Shell · ★ 56,769 · 🍴 6,117 · 📈 2,656 stars today
Production-grade engineering skills for AI coding agents.
中文介绍 面向 AI 编码 Agent 的生产级工程技能集合，由 Google Chrome 团队的 Addy Osmani 整理维护。提供一套可直接用于实际项目的最佳实践、规范和技巧，帮助 AI 编码助手（如 GitHub Copilot、Cursor 等）更好地理解工程上下文，提升代码质量与可维护性，适合希望将 AI Agent 落地到真实工程场景的开发者使用。
Python · ★ 1,774 · 🍴 422 · 📈 20 stars today
Music Assistant is a free, opensource Media library manager that connects to your streaming services and a wide range of connected speakers. The server is the beating heart, the core of Music Assistant and must run on an always-on device like a Raspberry Pi, a NAS or an Intel NUC or alike.
中文介绍 Music Assistant 是一款免费开源的媒体库管理工具，作为整个系统的核心服务端，负责连接 Spotify、Tidal 等流媒体服务，并将音乐推送至各类联网音箱（如 Sonos、Chromecast）。支持统一管理分散在多个平台的音乐资源，适合家庭音频爱好者和智能家居用户构建本地化音乐中心。
TypeScript · ★ 37,623 · 🍴 8,712 · 📈 388 stars today
Mattermost is an open source platform for secure collaboration across the entire software development lifecycle..
中文介绍 Mattermost 是一款开源的安全团队协作平台，覆盖软件开发全生命周期，支持私有化部署。提供频道消息、工作流自动化、与 DevOps 工具链（如 Jira、GitHub、PagerDuty）的深度集成，满足企业对数据主权和安全合规的要求，广泛应用于对数据安全敏感的政府、金融和科技团队。
Swift · ★ 35,065 · 🍴 978 · 📈 3,504 stars today
A tool for creating and running Linux containers using lightweight virtual machines on a Mac. It is written in Swift, and optimized for Apple silicon.
中文介绍 Apple 官方开源的 Mac 端 Linux 容器工具，使用轻量级虚拟机技术在 macOS 上创建和运行 Linux 容器，无需依赖 Docker Desktop。基于 Swift 编写，针对 Apple Silicon（M 系列芯片）深度优化，启动速度快、资源占用低，适合希望在 Mac 上进行容器化开发的工程师，尤其是 M 芯片用户。
TypeScript · ★ 118,023 · 🍴 6,301 · 📈 179 stars today
Collection of publicly available IPTV channels from all over the world
中文介绍 全球最大的公开 IPTV 频道资源聚合项目，收录来自世界各地数千个可免费访问的直播频道，按国家、语言、分类整理为 M3U 播放列表。用户可直接在 VLC、Kodi、IPTV Smarters 等播放器中加载使用，适合需要收看境外或特定地区电视直播的用户。
Shell · ★ 225,985 · 🍴 20,085 · 📈 1,275 stars today
An agentic skills framework & software development methodology that works.
中文介绍 一套面向 AI Agent 的技能框架与软件开发方法论，旨在解决当前 Agentic 编程实际落地效果不稳定的问题。通过结构化的 Agent 技能定义和开发流程规范，帮助工程师更系统地使用 AI 进行软件开发，强调"真正有效"的方法而非理论，适合在生产环境中探索 AI 辅助开发的团队和个人。
TypeScript · ★ 15,756 · 🍴 1,076 · 📈 369 stars today
Desktop app to manage markdown knowledge bases
中文介绍 Tolaria 是一款桌面端 Markdown 知识库管理应用，为个人或团队提供本地化的笔记与文档组织能力。支持管理以 Markdown 格式存储的知识库，定位类似 Obsidian，强调简洁的桌面体验，适合偏好本地存储、对数据隐私敏感的写作者、研究人员和知识工作者。
Python · ★ 3,185 · 🍴 304 · 📈 515 stars today
open-source healthcare ai
中文介绍 面向医疗健康领域的开源 AI 项目，致力于将大语言模型等 AI 技术应用于医疗场景，如辅助诊断、医学文献理解、患者问答等。作为开源方案，可供医疗机构、研究人员和开发者在合规前提下自由部署和定制，降低医疗 AI 的技术门槛与商业授权成本。
Python · ★ 8,631 · 🍴 1,289 · 📈 28 stars today
LMCache: Supercharge Your LLM with the Fastest KV Cache Layer
中文介绍 LMCache 是专为大语言模型推理设计的高性能 KV Cache 层，通过优化键值缓存的存储与复用，显著降低重复前缀的计算开销，提升 LLM 推理吞吐量并降低延迟。可与 vLLM 等主流推理框架集成，适合需要在生产环境中大规模部署 LLM 的工程师和 AI 基础设施团队。
★ 16,971 · 🍴 1,741 · 📈 827 stars today
PM Skills Marketplace: 100+ agentic skills, commands, and plugins — from discovery to strategy, execution, launch, and growth.
中文介绍 面向产品经理的 AI Agent 技能市场，涵盖 100 多个专为 PM 工作流设计的技能、命令与插件，覆盖用户调研、产品策略、需求执行、发布管理和增长分析等全阶段。可与 AI 编码助手或 Agent 框架配合使用，帮助产品经理通过 AI 提升日常工作效率，减少重复性文档和分析工作。
Go · ★ 6,006 · 🍴 540 · 📈 400 stars today
Advanced DNS tunneling VPN for censorship bypass, optimized beyond DNSTT and SlipStream with low-overhead ARQ, resolver load balancing, high packet-loss stability and speed.
中文介绍 基于 DNS 隧道技术的高级 VPN 工具，专为在网络审查严格环境下突破封锁而设计。相比 DNSTT 和 SlipStream 进行了多项优化，包括低开销 ARQ 重传机制、DNS 解析器负载均衡、高丢包环境下的稳定性增强，在恶劣网络条件下仍能保持较高速度，适合需要在受限网络中可靠翻墙的用户。
Shell · ★ 112,404 · 🍴 18,335 · 📈 1,026 stars today
A complete AI agency at your fingertips - From frontend wizards to Reddit community ninjas, from whimsy injectors to reality checkers. Each agent is a specialized expert with personality, processes, and proven deliverables.
中文介绍 一套完整的 AI 代理机构工具集，内置多个拥有专属人格和工作流程的专业 Agent，涵盖前端开发、社区运营、创意注入、事实核查等多种角色。每个 Agent 具备明确的职责边界和处理流程，适合需要用 AI 模拟完整代理团队工作模式的创业者、自由职业者和数字营销从业者。
C · ★ 134,327 · 🍴 8,063 · 📈 103 stars today
Microsoft PowerToys is a collection of utilities that supercharge productivity and customization on Windows
中文介绍 微软官方开源的 Windows 效率增强工具集，涵盖窗口管理（FancyZones）、批量文件重命名、颜色拾取器、键盘映射、快速启动器（PowerToys Run）等数十种实用工具。基于 C# 和 C++ 开发，持续迭代更新，是 Windows 高级用户和开发者提升桌面工作效率的必备工具。
👍 2
Multi-agent systems communicate mostly through text, paying a lossy and expensive decode and re-encode cost. KV-cache communication is a promising alternative, yet most prior work is homogeneous, using duplicate copies of the same model, and avoids the central challenge of cross-model latent alignme
👍 2
Interactive LLM agents are becoming part of daily work, but they do not reliably become easier to work with over time: a correction remembered in one session may still be violated in the next. We study this gap between preference access and preference compliance. In tasks derived from anonymized rea
👍 2
Multimodal Large Language Models (MLLMs) have shown promising reasoning capabilities in general domains, yet their performance remains limited in specialized settings such as healthcare, especially in multilingual and low-resource scenarios. This gap is critical in regions like rural India, where pa
👍 24
Holistic visual tokenizers are fundamental to unified multimodal models (UMMs) as they map diverse visual inputs into a unified representation space. In this paper, we present HYDRA-X, the first UMM that unifies image and video tokenization within a single Vision Transformer (ViT). Our design is dri
👍 80
Ultra-long-context capability is becoming indispensable for frontier LLMs: agentic workflows, repository-scale code reasoning, and persistent memory all require the model to jointly attend over hundreds of thousands to millions of tokens, yet the quadratic cost of softmax attention makes this untena
👍 4
Search Agents -- large language models augmented with search tools -- have intensified the need for future-proof evaluation benchmarks. Existing benchmarks such as BrowseComp rely on static knowledge, making them vulnerable to test-set contamination and parametric memorization. Consequently, models
👍 104
Large language model (LLM) agents have achieved strong performance on a wide range of benchmarks, yet most evaluations assume static environments. In contrast, real-world deployment is inherently dynamic, requiring agents to continually align their knowledge, skills, and behavior with changing envir
👍 16
Latent chain-of-thought compresses reasoning by replacing visible reasoning traces with continuous hidden-state recurrence, but existing formulations are difficult to optimize with standard on-policy reinforcement learning (RL) and hard to interpret causally. Our key insight is that a single pair of
👍 2
The potential impacts of world models (WMs, i.e., learned simulators) on robotics are far-reaching -- policy evaluation, policy improvement, and test-time planning -- all with limited real-world interaction. To unlock these downstream capabilities, a WM needs to jointly satisfy three desiderata: (i)
👍 19
LLM-based agents have shown increasing potential in automating scientific discovery. Given an optimizable metric and an execution environment, they can propose, validate, and iterate scientific solutions, and have produced results that outperform human-designed approaches. As model capabilities cont
👍 73
Recent image generators have demonstrated impressive photorealism and instruction-following capabilities in single-image generation and editing. However, constrained by their architectures, they cannot achieve interleaved generation (text-image sequence), which has crucial applications in visual nar
👍 3
Anticipating LLM behavioral tendencies from low-cost psychometric probes is critical for safe deployment, but only if self-reports (SR) reliably predict behavior. Recent work documented substantial SR-behavior dissociation in LLMs, but relied on broad personality traits (Big 5) that predict specific
👍 10
Multimodal image fusion aims to integrate complementary information from different modalities into a fused image that preserves rich local details while maintaining globally consistent appearance. Existing approaches build shared representations on 2D feature grids, which excel at modeling local str
👍 12
Speculative decoding (SD) addresses the high inference costs of LLMs by having lightweight drafters generate candidates for large verifiers to validate in parallel. Existing draft-verify methods use binary decisions: accept or fully recompute. Yet we find that many rejected tokens can be verified co
👍 10
Deep search requires agents to answer complex questions through multi-step web search, browsing, evidence comparison, and synthesis. A central challenge is deciding how to search when several directions look plausible but only some will later lead to reliable evidence. If an agent greedily follows t
👍 3
Compact language models (LMs) reduce cost, latency, and deployment risk for tool agents. Yet MCP-style tool use requires more than isolated function calling: an agent must discover tools from live catalogs, satisfy schemas, preserve dependencies across intermediate outputs, and ground final response
👍 1
Autonomous web navigation remains challenging for LLM agents, and the strongest generalist systems rely on proprietary reasoning models whose inference cost is prohibitive for the repetitive tasks where such agents would be most useful. We argue this gap stems not from insufficient model capability
👍 8
Adversarial robustness evaluations of large language models (LLMs) typically report attack success rate (ASR) under fixed query budgets, implicitly treating all attacks as equally costly. In practice, the computational expense of different attack strategies can vary by orders of magnitude. Consequen
👍 1
We present Flash-GMM, a fused Triton kernel for efficient computation of Gaussian Mixture Models (GMMs) over large-scale data in a single GPU pass. By eliminating the need to materialize the full responsibility matrix in GPU memory, Flash-GMM achieves a 20times speedup over existing implementations
👍 2
Built on pretrained vision foundation models (VFMs), representation autoencoders (RAEs) have recently emerged as a promising approach for constructing semantically rich latent spaces for image generation. However, their reconstruction quality often remains suboptimal, largely because deep VFM repres
👍 1
Advances in handwritten text recognition have enabled large-scale transcription of historical documents, but still provide limited access to interpretable visual measurements for paleography, the study of historical scripts. In this paper, our main insight is that morphological script analysis, in p
👍 7
Visual reasoning requires integrating evidence distributed across regions, attributes, and relations, making single-chain reasoning prone to early perceptual commitment and hallucination. We propose Visual Para-Thinker++, a single-policy multi-agent framework in which one shared MLLM policy is insta
👍 6
On-policy distillation (OPD) trains a student on its own trajectories with dense per-token supervision from a stronger teacher, and often outperforms off-policy distillation and standard reinforcement learning. However, we find that its effectiveness implicitly relies on two assumptions that frequen
👍 55
Computer-use agents (CUAs) increasingly operate in runtimes that combine visual desktop control, command-line execution, code editing, browsers, and external tools. Existing benchmarks, however, often evaluate these interfaces as separable capabilities, leaving long-horizon cross-interface orchestra
👍 4
Representation alignment with pretrained vision models has recently shown strong potential for accelerating diffusion transformer training. By aligning intermediate diffusion features with clean-image representations from self-supervised vision encoders, existing methods improve convergence and gene
👍 71
Multimodal Large Language Models (MLLMs) have demonstrated remarkable success in visual understanding, yet their performance degrades significantly under real-world visual corruptions. While existing robustness enhancement approaches exist, they are limited: black-box feature alignment lacks interpr
👍 3
Robotic simulators are a cornerstone of modern research in aerial robotics, serving both as a vehicle for the development of new control algorithms and as the data source for training reinforcement learning (RL) policies. Yet, existing quadcopter learning environments often face a trade-off between
👍 2
Are tool-calling LLM agents equally safe throughout a conversation? We discover they are not: agents are most vulnerable at the very start of a session and become substantially safer after a few regular agentic tasks -- a phenomenon we term the cold-start safety gap. To study this systematically, we
👍 2
Large language models deployed as agents over large tool catalogs face a critical tool-retrieval bottleneck. As embedding-based retrieval approaches rely on compact encoders that may under-capture specialized tool semantics, parametric tool retrieval addresses this by encoding each tool as a virtual
👍 0
Large Language Models (LLMs) are increasingly used for zero-shot annotation and LLM-as-a-judge tasks, yet their reliability hinges on how model-internalized priors interact with user-provided instructions. We investigate three dimensions of this interaction: (1) how an LLM's familiarity with data an
@aiedge_ · 69.5K 粉丝 · 700.1K 阅 · 506 赞 · 68 转
TLDR: Anthropic just published the official playbook for prompting the most powerful AI model on earth - I translated it. Most people won't read this guide (it's buried in the API docs), which is
中文介绍 作者将 Anthropic 官方 API 文档中埋藏较深的 Claude 提示词指南整理翻译，内容涵盖提示词工程核心方法论。原文是 Anthropic 为其最新旗舰模型发布的官方「提示词手册」，因藏于技术文档而鲜有人读。适合想系统掌握 Claude 提示技巧的开发者参考。
@dhaber · 50.0K 粉丝 · 497.3K 阅 · 500 赞 · 57 转
One of the biggest ways that AI is transforming work (and also one of the most taboo subjects inside companies at the moment) is that most work discussions are being recorded now by default. This
中文介绍 探讨 AI 工具让职场沟通「默认全程录音」这一新现象——会议、讨论越来越多被自动记录，正成为 AI 改变工作方式中最显著却也最敏感的趋势之一。作者认为这一话题在企业内部属于禁忌，但影响不可忽视。
@Recursive_SI · 6.3K 粉丝 · 465.1K 阅 · 516 赞 · 71 转
Early results from Recursive’s automated AI research system on model training and GPU kernel benchmarks Today we are releasing early results from Recursive’s automated AI research system. Across three
中文介绍 Recursive 公司发布自动化 AI 研究系统的早期成果，方向涵盖模型训练与 GPU 内核基准测试共三个领域。这是该系统首次对外披露结果，属于 AI 自主做 AI 研究（automated AI research）的早期探索案例，具体数据尚未全部公开。
@0xCodez · 6.4K 粉丝 · 371.8K 阅 · 515 赞 · 56 转
Most people are using Claude Fable 5 like Sonnet 4.6 with a bigger context window. They prompt it. It works for 5 minutes. They close the tab. 9 out of 10 users have never run an agent system that
中文介绍 14 步搭建基于 Claude 的自我改进 Agent 系统，涵盖循环（loops）、动态工作流与例程（routines）设计。作者指出大多数人把 Claude 当普通聊天工具在用，从未真正运行过 Agent 系统。教程针对如何让 Agent 持续自我优化而非「用完就关」。
@saranormous · 143.5K 粉丝 · 194.8K 阅 · 614 赞 · 40 转
The mid-2026 investor's version of AI psychosis is a despair that nothing is investable, that we should put all our money into Anthropic and Nvidia and go home. I have never felt it. I have been sure
中文介绍 作者反驳「2026 年中期只有 Anthropic 和 Nvidia 值得投资」的悲观论断，认为这是投资者版本的 AI 焦虑症。本帖属于对当前 AI 投资情绪的观察与反思，讨论哪些方向仍有机会，语气偏个人判断。
@0x_rody · 1.7K 粉丝 · 193.2K 阅 · 513 赞 · 72 转
Claude writes your code, hands it over, and 3 tests are failing. You paste the errors back, it fixes one thing, breaks another, and you spend the evening as a messenger between Claude and your
中文介绍 分享在 Claude Code 中搭建「自我修复循环」的具体配置方案，解决代码生成后测试失败、来回传递错误信息效率极低的痛点。核心思路是让 Claude 自动捕获报错、重试修复，而非人工当「错误信使」，附实际搭建步骤。
@BrainsAndTennis · 10.5K 粉丝 · 187.4K 阅 · 539 赞 · 45 转
How do you build an agent that actually performs in a domain — one customers pick because it's better? The basics have been standardized over the past year: an agent is a while-loop around a model
@TheAhmadOsman · 61.0K 粉丝 · 74.9K 阅 · 507 赞 · 98 转
Anthropic wants the public to see one thing: the careful lab, the safety lab, the grown-up in the room trying to keep frontier AI from running off a cliff. However, the pattern around Anthropic does
中文介绍 批评 Anthropic 对外塑造「安全实验室」形象，但实际行为模式与开源 AI 社区存在冲突。作者梳理 Anthropic 围绕开源问题的一系列动作，认为其公众形象与实际立场之间存在落差，观点偏向质疑与批评。
@coinbase · 7.0M 粉丝 · 72.8K 阅 · 500 赞 · 62 转
TL;DR: Coinbase for Agents connects your AI agent directly to your Coinbase account so it can trade, pay, and execute workflows on your behalf, all within limits you control. Available today as an MCP
中文介绍 Coinbase 推出面向 AI Agent 的基础设施「Coinbase for Agents」，通过 MCP 协议连接 Agent 与用户 Coinbase 账户，允许 Agent 在用户设定的权限范围内自主完成交易、支付和工作流执行，当日起正式上线。
@RayDalio · 2.2M 粉丝 · 72.6K 阅 · 515 赞 · 93 转
What is the best approach to being effectively intelligent now that human intelligence and artificial intelligence are merging? Because I have been building computerized investment decision-making
中文介绍 Ray Dalio 结合自身数十年量化投资决策系统的构建经验，探讨人类智能与 AI 融合时代如何保持「有原则的思考」。核心观点是：AI 能力越强，越需要清晰的原则框架配合，而非单纯依赖模型输出。
@ORACLEAIFND · 31.9K 粉丝 · 63.6K 阅 · 1.5K 赞 · 563 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 ORACLE 项目宣称其自主 AI Agent 已在预测市场 Polymarket 上交易，并援引数据称 2026 年 Polymarket 逾 30% 活动来自算法及 AI 钱包。内容疑似项目推广，信源可信度存疑，需谨慎甄别。
@OracleTrdading · 39.7K 粉丝 · 61.5K 阅 · 1.5K 赞 · 577 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 与 ORACLEAIFND 账号内容高度重复，同样宣传 ORACLE AI Agent 在 Polymarket 自主交易，援引「30% 活动来自 AI 钱包」数据。多账号同步发布相同内容，疑似营销宣传，内容可信度存疑。
@OracleTrdade · 41.7K 粉丝 · 60.6K 阅 · 1.5K 赞 · 560 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 与 ORACLEAIFND 等账号内容完全一致，第三次重复发布 ORACLE AI Agent 在 Polymarket 交易的宣传内容。多账号矩阵同步推送，高度疑似营销刷量行为。
@OracleMarkett · 45.0K 粉丝 · 60.1K 阅 · 1.5K 赞 · 568 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 第四个账号发布与前三条完全相同的 ORACLE Polymarket AI Agent 宣传内容。结合多账号同步发布的模式，判断为协同营销推广，内容独立价值极低。
@elvissun · 45.0K 粉丝 · 42.0K 阅 · 502 赞 · 45 转
99% people are using /goal and loops wrong. The hype they hear is "long-running loops prompting autonomous agent": point it at a task, walk away, come back to working code. But top agentic engineers
中文介绍 针对 Claude 的 /goal 指令与循环机制的误用提出纠正，分享「30 小时内用一条 prompt 提炼完整产品」的实战方法论。作者认为顶级 Agentic 工程师的做法与大众认知有根本差异，附完整 playbook，适合深度使用 Claude Agent 的开发者。
@leerob · 258.6K 粉丝 · 36.8K 阅 · 586 赞 · 40 转
At Cursor, we run thousands of agents to help us train the next version of Composer. We give them research tasks, and if they aren't succeeding or run into issues, they DM us on Slack or page us via
中文介绍 Cursor 工程师分享公司内部递归 Agent 系统的实践：每天运行数千个 Agent 辅助训练下一版 Composer，Agent 遇到问题会主动通过 Slack 或 PagerDuty 通知团队。是目前少见的大规模 Agent 系统真实落地案例披露。
中文介绍 AllenAI推出olmo-eval评估工作台，专为模型开发循环设计，旨在为开发者提供系统化的模型评估工具，支持在模型迭代过程中进行全流程性能测试与对比分析。
OpenAI introduces three Academy courses that help people build practical AI skills, create repeatable workflows, and apply agents in everyday work.
中文介绍 OpenAI发布三门Academy新课程，帮助用户在工作中建立实用AI技能，内容涵盖构建可复用工作流程及在日常工作中应用AI智能体，旨在推动AI能力在职场中的普及与落地应用。
a quiet day lets us highlight a great concept from Peter Steinberger, Boris Cherny, and Andrej Karpathy
中文介绍 Latent Space发布AINews专题，聚焦Peter Steinberger、Boris Cherny及Andrej Karpathy提出的「循环堆叠」（Loopcraft）概念，探讨AI开发中循环结构的设计思路与实践价值。
Preply uses OpenAI to launch AI-generated lesson summaries, providing personalised feedback and language learning exercises.
中文介绍 语言学习平台Preply借助OpenAI技术推出AI生成课程摘要功能，为学习者提供个性化反馈和语言练习，将AI辅助与人类导师相结合，提升个性化学习体验。
中文介绍 TLDR AI日报（2026年6月12日）主要内容包括：OpenAI宣布收购Ona，Anthropic就此前政策立场作出调整，小米发布MiMo代码模型，涵盖当日AI领域重要动态。
Google DeepMind is funding research into the potential dangers of situations where millions of different AI agents interact with each other online. According to Rohin Shah, who directs the company’s AGI safety and alignment research, the mass-market arrival of agents that can carry out tasks without
中文介绍 Google DeepMind正资助研究数百万AI智能体大规模联网交互可能带来的安全风险。AGI安全与对齐研究负责人Rohin Shah表示，随着智能体进入大众市场，其自主行动的协同与冲突问题已引发公司高度关注。
a quiet day lets us reflect on a great essay
中文介绍 Latent Space发布AINews专题，结合Sarah Guo的深度文章，探讨开放模型生态、模型实验室与智能体实验室的竞争格局，以及AI训练中「不可训练」能力的边界问题。
OpenAI supports the EU Code of Practice on AI content transparency, advancing provenance standards and tools to help people understand AI-generated content.
中文介绍 OpenAI宣布支持欧盟AI内容透明度行为准则，积极参与AI内容溯源标准的制定，并推进相关工具建设，帮助用户识别和理解AI生成内容，以推动欧洲可信AI生态系统建设。
Learn how BBVA scaled ChatGPT Enterprise to 100,000 employees and partnered with OpenAI to accelerate AI-powered banking transformation worldwide.
中文介绍 西班牙大型银行BBVA已将ChatGPT Enterprise部署至全球10万名员工，并与OpenAI深度合作，推动AI驱动的银行业务转型，将人工智能定位为其核心战略支柱。
Discover how astrophysicist Chi-kwan Chan uses Codex to build black hole simulations, helping scientists study extreme physics and test Einstein’s theory of general relativity.
中文介绍 天体物理学家Chi-kwan Chan利用OpenAI Codex辅助构建黑洞模拟程序，用于研究极端物理现象并验证爱因斯坦广义相对论，展示AI编程工具在前沿科学研究中的实际应用价值。
OpenAI plans to acquire Ona to expand Codex with secure, persistent cloud environments, enabling long-running AI agents across enterprise workflows.
中文介绍 OpenAI宣布计划收购云基础设施公司Ona，旨在为Codex扩展安全、持久的云端运行环境，支持长时运行的AI智能体在企业级工作流程中的部署与应用。
中文介绍 Hugging Face发布PyTorch性能分析系列第二篇，详细介绍从标准nn.Linear层到融合MLP（多层感知机）的优化路径，探讨通过算子融合提升模型推理效率的方法与实践技巧。
中文介绍 TLDR AI日报（2026年6月11日）主要内容包括：Anthropic CEO Dario Amodei就AI政策发表看法，Google DeepMind发布DiffusionGemma模型，WhatsApp宣布将开放对AI机器人的支持等当日重要AI资讯。
Access OpenAI models and Codex through Oracle Cloud, using existing commitments to build and deploy AI with enterprise security and governance.
中文介绍 OpenAI与Oracle达成合作，企业客户可通过现有Oracle Cloud承诺额度直接调用OpenAI模型及Codex服务，在企业级安全与治理框架下构建和部署AI应用，降低企业上云使用门槛。
中文介绍 Google DeepMind发布DiffusionGemma模型，采用扩散生成机制实现文本生成速度较传统方法提升约4倍，为大语言模型推理效率的提升提供了新的技术路径。
美股方面，SPY现报741.75美元，RSI14为52.9，SMA20/50/200多头排列支撑，近5日上涨0.57%，整体技术结构稳健，距52周高位仅2.45%。QQQ近5日涨幅2.31%，技术面同样维持多头排列。然而个股分化明显：MSFT的RSI14跌至37.1，MACD死叉（-4.135 vs 信号线1.271），空头排列压制；META的RSI14仅34.7，MACD柱状深度为负。VIX单日暴跌9.05%至17.68，5日累跌17.81%，市场对短期波动的定价快速回落，与标普技术面改善相互印证。加密市场整体处于显著压力区间：当前加密恐慌贪婪指数仅为13（极度恐慌），加密总市值2.26T美元，24小时微跌0.16%，BTC主导率56.4%，显示资金向比特币集中避险。BTC现报63512美元，RSI14为32.8，位于SMA20（68140）、SMA50（74418）及SMA200（77884）三线全面下方，空头排列形态完整。ETH的RSI14更低至30.9，较52周高位下跌66.38%，技术面承压尤重。中概股普遍弱势，阿里巴巴RSI14触及超卖（29.6），但空头排列未破；腾讯控股近5日小幅回升2.29%，MACD出现改善但仍处负值区间。商品外汇方面，黄金期货单日大涨3.66%，但MACD仍为负值，SMA50（4588）远高于现价（4239.9），中期趋势有待确认；WTI原油近5日跌幅达6.9%，RSI14为37.4。美元指数DXY的RSI14升至60，多头排列维持，接近52周高位。
当前价格566.98美元，RSI14读数34.7逼近超卖区间，MACD为-12.809且信号线为-7.969，柱状值持续为负，动量偏弱。SMA20/50/200分别为604.21、621.83、658.09，呈完整空头排列，价格较52周高位回撤28.79%。近5日跌幅4.39%，技术结构整体偏空。
RSI14为29.6，已触发超卖信号，但价格仍处空头排列之中（SMA20=125.81、SMA50=130.33、SMA200=149.57均高于现价112.82）。MACD=-4.954，信号线=-3.354，死叉状态延续。近5日跌幅6.81%，较52周高位下跌41.44%，超卖信号虽出现，但趋势下行动能尚未消退。
当日大涨3.66%至4239.9美元，但RSI14仅37.3，仍处中性偏低区间。MACD=-116.35，信号线=-87.26，仍处绿柱阶段，短期均线SMA20=4423.13与SMA50=4588.26均高于现价，近5日跌幅2.24%，单日反弹是否能扭转中期下行趋势尚需观察，目前技术状态中性。
近5日上涨2.31%，SMA20=721.5、SMA50=681.81、SMA200=625.38形成多头排列，现价721.34略低于SMA20。RSI14为55.2，处中性偏强区间。MACD=8.315虽低于信号线13.604，MACD柱状值为负，但整体趋势标注为看涨，较52周低点上涨37.75%，技术结构偏上行。
VIX当前报17.68，单日下跌9.05%，近5日下跌17.81%，为本次扫描中单日跌幅最大标的。MACD=0.309金叉信号线-0.098，出现短期金叉形态，但整体趋势标注为「bearish」，SMA50(18.25)与SMA200(18.54)均高于现价，且触发「死叉(SMA50↓SMA200)」信号，VIX持续走低对应股市波动预期收窄，与美股技术面形成呼应。
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
本报告仅供技术指标解读参考，所有内容均基于历史行情数据计算所得的技术指标读数，过去走势不代表未来表现。报告不构成任何投资建议或买卖依据，投资者应结合自身风险承受能力及基本面信息独立判断。
Iran’s foreign minister says end to war on all fronts includes Lebanon but Tehran won’t give up nuclear programme; US president has reportedly talked to Netanyahu Iran’s official Islamic Republic News Agency (IRNA) has cautioned against media speculation about a potential memorandum of understanding
中文摘要 伊朗外长表示，与美国达成的停火协议涵盖黎巴嫩战线，但伊朗不会放弃核计划。伊朗官方通讯社伊通社警告外界不要对潜在备忘录进行媒体炒作。据报道，美国总统特朗普已与以色列总理内塔尼亚胡通话，各方就协议细节仍在协商之中。
Climate minister Chris Bowen says country must prepare for changing world and can play bigger role in reducing emissions Australia will find exporting fossil fuels increasingly difficult but can switch to exporting clean energy products, the president of the next UN climate negotiations has declared
中文摘要 澳大利亚气候部长、下届联合国气候大会（COP）主席克里斯·鲍文表示，澳大利亚出口化石燃料将愈发困难，但该国可转型为清洁能源产品出口国，在全球减排中发挥更大作用，需为不断变化的世界做好准备。
The arrest of U Min Zin, who did graduate studies at U.C. Berkeley and directs a research group on Myanmar, took place soon after President Trump met with Xi Jinping in China.
中文摘要 中国以间谍罪逮捕美国学者吴敏津（U Min Zin）。他曾就读于加州大学伯克利分校，现任一缅甸研究机构负责人。逮捕发生在美国总统特朗普访华并与习近平会面后不久，引发外界关注。
Data analytics company loses on 22 out of 23 counts in lawsuit disputing how Swiss government rejected firm’s services The US technology company Palantir has lost a legal challenge to force a Swiss independent magazine to publish its responses to articles about how the Swiss government rejected its
中文摘要 美国数据分析公司Palantir在瑞士提起的诉讼中，23项诉求中22项败诉。该诉讼旨在强制一家瑞士独立杂志刊登公司回应文章，内容涉及瑞士政府拒绝采用Palantir服务的相关报道。
The United Nations said it was “deeply concerned” about the arrests of dozens of women, and reported that two people were killed in protests organized to support them.
中文摘要 阿富汗爆发罕见公开抗议活动，民众反对塔利班统治规定。联合国对数十名女性遭逮捕表示「深切关注」，并报告称已有两人在支持被捕女性的抗议活动中遇难。
U.S. and Iranian officials said final details were still being worked out, but President Trump and Iran’s foreign minister both said they were close to an agreement. Previous potential deals have evaporated at the last minute.
中文摘要 美伊官员表示，停火协议最终细节仍在磋商，但特朗普总统与伊朗外长均称双方已接近达成协议。官员指出协议触手可及，但也提醒此前多次潜在协议均在最后关头破裂。
A senior administration official said the two sides were “not quite at the finish line yet.”
中文摘要 美国白宫表示，美伊停火协议「非常接近」达成。一名美国高级政府官员称，双方目前「尚未完全到达终点线」，谈判仍在进行中，协议最终细节有待确认。
中文摘要 该条目为伊朗与美国停火谈判实时动态更新页面，暂无具体摘要内容，为相关新闻直播报道的最新进展汇总入口。
Musk is now worth $1.11tn according to the Bloomberg rich list, while SpaceX listed on the Nasdaq stock exchange with a value of $2.2tn.
中文摘要 埃隆·马斯克成为全球首位万亿美元富翁。据彭博富豪榜数据，其净资产达1.11万亿美元。SpaceX在纳斯达克交易所上市，估值达2.2万亿美元，股价大涨是推动马斯克财富突破万亿门槛的主要因素。
The lawsuit accuses OpenAI of failing to intervene despite warning signs in daughter's ChatGPT conversations.
中文摘要 一名母亲在美国对OpenAI提起诉讼，指控其女儿的死亡与使用ChatGPT有关。诉状称，OpenAI未能在发现女儿与ChatGPT对话中出现的警示信号后及时干预，存在重大过失。
The US president has sought to reshape the capital city's image and institutions through series of plans and projects.
中文摘要 美国一名法官维持原有命令，要求将特朗普名字从肯尼迪中心移除。特朗普总统此前推行一系列计划，试图重塑首都华盛顿的城市形象及相关机构，此次司法裁定令其部分意图受阻。
The deal which will pave the way for hostilities to end is close to being finalised, the US, Iran and mediators Pakistan say.
中文摘要 伊朗表示，若停火协议达成，霍尔木兹海峡将重新开放。美国、伊朗及调停方巴基斯坦均称，结束敌对行动的协议已接近最终敲定，各方正就最后细节展开协商。
For months, mice have been found in tea kettles, crunched by car tires and even appeared in people’s beds. In one town, the end might be in sight.
中文摘要 澳大利亚西部正遭受严重鼠患，数月来老鼠出没于水壶、汽车轮胎下乃至居民床上。目前部分城镇出现鼠患缓解迹象，当地居民和当局正努力应对这一持续数月的鼠灾。
Reuters reports UAE agreed to unlock billions for Iran, but Abu Dhabi swiftly issued a categorical denial.
中文摘要 有消息人士援引路透社报道称，阿联酋同意解冻数十亿美元伊朗被冻结资金，以配合美国推动停火的努力。但阿布扎比方面随即发表声明，对此予以断然否认。
Progress for Kyiv's membership bid given the green light after Hungary's new government lifts Budapest's veto.
中文摘要 欧盟正式同意启动乌克兰和摩尔多瓦的入盟程序。在匈牙利新政府解除布达佩斯的否决立场后，乌克兰入盟进程获得放行，此举标志着两国加入欧盟的谈判进程向前迈出重要一步。
SpaceX’s first day on the stock market transformed the startup into one of the world’s most-valuable public companies, handed buyers of the IPO a 19% return and turned its founder Elon Musk into the world’s first trillionaire. Ed Ludlow reports. (Source: Bloomberg)
中文摘要 SpaceX完成史上最大规模IPO，首日上市融资750亿美元，股价上涨19%，市值跻身全球最有价值上市公司之列。创始人埃隆·马斯克凭借此次上市身家突破1万亿美元，成为全球首位万亿富翁。
The US Commodity Futures Trading Commission is considering whether to block CME Group Inc.’s bid to launch a round-the-clock oil contract, heightening tensions between the market stalwart and its regulator.
中文摘要 美国商品期货交易委员会（CFTC）正考虑阻止CME集团推出全天候24小时石油期货合约的计划，此举加剧了监管机构与CME之间的紧张关系。
Bill Bradley, former US Senator and two-time NBA champion, discussed the New York Knicks' recent success, highlighting how Jalen Brunson gave up millions in free agency to allow the team more financial flexibility to build a stronger roster. He emphasized that this special group's dedication and sha
中文摘要 前美国参议员、两届NBA冠军比尔·布拉德利评价纽约尼克斯队近期成功，指出球星贾伦·布朗森在自由市场放弃数百万美元薪资，为球队腾出财务空间组建更强阵容，团队牺牲精神激励了纽约市民。
SpaceX had the largest stock-market debut in history when it went public on June 12. The company raised $75 billion in the initial public offering and ended its first day on the public markets with a market capitalization of around $2.2 trillion. The IPO turned SpaceX founder Elon Musk into the worl
中文摘要 SpaceX于6月12日完成史上规模最大的股市首发，融资750亿美元，首日收盘市值约达2.2万亿美元，刷新IPO历史纪录。此次上市使创始人马斯克成为全球首位净资产超过万亿美元的富豪。
Heather Pringle, CEO of the Space Foundation, discussed the impact of SpaceX's historic $75 Billion IPO on the broader space industry and investment landscape. She highlighted that SpaceX's public listing marks a significant milestone for a maturing space sector, demonstrating viable pathways to pro
中文摘要 太空基金会CEO希瑟·普林格尔表示，SpaceX 750亿美元的历史性IPO对整个航天行业及投资格局具有深远影响，标志着太空产业走向成熟，预计将带动更多资本涌入航天领域。
Musk is now worth $1.11tn according to the Bloomberg rich list, while SpaceX listed on the Nasdaq stock exchange with a value of $2.2tn.
中文摘要 SpaceX在纳斯达克上市首日市值达2.2万亿美元，马斯克个人财富依据彭博富豪榜升至1.11万亿美元，成为全球有史以来首位净资产突破万亿美元的富翁。
Deal is critical to David Ellison’s ambition to create new media empire
中文摘要 派拉蒙以1110亿美元收购华纳兄弟的交易获得美国反垄断监管机构批准。此次并购是派拉蒙首席执行官大卫·埃里森打造新媒体帝国战略的核心举措。
Barclays and LSE Group bosses miss out despite early-stage approval
中文摘要 花旗集团首席执行官简·弗雷泽在英国国王生日授勋名单中获封女爵士（DBE）荣衔。巴克莱银行及伦敦证券交易所集团负责人尽管此前获得初步批准，最终未能上榜。
New York City Comptroller Mark Levine discussed the fast-tracking of SpaceX's inclusion into major indexes such as MSCI Global Standard and FTSE Russell, highlighting concerns about the abandonment of traditional criteria like seasoning periods and earnings track records. Levine emphasized the unpre
中文摘要 纽约市财务总监马克·莱文对SpaceX被快速纳入MSCI全球标准指数及富时罗素指数表示担忧，认为相关机构放弃了传统的上市缓冲期及盈利记录等纳入标准，存在公司治理结构方面的隐患。
Barry speaks with Jean Eric Salata, chair of EQT group. They discuss his time working in Asian private equity investment along with what he sees as necessary to become a good investor across different cultures including what he learned in Japan and Hong Kong. (Source: Bloomberg)
中文摘要 彭博《商业大师》节目专访EQT集团主席让-埃里克·萨拉塔，两人探讨其亚洲私募股权投资经历，以及在日本、香港等不同文化背景下成为优秀投资者所需具备的素质与经验。
Jeff Yass, the founder of the Wall Street trading firm Susquehanna International Group, has long been a New York Knicks fan. But he couldn’t fully savor the moment when the team completed its epic comeback to win game four of the NBA finals.
中文摘要 华尔街交易公司苏斯奎哈纳国际集团创始人杰夫·亚斯是纽约尼克斯球迷，但尼克斯在NBA总决赛第四场完成史诗级逆转夺胜，却令其公司体育交易部门蒙受创纪录亏损，使他无法尽情享受球队胜利的喜悦。
Rocket and AI group’s shares jumped by nearly a fifth after raising $75bn in record initial public offering
中文摘要 SpaceX以750亿美元创纪录IPO完成上市，首日股价涨幅近20%，市值大幅攀升。这家火箭与人工智能集团的上市使创始人马斯克成为全球首位万亿美元富翁。
15 回复 · 程序员 节点
18 回复 · 程序员 节点
5 回复 · Apple 节点
23 回复 · 程序员 节点
22 回复 · 程序员 节点
23 回复 · Apple 节点
8 回复 · Apple 节点
19 回复 · Apple 节点
8 回复 · Linux 节点
7 回复 · Linux 节点
「君の公益」 上架 claude-fable-5 地址 muyuan.do 不要再给我发私信或者艾特我，把我惹急了我就开三级登录了 公益站的本质是让没钱的佬友也能体验一下大模型，不得滥用！ 分发我一直有安排佬友去查，不要拿我的公益站去搞黄色，搞政治敏感，不要挑战我的底线 142 个帖子 - 125 位参与者 阅读完整话题
目前从事这方面工作，因为架构设计需求分析都差不多了，顺手就实现了一套，功能日渐完善，计划开源。到时候欢迎大家试用呀。 12 个帖子 - 12 位参与者 阅读完整话题
周五的夜，马上是周末了， 85 个帖子 - 82 位参与者 阅读完整话题
项目我一直是拿5.5xhigh开发的，系统里的tts一直有问题，因为是本地部署的所以一直在喊5.5改框架改参数修bug，但是一直有几个问题解决不了，但也能正常使用也就算了 但今天生成的音频又出问题，我真是艹了 又一次喊5.5定位问题的时候，看着没几个能用的公益站 ，突然想起来any大善人有fable5能用，于是赶快更新cc，接入ccs使用。retry几次后，从线程调用入手直接给我列了4个点，那是字字珠玑，一看5.5感觉纯在说废话（也有可能是对自己写的东西太信任了） 那还说啥了，赶紧给fable去写。才修了两步，完美解决了问题的同时，tts的生成速度还快了不少，给我高兴坏了 。而且在我指出之后，
这个方法适用于可以使用 至于具体切换是否需要在Claude重新调整（我想大概是要的 但是不麻烦） 首先就正常填写各项内容 无需模型映射 手动指定其实也无需 直接使用就行 路由也无需打开 打开Claude code（不好截图就不截图了） 选择 输入模型名称 可以参考我下面的 注意选 Offer 1M-context variant 然后Apply 最后记得选1M上下文的哦 38 个帖子 - 19 位参与者 阅读完整话题
之前理解错了规则，已经按照规则编辑了，我与ai讨论产生的总结，拿来一起讨论。 24 个帖子 - 22 位参与者 阅读完整话题
今天，我们发布并开源 Kimi K2.7 Code 编程模型。 内外部基准评估显示，Kimi K2.7 Code 相比 K2.6 模型显著提升了长上下文编程场景的指令遵循能力、长程编程任务的性能表现，并且大幅改善了在长程任务中的过度思考倾向，平均 token 消耗减少 30%。 在评估代码能力的内部外基准测试中，K2.7 Code 相比 K2.6 性能显著提升：Kimi Code Bench v2 提升 21.8%、Program-Bench 提升 11%、MLS Bench Lite 提升 31.5%。 模型代码能力的进化带来了 agentic 能力的提升。在评估 Agent 自主化执行能力
docs.oracle.com Always Free Resources Learn what Always Free resources are available to all Oracle Cloud Infrastructure users. 甲骨文云的免费账号继arm机器限速和韩国春川地区无法注册和开出arm机器后，现在直接将之前的永久4C24G的额度砍半，最多只能2C12G。 6.13更新，之前在oracle系统里显示的4c24g的实例成本分析也跟着文档更新有计费了，接下来就看15号一个月过半后有没有账单了。 166 个帖子 - 110 位参与者 阅读完整话题
从fable 12万字提示词里，已经对侮辱谩骂ai的行为做了定性： 所以再骂ai有可能会被拒绝服务。 32 个帖子 - 27 位参与者 阅读完整话题
本篇关于GPT接码问题和开通订阅后是否还需要接码的问题 遇到的问题 接码是不是必须要有国外的卡或者是gg卡（giffgaff） 不管是通过任何渠道去订阅plus或者pro5X以及pro20X的账号前，在终端codex login或者codex App内进行登录时，跳出来需要接码，此时此刻如果去购买订阅，然后有没有可能就不需要接码了？ 通过小白卡如何接码？以及充值花费是否必须是国外的实体卡或者虚拟卡？ 解答环节 接码需要用到国外的卡号即可，并非是实体卡 假设你在付款前，也就是没有订阅前，需要接码，然后后续不管是通过Apple ID的方式还是虚拟卡的方式，以及国内信用卡套Google play的方