推理与感知在通用智能中的核心地位。
@sydneyrunkle · 7.5K 粉丝 · 69.5K 阅 · 511 赞 · 74 转
Building useful agents is largely about customization: connecting your agent to the right context, data, and environment(s) for the task at hand. At its core, an agent is a model calling tools in a
中文介绍 构建自定义 agent harness 的实操教程：核心思路是将 agent 与任务所需的正确上下文、数据和运行环境对接。文章拆解 harness 的基本组成——模型调用工具的循环机制——并给出针对不同任务场景的定制化配置方法，适合希望超越默认 harness 限制的开发者。
@mvanhorn · 27.6K 粉丝 · 54.5K 阅 · 545 赞 · 44 转
Three months ago I posted "Every Claude Code Hack I Know." It hit 913K views. @kevinrose had asked what IDE to use, and my answer was: "No IDE. Just plan.md files and voice." This used to be called
中文介绍 作者汇总截至 2026 年 6 月的全部 agentic 工程技巧，是其三个月前获 91.3 万阅读的「Claude Code 技巧」帖的升级版。核心理念延续「无 IDE、只用 plan.md 文件加语音」，并加入多 agent 协调、长任务规划等新实践，覆盖内容、开发、运营多个场景。
@jainarvind · 9.3K 粉丝 · 53.7K 阅 · 505 赞 · 68 转
Enterprise AI token spend is scaling quickly, especially as the technology shifts from simple chat assistants into coding agents, AI coworkers, and long-running workflows. These systems do far more
中文介绍 企业 AI token 开销激增的根源是架构问题而非模型选择问题。随着 AI 从聊天助手演进为编程 agent 和长流程 AI 协作者，系统调用量级大幅提升。文章分析 token 消耗的架构性驱动因素，提出通过任务分解、缓存和 agent 粒度优化来控制成本。
@IBuzovskyi · 1.2K 粉丝 · 50.1K 阅 · 500 赞 · 51 转
These 10 Hermes Agent hacks saved me 15+ hours every week - and they work for any workflow you run repeatedly. Content, software development, business operations, client management, research, sales.
中文介绍 分享 10 个 Hermes Agent 使用技巧，作者声称每周节省 15 小时以上，适用于内容创作、软件开发、客户管理、销售等重复性工作流。文章将普通对话式 agent 改造为 7×24 小时自动运行系统，提供可直接复用的配置思路。
@delba_oliveira · 74.0K 粉丝 · 37.4K 阅 · 533 赞 · 38 转
As we delegate more ambitious tasks to Claude, it becomes increasingly important that it can verify its own work. The more Claude can self-verify: the more independently it can work on long-running
中文介绍 探讨如何通过「反馈回路」让 Claude Code 在复杂任务中减少人工干预：关键是让 Claude 具备自我验证能力。文章介绍在 CLAUDE.md 中设置可执行验证步骤、利用测试和 lint 作为内置检查点等具体方法，使 Claude 能独立完成长流程任务而无需频繁监督。
@ericzakariasson · 67.9K 粉丝 · 37.3K 阅 · 507 赞 · 27 转
If you've ever watched an agent try to fix a bug, you've watched it guess. It reads the code, comes up with a theory, makes an edit, and hopes. Sometimes it's right. A lot of the time you get a fix
中文介绍 指出 agent 调试时依赖猜测的根本原因是缺乏运行时上下文：agent 只能读代码、构建假设、做修改，却看不到实际运行状态。文章提出为 agent 注入运行时上下文（如日志、错误栈、环境变量）的具体方案，将「猜测式修复」变为「诊断式修复」。
@OptimaiNetwork · 98.8K 粉丝 · 5.7K 阅 · 542 赞 · 110 转
May was an important month for OptimAI because it marked a shift from proving the product to expanding the platform. The broader market is moving in the same direction: AI is no longer just about
中文介绍 OptimAI 网络发布 5 月进展与 6 月方向：5 月完成从产品验证到平台扩张的阶段转型，定位「agent 原生互联网」基础设施。文章同步市场观察——AI 正从单点工具演进为协作平台——并说明下一阶段的技术与生态建设重点。
We talk with the VendingBench authors on evaling Claudes from Haiku to Mythos, and how they build leading, and lasting, frontier evals from scratch.
中文介绍 Latent Space播客采访了Andon Labs的Lukas Petersson与Axel Backlund，两人是VendingBench评测基准的作者。对话围绕如何从零构建前沿AI模型评测体系展开，涵盖对Claude系列（从Haiku到Mythos）的评估方法，以及如何打造具有持续性和引领性的基准测试框架。
中文介绍 英伟达在Hugging Face发布Nemotron 3.5内容安全模型，主打可定制化的多模态安全能力，面向全球企业级AI应用场景，支持跨语言、跨模态的内容安全审核需求。
中文介绍 ServiceNow AI在Hugging Face发布EVA-Bench Data 2.0评测数据集，覆盖3个领域、121种工具及213个场景，旨在为企业AI智能体的能力评估提供更全面的基准测试支持。
Learn how Endava is using AI agents, ChatGPT Enterprise, and Codex to accelerate software delivery, automate workflows, and build an AI-native culture across the enterprise.
中文介绍 科技服务公司Endava正将AI智能体、ChatGPT Enterprise及Codex整合至软件交付流程，通过自动化工作流加速开发效率，并在企业内部推进AI原生文化建设，重新设计以AI为核心的软件交付体系。
Most days in her chambers, Judge Maritza Braswell, a federal magistrate judge in Colorado, sifts through stacks of documents written by people without a lawyer. Many of them can’t afford to hire a lawyer, and others have cases too weak or too small to interest one. She reads each one carefully, mind
中文介绍 MIT Technology Review报道，美国法院正面临大量AI生成诉状的涌入。科罗拉多州联邦治安法官Maritza Braswell每日需审阅大批无律师代理当事人提交的文件，其中许多疑似由AI生成。此现象引发司法界对诉讼质量与效率的广泛担忧。
ChatGPT introduces a new memory system to better remember preferences, keeping context fresh and relevant across conversations.
中文介绍 OpenAI为ChatGPT推出名为「Dreaming」的新记忆系统，能更好地记住用户偏好，使上下文在多次对话中保持新鲜且相关，提升ChatGPT的个性化与持续服务能力。
a quiet day.
中文介绍 Latent Space AINews简报报道，图像生成领域迎来Reve 2与Ideogram 4两款新模型，均新增布局（Layout）控制功能，可对生成图像的构图与元素排列进行更精细的控制。整体为较平静的一个行业日。
An action plan for AI-powered biological resilience
中文介绍 OpenAI发布「智能时代的生物防御」报告，提出以AI技术增强生物安全韧性的行动计划，探讨如何利用人工智能应对生物威胁，涵盖监测、预警及防御等多个维度的政策与技术建议。
中文介绍 Hugging Face发布博文，介绍其将hf命令行工具（CLI）设计为面向AI智能体优化的Hub交互方式，旨在使智能体能够更高效、自动化地访问和操作Hugging Face Hub上的模型、数据集等资源。
中文介绍 TLDR AI日报（2026年6月4日）要点：DeepSeek正在进行新一轮融资；Meta旗下部分模型发布时间出现延期；谷歌发布Gemma 4 12B新模型。此外收录当日其他AI行业动态。
Verified Generation and Compounding Intelligence
中文介绍 Latent Space播客采访Axiom Math创始人Carina Hong，探讨如何超越非正式AI，通过可验证生成（Verified Generation）与复合智能（Compounding Intelligence）构建更严谨、可信赖的数学推理AI系统。
The legendary Microsoft CEO makes his first Latent Space appearance!
中文介绍 微软CEO萨提亚·纳德拉（Satya Nadella）首次亮相Latent Space播客，本期为No Priors与Latent Space联合特别节目，在微软Build大会期间录制，就AI战略、企业转型等议题展开深度对话。
GPT-Rosalind advances life sciences research with enhanced biological reasoning, medicinal chemistry expertise, genomics analysis, and experimental workflow capabilities.
中文介绍 OpenAI发布GPT-Rosalind新能力更新，该模型专注生命科学研究领域，新增增强型生物推理、药物化学专业知识、基因组学分析及实验工作流支持等功能，进一步强化其在科研场景中的应用能力。
中文介绍 Hugging Face发布博文，探讨直接偏好优化（DPO）技术在聊天机器人以外更广泛场景中的应用，分析DPO如何用于专业任务的模型对齐，拓展其在企业级AI和垂直领域的落地潜力。
See how Wasmer used Codex with GPT-5.5 to build a Node.js runtime for the edge, accelerating development 10x to 20x and shipping in weeks instead of months.
中文介绍 WebAssembly运行时公司Wasmer借助OpenAI Codex（基于GPT-5.5）构建了面向边缘计算的Node.js运行时，开发效率提升10至20倍，原本需数月完成的项目仅用数周即完成交付。
本期技术面呈现明显的资产分化格局。美股方面，SPY 与 QQQ 均接近 52 周高位，多头排列结构完整，SPY 当前价格 757.09 高于 SMA20（745.99）、SMA50（711.9）及 SMA200（683.44），RSI14 达 69.7；QQQ 的 RSI14 更高至 73.5，已触及超买状态并触发 MACD 死叉，短期上行动能出现分化信号。NVDA 近 5 日涨幅 +2.06%，均线多头排列维持，但 MACD 仍低于信号线；GOOGL 单日+3.68% 但 5 日累跌 -4.6%，信号混杂。加密市场整体承压显著：BTC 现报 63,423 美元，RSI14 仅 17.9，ETH 同样降至 1,759 美元、RSI14 17.9，SOL 报 68.16 美元、RSI14 19.6，三者均深陷超卖且空头排列。加密恐慌贪婪指数降至 12（极度恐慌区间），加密总市值 24 小时内下滑 -1.88% 至 2.28 万亿美元，BTC 主导率维持在 55.8%，显示市场避险情绪向比特币集中。中概股板块整体偏弱，阿里巴巴、拼多多、京东、腾讯均处于空头排列，RSI14 普遍在 37-49 范围内震荡，无明显反转信号。商品与外汇层面，黄金期货在 4,499 美元附近震荡整理，原油 5 日涨幅 +4.42% 但 RSI14 仅 46，动能有限；美元指数 DXY 报 99.43，接近 52 周高位且多头排列，10Y 美债收益率 4.48% 亦维持多头排列，宏观流动性环境对风险资产仍构成一定压制背景。
QQQ 当前 RSI14 达 73.5，处于超买区域，MACD 已形成死叉（MACD 21.055 低于信号线 21.514），同时距 52 周高点仅 -1.07%，上行空间受压。尽管均线呈多头排列（SMA20 721.5 / SMA50 665.46 / SMA200 621.14 依次走高），但短期动量指标已出现顶部信号，技术结构显示超买叠加死叉的负背离特征，回调压力值得关注。
BTC 当前报价 63,423 美元，RSI14 仅 17.9，深度超卖，但价格已跌破 SMA20（74,028）、SMA50（76,656）、SMA200（78,931），三线呈空头排列，MACD 为 -2,984，信号线 -1,613，柱状图负值持续扩张。5 日跌幅 -14.01%，距 52 周高点下行 -49.74%。配合加密恐慌贪婪指数仅 12（极度恐慌），加密总市值 24h 下滑 -1.88%，整体空头结构明显。
GOOGL 当日大涨 +3.68%，但 5 日累计仍跌 -4.6%，短期波动剧烈。RSI14 为 48.6，处于中性区间，均线多头排列（SMA20 386.85 / SMA50 352.95 / SMA200 303.21），然而当前价格 372.19 已跌破 SMA20，MACD（2.60）与信号线（8.14）仍有较大负差，动量偏弱。技术结构多空信号交织，方向有待进一步确认。
黄金期货现报 4,499.1 美元，当日涨 +1.41%，但 5 日变动为 0%，短期方向震荡。当前价格低于 SMA20（4,565）与 SMA50（4,629），但高于 SMA200（4,394），形成上下分层格局。RSI14 为 44.3，处于中性偏弱区间，MACD -53.58 低于信号线 -51.36，短期动量仍偏负。整体呈中性整理态势，无明确突破信号。
SPY 当前报价 757.09，距 52 周高点仅 -0.44%，距 52 周低点累涨 28.09%。SMA20（745.99）/ SMA50（711.9）/ SMA200（683.44）形成标准多头排列，价格位于三条均线之上。RSI14 为 69.7，尚未突破超买阈值 70。MACD（11.95）略低于信号线（12.58），短线动量轻微回落，但整体趋势结构仍维持偏强特征。
VIX 恐慌指数
10Y 美债收益率 (%)
美元指数 DXY
S&P 500 ETF
Nasdaq 100 ETF
Apple
Microsoft
Nvidia
Alphabet
Tesla
Meta
Bitcoin
Ethereum
Solana
阿里巴巴 (BABA)
拼多多 (PDD)
京东 (JD)
腾讯控股 (0700.HK)
黄金期货
WTI 原油期货
美元 / 人民币
本报告所有内容均基于公开行情数据计算所得技术指标，仅供技术指标解读参考，不构成任何投资建议或买卖信号。过去走势不代表未来表现，市场存在不确定性，读者应自行判断风险并承担相应责任。
Swampy mangrove forests are staging a surprise comeback - which is good news for coastal communities and the climate.
中文摘要 全球红树林正在经历意外复苏。这种生长于沼泽地带的沿海森林在遭受数十年人为破坏后，近年呈现恢复迹象，对沿海社区防灾及应对气候变化均具有重要意义。
Follow today’s news live Get our breaking news email, free app or daily news podcast Jane Hume, the deputy opposition leader, accused the government of being “sneaky”, saying discussions around tax reforms in parliament were much too short to properly address so-called “generational” tax reforms. “W
中文摘要 澳大利亚副反对党领袖简·休谟指责工党政府在税改问题上态度"鬼鬼祟祟"，称议会就所谓"代际"税改的讨论时间远远不足。前总理约翰·霍华德亦联署公开信，呼吁推进博彩业改革。
Republicans seek to pass $70bn immigration-enforcement funding, but must endure series of votes on controversial issues.
中文摘要 美国民主党在国会"马拉松投票"程序中，强制就特朗普180亿美元和解基金进行表决。共和党方面正寻求通过700亿美元移民执法拨款，但须在一系列争议性议题投票中逐一过关。
Kenyan President William Ruto said allowing the US to build an Ebola quarantine facility in Kenya was the “right thing”.
中文摘要 肯尼亚总统威廉·鲁托为允许美国在肯尼亚建设埃博拉隔离设施的决定进行辩护，称此举"是正确的"。此前该决定引发民众抗议，并造成人员伤亡。
Ukrainian president proposes meeting in neutral third country as Trump says both sides have to ‘make compromises’ The Ukrainian president, Volodymyr Zelenskyy, has called for face-to-face negotiations in a public letter addressed directly to the Russian president, Vladimir Putin. The letter, the fir
中文摘要 乌克兰总统泽连斯基致信俄罗斯总统普京，公开呼吁在中立第三国举行面对面谈判，以推动结束俄乌战争。特朗普表示，双方均须作出"妥协"。
The K-pop band say accepting their flaws and embracing humour took them to a new level of success.
中文摘要 韩国女团LE SSERAFIM（勒色菲）讲述乐队内部矛盾及应对网络恶意攻击的经历，成员表示通过接纳自身不足、以幽默化解压力，最终推动乐队走向新的发展阶段。
It could take weeks - or even a month - before we know who won elections in California and Los Angeles.
中文摘要 美国第二大城市洛杉矶及加利福尼亚州选举结果公布极为缓慢，预计需数周甚至一个月方能确定最终胜选者，文章对此现象背后的制度原因进行了解析。
Hezbollah’s leader said the Iran-backed group, which was not included in U.S.-brokered talks, said a truce worked out between Israel and the Lebanese government amounted to surrender for his group.
中文摘要 真主党领导人宣布拒绝以色列与黎巴嫩政府之间由美国斡旋达成的停火协议，称该协议对真主党而言无异于投降。真主党未被纳入相关谈判，以色列对黎巴嫩的军事行动仍在持续。
Al Jazeera’s Hala Al Shami explains why US officials are invested in Colombia's presidential elections.
中文摘要 半岛电视台记者哈拉·沙米解析美国高度关注哥伦比亚总统大选的深层原因，指出美国官员对此次选举结果高度投入，背后涉及地区战略利益考量。
In a remote mining town at the center of the Democratic Republic of Congo’s Ebola outbreak, grief and mistrust are complicating efforts to stop the virus. Health workers are trying to treat the ill and bury the dead, but some residents are still in denial that the disease is even real.
中文摘要 刚果（金）埃博拉疫情中心、一处偏远矿业小镇内，民众对疾病存在的否认情绪及对卫生部门的不信任正在蔓延，给防疫工作带来严重阻碍。当地卫生工作者同时面临救治病患和处理遗体的双重压力。
The Trump administration’s efforts to tighten the economic noose on Cuba appear to be working, as more international firms announced they would leave the island.
中文摘要 特朗普政府持续收紧对古巴的经济制裁，多家国际连锁酒店宣布撤出古巴市场。此举被视为美方加紧经济施压策略初见成效的信号。
Al Jazeera’s Obaida Hitto has the latest from Lebanon where a new diplomatic push has done little to quell violence.
中文摘要 真主党明确拒绝美国提出的黎巴嫩停火方案，与此同时以色列对黎巴嫩的军事打击仍未停止。半岛电视台记者报道，新一轮外交斡旋努力收效甚微，局势未见缓和。
Gun battles erupted on Thursday, as armed groups loyal to rival politicians clashed. Residents said it was the worst fighting in years.
中文摘要 索马里首都摩加迪沙爆发激烈武装冲突，效忠不同政治派系的武装团体之间发生枪战，当地居民表示这是多年来最严重的一次交火，当局正努力恢复秩序。
Zeynab Javadli had not been in contact with any of her friends since Tuesday.
中文摘要 检察官透露，迪拜统治者侄子的前妻泽纳布·贾瓦德利已被羁押，自周二起与所有友人失去联系，案件具体详情目前仍在调查之中。
Ukraine's president tells the Russian leader that only "direct engagement" between the two countries could end the war, with the US focused on Iran.
中文摘要 乌克兰总统泽连斯基发表公开信，直接致函普京，提议举行面对面直接谈判，强调只有两国直接接触才能结束战争。当前美国外交重心已转向伊朗问题。
Instead of throwing away byproducts of food processing, fermentation is making them valuable.
中文摘要 发酵技术正被用于将食品加工过程中产生的副产品转化为有价值的食材，而非直接丢弃。这一古老工艺有助于减少食物浪费，同时将原本废弃的原料变为可食用且具商业价值的产品，为食品行业的可持续发展提供新思路。
Jackie Jantos says loneliness and lack of confidence were challenges for young adults looking for relationships.
中文摘要 交友应用Hinge首席执行官杰基·詹托斯表示，20多岁的单身年轻人因孤独感和缺乏自信，在交友软件上越来越依赖AI功能来"迈出第一步"。她指出，孤独与自信不足是当代年轻人建立亲密关系面临的主要挑战。
Naomi Gleit has weathered many controversies at Meta, but remains in what she tells the BBC is her "dream job".
中文摘要 Meta任职时间最长的员工娜奥米·格莱特接受BBC采访，谈及人工智能、就业影响及对其上司扎克伯格的看法。格莱特经历了Meta多次争议风波，仍将这份工作称为自己的「梦想职业」。
Sempra Infrastructure said it began producing liquefied natural gas from Mexico’s first West Coast export terminal, with plans to ultimately ship the fuel to Asia in a move that could add relief to a market hurt by tightened global supply due to the Iran war.
中文摘要 森普拉基础设施公司宣布，墨西哥西海岸首座液化天然气出口终端已正式投产。该设施计划向亚洲市场出口LNG，有望缓解因伊朗战争导致全球供应趋紧的能源市场压力。
Mira Murati, Co-Founder & CEO of Thinking Machines Lab, discusses the future of human-AI interaction, evolving AI products and the biggest opportunities ahead with Bloomberg’s Emily Chang at Bloomberg Tech 2026 in San Francisco. (Source: Bloomberg)
South Korea’s $4.9 trillion stock market is starting to flash signs of strain after a world-beating rally.
中文摘要 韩国股市在累计涨幅达105%的强劲反弹后，开始出现疲态迹象。市值约4.9万亿美元的韩国股市上涨广度收窄，市场担忧此轮全球领先的涨势能否持续，隐患逐渐浮现。
Research analysts across Wall Street are telling would-be SpaceX IPO buyers that they’re modeling for the company’s artificial intelligence division to see 100 times revenue growth at the end of the decade, to help justify a targeted $1.8 trillion valuation.
中文摘要 华尔街多家研究机构预测，SpaceX人工智能部门营收到2030年将增长100倍，以支撑该公司1.8万亿美元的目标估值。这一预测主要面向有意购买SpaceX IPO股票的潜在投资者。
Hit-Boy, Grammy-Winning Artist & Producer discusses AI tools, music production and creative experimentation with Bloomberg’s Emily Chang and Tom Giles at Bloomberg Tech 2026 in San Francisco. (Source: Bloomberg)
中文摘要 格莱美奖得主音乐制作人Hit-Boy在旧金山彭博科技2026峰会上，与彭博记者探讨人工智能工具在音乐制作中的应用，以及AI如何推动创意民主化与创作实验的新可能。
Fei-Fei Li, Co-Founder & CEO at World Labs discusses spatial intelligence, large world models and real-world AI applications with Bloomberg’s Emily Chang at Bloomberg Tech 2026 in San Francisco. (Source: Bloomberg)
中文摘要 World Labs联合创始人兼CEO李飞飞在旧金山彭博科技2026峰会上，就空间智能、大型世界模型的构建以及人工智能的现实世界应用场景展开深入讨论。
Stocks in Asia were set to edge lower on Friday as investors grappled with a pullback in artificial-intelligence shares and hopes of a diplomatic resolution to tensions in the Middle East.
中文摘要 受人工智能相关股票回调以及中东局势外交解决前景影响，亚洲股市周五料小幅走低，市场情绪偏谨慎。与此同时，油价趋于稳定，投资者持观望态度。
Oil was steady after its first decline this week on optimism that talks between the US and Iran will see energy flows resume through the Strait of Hormuz.
中文摘要 油价在本周首次下跌后企稳，市场对美伊谈判进展持乐观态度，预期霍尔木兹海峡能源通道或将恢复畅通，从而缓解全球原油供应紧张局面。
S&P Dow Jones Indices will keep its existing eligibility requirements for main benchmarks like the S&P 500 Index, rejecting proposals that would have made it faster for mega-cap companies such as Elon Musk’s SpaceX to gain rapid entry into the benchmark after going public.
中文摘要 标普道琼斯指数公司经咨询后决定维持现行标普500指数纳入资格标准，拒绝了相关修改提案。这意味着SpaceX等大型公司上市后仍无法快速纳入该基准指数，马斯克旗下SpaceX的入指之路短期内不会提速。
4 回复 · 程序员 节点
26 回复 · 程序员 节点
6 回复 · 程序员 节点
15 回复 · 程序员 节点
9 回复 · 程序员 节点
5 回复 · 程序员 节点
11 回复 · Apple 节点
10 回复 · Apple 节点
26 回复 · Apple 节点
17 回复 · Apple 节点
应该没问题了 muyuan.do 不要在我这里浪费时间，去做你该做的事情，去爱你该爱的人 123 个帖子 - 122 位参与者 阅读完整话题
L站的佬友们好，我入行这个AI漫剧这行已经三个月了，基本都是接触的海外项目，然后这个月刚做了海外阅文和抖音合作的海外项目，在审核阶段，就和大家分享一下自己做真人海外的心得 人物站位和人物走位控制 情绪参考 工具使用 画面，镜头的参考 最后，大家做这个真人时候，我建议大家看着提示词过一遍画面，脑子里面有这个画面的详细情况，你能知道这个画面是不是符合你的想法，还有就是，剪辑，配乐真的很重要，有能力的佬友可以去抖音找找suno ai音乐的教程，我这几天都在做这个的广告项目 欢迎佬友来问我问题，但是我上班时间一般不会登陆L站，但是晚上9点钟后到家一般会回复，还有如果是小白要入这行现在，我的建议是，先按
l0veyou公益站推出了数字人模型，一次可生成10秒，效果非常逼真，没过多久我会给它支持到生成20秒一个视频，这样更好 在这里选择音频（上传的音频需要大于两秒，否则会生成视频失败），然后点击上传模型，我需要在右下角的加号，点击上传参考图，然后提示词可以这样填：她在说话。 另外提醒一下， Ai生图暂时用不了了。我今天修了一天也没修好，可能至少要明天才能好了。今晚也不知道还有没有得睡，因为明天我还要上架新模型，明天我要上架gemini3.1pro（0.1ldc一句话）和3d模型生成（生成出来的3d模型，它的零件可以直接拆分，而不是一整块模型，效果非常好） 14 个帖子 - 9 位参与者 阅读完整
停机时间延长至今晚10点 159 个帖子 - 152 位参与者 阅读完整话题
今日早上八点半奥特曼重置了所有账号，当前正在管理9个Pro20x，刚好可以拿来估计一波GPT订阅的额度。 账号： 到截图时间的用量数据： 模型 输入 输出 缓存读取 合计 Token GPT5.5 62,178,157 1,759,623 245,985,792 309,923,572 GPT5.4 284,049,752 12,110,928 1,555,673,600 1,851,834,280 平均缓存命中率： 83.2% 到截图时间的费用数据： GPT5.4：1313 美元 到截图时间9个账号合计使用周限制： 7% + 9% + 8% + 8% + 8% + 7% + 11% + 10
转自新华社 记者从教育部获悉，今年全国高考报名人数为1290万人。 对于高校招生宣传工作，教育部强调，严禁虚假宣传、违规承诺，严禁以任何形式炒作“高考状元”“高分考生”“升学率”。推动高校录取通知书回归“一页纸”，坚决纠治奢华录取通知书、新生礼盒等不良风气。 来源：新华社微信公众号综合人民日报客户端（记者：吴丹） 有佬友知道这是何意味吗？ 86 个帖子 - 38 位参与者 阅读完整话题
考虑休息一会儿 16 个帖子 - 15 位参与者 阅读完整话题
我现在是浑身冒火 44 个帖子 - 42 位参与者 阅读完整话题
我合计合计得买个新服务器了 极其的夸张 137 个帖子 - 128 位参与者 阅读完整话题
美国企业支出管理平台 Ramp 发布的 2026 年 6 月报告显示，中国 AI 公司 DeepSeek 登上了热门软件榜首。尽管美国官方先前高度防范中国大模型，但真实的商业交易数据却揭示了相反的现状。Ramp 分析了平台上 5 万多家企业的信用卡消费记录，发现许多美国公司并未在本地部署开源模型，而是直接掏钱购买 DeepSeek 官方的托管 API 服务。这意味着，大量美国企业的数据正直接发送并存储在位于中国的服务器上。 真实的资金流向与一年多前美国社会对 DeepSeek R1 刚发布时的警惕态度形成了强烈反差。当时出于对泄密和安全的担忧，美国大公司和政府机构普遍限制使用中国模型。然而，面