swyx@swyx · 1天前61

congrats to our friends @ona_hq on joining @openai! see their talk here for alpha on what’s next for Codex 👀

译swyx 祝贺 ONA 团队加入 OpenAI，并引用其演讲透露 Codex 的下一步方向。同时引述 @aiDotEngineer 观点：运行 Agent 集群需要三层——Runtime 与编排触发器已解决，但协调层（Agent 间任务交接、验证阶段、继续执行）仍未解决。Stripe 和 RAMP 各自自建了内部方案 Minions 和 Inspect。@loujaybee 指出 GitHub 作为协调层很差——噪音大、仅为人类设计，不适合数百个并行 PR。他提出的修复方案是构建一个 CLI 网关，供本地编码 Agent 调用以检查当前 SDLC 阶段是否已通过、能否继续。

Rohan Paul@rohanpaul_ai · 1天前67

OpenAI is buying Ona to give Codex agents a secure cloud desk that stays open after humans leave. Codex already has 5M weekly users, up 400%, but harder work breaks the old chat pattern because agents need tools, files, credentials, logs, and time. Ona adds persistent cloud workspaces, meaning an agent gets a controlled place to run commands, inspect systems, preserve context, and resume work without depending on one device. The enterprise angle is the real acquisition target: companies want agents inside their own cloud boundary, with scoped credentials, review trails, access limits, and auditable activity. This makes Codexmore like a managed execution layer for tests, bug fixes, refactors, vulnerability work, migrations, and multi-step knowledge tasks.

译OpenAI 宣布收购 Ona，其安全云端执行技术可为 Codex 智能体创建持久云端工作空间——用户离开后，智能体仍可持续运行命令、检查系统、保留上下文并跨设备恢复任务。目前 Codex 周活用户达 500 万（增长 400%）。收购旨在强化企业级部署：智能体可在企业云边界内运行，具备作用域凭证、审核追踪、访问限制和可审计活动，适用于测试、漏洞修复、重构、迁移等多步骤任务。收购完成后，Ona 团队将加入 OpenAI Codex 团队。

Rohan Paul@rohanpaul_ai · 1天前29

Guangzhou Auto Show in China hyper-realistic robotic suit that many initially mistook for XPeng’s IRON humanoid robot.🙂

译中国广州车展 一个超逼真的机器人服装，许多人最初误以为是小鹏的IRON人形机器人。🙂

Sam Altman@sama · 1天前41

really looking forward to working together!

译非常期待合作！

Greg Brockman@gdb · 1天前69

welcome @ona_hq to the team, to help organizations deploy agents securely in production!

译欢迎 @ona_hq 加入团队，帮助组织在生产环境中安全部署智能体！

Rohan Paul@rohanpaul_ai · 1天前82

WSJ: OpenAI is considering deep price reductions as competition with Anthropic intensifies. Anthropic is pressuring OpenAI because its strongest growth is coming from developer and coding workflows, specially with Claude Code, where users can generate huge token volume every day and quickly make Claude part of their normal work. OpenAI is still the bigger consumer brand, but in this fight the valuable prize is not casual chat users, it is enterprise teams paying metered bills for coding agents, automation, and internal tools. The difference is that Anthropic seems to have a sharper wedge in high-spend technical work, while OpenAI has to defend ChatGPT’s broad lead and stop Claude from becoming the default tool inside companies. --- wsj. com/tech/ai/openai-considers-drastic-price-cuts-anticipating-war-for-users-with-anthropic-9b8c178e

译WSJ 报道，OpenAI 正考虑大幅降价以应对与 Anthropic 的竞争。Anthropic 增长主要来自开发者和编码工作流，Claude Code 消耗大量 token，已让企业团队将其融入日常工作。OpenAI 虽在消费品牌上更大，但企业市场才是关键——企业为编码智能体、自动化等工具付费。同时，OpenAI 在 IPO 前准备对 ChatGPT 进行史上最大改版，将其打造成涵盖编码、AI 智能体、图像生成和商业软件的超级应用，改版将在未来几周陆续推出。OpenAI 将更多资源投入编码工具 Codex，目标实现 Codex 工程负责人所说的“个人智能体”。

SemiAnalysis@SemiAnalysis_ · 1天前67

GPU Racks hitting 400kW? Legacy data centers wont be able to handle it and the grid WILL get throttled. Radiant's 12 month, dirt to AI production, was made possible by bypassing the grid. Head of Infrastructure, Patrick Wohlschlegel tells @JordanNanos https://youtu.be/SQtavfviwrs

译GPU 机架达到 400kW？传统数据中心无法应对，电网将被限流。 Radiant 耗时 12 个月，从零到 AI 生产，正是因为绕过了电网。基础设施主管 Patrick Wohlschlegel 告诉 @JordanNanos

jason@jxnlco · 1天前61

I met @jolandgraf et la with @humford and Sandeep over a year ago and im even more excited to see them at the office soon! https://openai.com/index/openai-to-acquire-ona/

译一年多前我见到了@jolandgraf等人、@humford和Sandeep，现在更兴奋很快就能在办公室见到他们！ https://openai.com/index/openai-to-acquire-ona/

Chubby♨️@kimmonismus · 1天前62

Anthropic makes more revenue than any other AI model company right now, and it still can't get its new data centers funded on its own. The Information report says lenders want Google to guarantee the lease payments first. This is the same Google that helps design Anthropic's chips and is selling it around $200 billion in computing power. Odd position for the revenue leader to be in.

译Anthropic 目前营收超过任何其他 AI 模型公司，却仍无法靠自身获得新数据中心的融资。《The Information》报道称，贷款机构要求 Google 先担保租赁付款。正是这家 Google，协助设计 Anthropic 的芯片，并向其出售约 2000 亿美元的计算能力。营收领先者竟处于这种尴尬境地。

Runway@runwayml · 1天前25

The sold out New York premiere of the 2026 Runway AI Festival is tonight. See you at the movies.

译2026 年 Runway AI Festival 纽约首映门票已售罄，今晚举行。电影院见。

Berryxia.AI@berryxia · 1天前59

这个创新点可以啊！ 叫做River AI直接把个人AI的所有权塞回用户手里，把大公司垄断AI的局面打破了！ 现在最好的模型都掌握在少数几家巨头手里，你的对话、数据、偏好、决策权全在他们服务器上。 而River AI要建的，是一个完全属于你的AI栈，它只为你服务，只按你的价值观运行，只在你的控制下进化。 他们今天宣布的，就是要把AI从“租来的工具”变成“你的个人智能伙伴”。 不再是公司告诉你AI能做什么，而是你决定AI长什么样、怎么想、怎么帮你。 最狠的是这不是空谈，他们已经在招顶尖研究员和工程师，一起把这个愿景落地。 个人AI的时代，不是远景，是正在发生的现实。 以前大家默认AI的未来一定是越来越中心化、越来越被大厂掌控，结果River AI直接反其道而行，把主权还给每一个普通人。

译River AI 今日宣布成立，致力于创建完全由用户拥有和塑造的个人 AI。创始人指出，当前最好的 AI 模型由少数大企业控制，用户的对话、数据和决策权都掌握在它们手中。River AI 正在构建一套全新的个人化 AI 栈，完全按用户价值观运行，只在用户控制下进化，将 AI 从“租来的工具”变为“专属智能伙伴”，把主权还给每一个普通人。

Luma@LumaLabsAI · 1天前62

"AI can make it. The great idea still has to be human". Dream Brief, named one of five Cannes contenders for shots by Lora Schulson of 72andSunny. - https://shots.net/news/view/cannes-contenders-2026-usa-2

译"AI可以制作。但伟大的创意仍必须来自人类。" Dream Brief，被72andSunny的Lora Schulson评为五个戛纳竞争者之一。 - https://shots.net/news/view/cannes-contenders-2026-usa-2

🚨 AI News | TestingCatalog@testingcatalog · 1天前71

OpenAI ❤️ Ona OpenAI acquired Ona, a cloud execution platform for AI agents. > Its secure cloud execution technology will help Codex take on longer-running work, even when laptops are closed, and help more organizations deploy agents securely in production. Managed OpenAI Agents soon?

译OpenAI 已达成协议收购云执行平台 Ona，其安全云执行技术将帮助 Codex 处理更长时任务（即使笔记本关闭也能继续运行），并让更多组织在生产环境中安全部署智能体。收购完成后，Ona 团队将加入 OpenAI 的 Codex 部门，此举或预示着托管 OpenAI 智能体服务的到来。

PixVerse@PixVerse_ · 1天前49

\🎬 PixVerse、ついに東京へ上陸！ 🗼／ 来週6/16(火)の夜、@PixVerse_ 初となる東京オフラインイベントを開催します！✨ AI動画の最前線で活躍するクリエイターや業界関係者が大集結🤝 @btrax_jp 様 、 @AICollectiveCo 様とともに、熱いクリエイティブの夜をお届けします🔥

Chubby♨️@kimmonismus · 1天前56

Iran threatens Elon Musk’s companies in Middle East via Iranian state media. Data Center and Starlink factory aimed for. Donald Trump recently wrote on TruthSocial that "The United States will be hitting Iran (whose Navy, Air Force, radar, anti-aircraft capabilities, and all other forms of defense, together with most of its offensive capability, are GONE!), VERY HARD TONIGHT." In response, Iran sent a clear message to the US, targeting Elon Musk's companies. Musk's largest asset in the region is the xAI data center in Saudi Arabia: a 500 MW AI data center built in partnership with HUMAIN, the Saudi state's AI vehicle, with NVIDIA as its primary chip supplier. This would make it the largest xAI data center outside the US - for comparison, the Colossus-1 data center in Memphis delivers around 300 MW. We are in for a very difficult night.

译伊朗通过官方媒体威胁马斯克在中东的公司，目标指向xAI在沙特阿拉伯的500 MW AI数据中心与Starlink工厂。该数据中心与沙特国家AI机构HUMAIN合作建设，NVIDIA为主要芯片供应商，是目前xAI在美国以外最大的数据中心（对比孟菲斯Colossus-1约300 MW）。此举是回应特朗普在TruthSocial发文称“美国今晚将狠狠打击伊朗”。局势紧张。

Chubby♨️@kimmonismus · 1天前75

Jeff Bezos raised $12B for Prometheus at a $41B valuation, seven months after launching it at $6.2B with no shipped product. The pitch is an "artificial general engineer" that compresses the design-to-build loop by 10x or more. The problem is that the physical economy can't be scraped. There's no internet of manufacturing data to train on, which is exactly why the reported $100B vehicle to buy up legacy industrial companies is interesting. You don't find that data. You acquire the factories that generate it. Could be an interesting moat.

译杰夫·贝佐斯旗下AI公司Prometheus在成立仅7个月、尚无任何产品交付的情况下，以410亿美元估值完成120亿美元融资（最初估值62亿美元）。该公司定位为“人工通用工程师”，目标是将设计到制造的循环压缩10倍以上。但物理经济无法像互联网数据那样抓取，缺乏制造业训练数据。为此，Prometheus计划斥资1000亿美元收购传统工业企业，通过获取工厂生成的数据构建护城河。

Google DeepMind@GoogleDeepMind · 1天前60

We’re teaming up @Palmeiras, the first football club to meaningfully build upon TacticAI: our AI system that can help simulate field scenarios and predict open play dynamics up to 8 seconds in advance. ⚽

译我们正与@Palmeiras合作，他们是第一家有意义地基于TacticAI构建的足球俱乐部：我们的AI系统可以模拟场上情景并提前最多8秒预测开放比赛动态。⚽

Luma@LumaLabsAI · 1天前35

Partnership is the new power move. Luma is at Cannes Lions 2026 with the partners building the next era of creative work. June 22–26, AI & Tech Sandbox and Luma House. Full schedule below. RSVP to join us → http://lumalabs.ai/cannes

译合作是新的力量之举。 Luma 将参加 2026 年戛纳创意节，与合作伙伴共同构建创意工作的新时代。6 月 22–26 日，AI & Tech Sandbox 和 Luma House。完整日程如下。 RSVP 参加 → http://lumalabs.ai/cannes

Runway@runwayml · 1天前65

Today, we’re deepening our partnership with Lionsgate with a slate of new initiatives, including a joint development program focused on creating original IP together. Learn more at the link below.

译今天，我们正在深化与狮门影业的合作，推出一系列新举措，包括一个专注于共同打造原创IP的联合开发项目。详情请见下方链接。

Anthropic@AnthropicAI · 1天前46

We’re launching Claude Corps, a national fellowship program matching people early in their careers with US nonprofits. We'll teach 1,000 people to use Claude, and pay them to use AI to advance their hosts’ missions. https://www.anthropic.com/claude-corps

译我们正在发起 Claude Corps，一个面向职业生涯早期人才与美国非营利组织配对的国家奖学金项目。 我们将教授 1,000 人使用 Claude，并支付他们用 AI 推进雇主使命的酬劳。 https://www.anthropic.com/claude-corps

StepFun@StepFun_ai · 1天前52

Great to see Step 3.7 Flash available on @ZenMuxAI! Fast, multimodal, and built for real-world workflows — now free to try on ZenMux for 1 month. Excited to make StepFun models more accessible to builders.

译阶跃星辰旗下多模态模型 Step 3.7 Flash 现已登陆 AI 平台 ZenMux，提供一个月免费试用。该模型专为快速迭代、编码、文档分析和多语言任务打造，面向开发者开放体验。

Chubby♨️@kimmonismus · 1天前55

Huge: OpenAI is considering drastically lowering the prices it charges users as it seeks to win customers from its rival Anthropic. The company is weighing significant cuts to what it charges for tokens, the unit of measurement AI firms use to bill for their products, according to people familiar with the matter. The move would be in anticipation of similar cuts the company expects at Anthropic, the people said.

译重磅：OpenAI正在考虑大幅降低用户收费，以从竞争对手Anthropic那里赢得客户。 据知情人士透露，该公司正在考虑大幅削减其token收费——token是AI公司用于计费的计量单位。知情人士称，此举是为了应对预计Anthropic也将做出类似降价。

Google DeepMind@GoogleDeepMind · 1天前62

When millions of AI agents interact with each other, new collective behaviors can emerge. 🌐 Together with @schmidtsciences, @coop_ai, @ARIA_research and supported by @GoogleOrg, we’re launching a $10M research fund to help understand how AI systems behave as a group. → https://goo.gle/3Si6rCl

译当数百万个AI智能体相互交互时，新的集体行为可能涌现。🌐 与 @schmidtsciences、@coop_ai、@ARIA_research 合作，并由 @GoogleOrg 支持，我们正在启动一个1000万美元的研究基金，以帮助理解AI系统在群体中的行为。→ https://goo.gle/3Si6rCl

Chubby♨️@kimmonismus · 1天前51

That was quick: Anthropic reversed a controversial policy that would have secretly degraded Claude Fable 5 for users doing frontier AI research after backlash from researchers who saw it as covert sabotage of competing AI development.

译真快：Anthropic 撤销了一项有争议的政策，该政策原本会在用户进行前沿 AI 研究时秘密降级 Claude Fable 5。此前研究人员强烈反对，认为这是对竞争 AI 开发的隐蔽破坏。

Alibaba Cloud@alibaba_cloud · 1天前42

🎙 Alibaba Cloud ClawTalks EP6 | Data + Agent = Your AI Workforce: Launch of Alibaba Cloud AI-native Database Service 📅 June 24, 2026 | 10:00 AM (UTC+8) | 30 min 🔗 Register now → https://int.alibabacloud.com/m/1000414360/ Your database shouldn't just store data—it should work for you. Introducing ApsaraDB Enterprise Agents: AI-native agents that live inside your database, think with context, and act autonomously. What you'll see in 30 minutes: ✅ Autonomous ops — analytics, governance, data prep, zero hand-holding ✅ Enterprise-grade security — granular access, data masking, token controls ✅ Self-improving — agents that learn and adapt on the job #AlibabaCloud #ClawTalks #ApsaraDB #AIAgents #DataIntelligence

译阿里云宣布推出ApsaraDB Enterprise Agents，即内置于数据库中的AI原生智能体，可自主执行分析、治理、数据准备等运维任务，无需人工干预；具备企业级安全能力（细粒度访问控制、数据脱敏、token管控），并能自我学习适应。相关活动将于2026年6月24日10:00（UTC+8）线上举办，时长30分钟。

Alibaba Cloud@alibaba_cloud · 1天前63

👏#ApsaraDB has 10 papers accepted to SIGMOD 2026—DB×AI, cloud-native storage & intelligent tooling. From paper to product: Beluga's CXL memory pool is in engineering validation; CloudJump III now powers #PolarDB's tiered storage. #AlibabaCloud keeps pushing the database frontier.🚀

译👏#ApsaraDB 有10篇论文被SIGMOD 2026接收——DB×AI、云原生存储与智能工具。从论文到产品：Beluga的CXL内存池正处于工程验证阶段；CloudJump III现已为#PolarDB的分层存储提供动力。 #AlibabaCloud 持续推动数据库前沿。🚀

🚨 AI News | TestingCatalog@testingcatalog · 1天前51

A new AI lab called River AI has been established in US by former folks from xAI. River AI will be focused on building “personal AI that is owned and shaped by you” Curious to test what’s coming 👀

译前xAI成员在美国成立新AI实验室River AI，致力于构建“由你拥有和塑造的个人AI”。River AI指出当前最优秀的AI被少数大公司控制，他们将打造全新的个人化AI堆栈，让AI完全为你工作、分享你的价值观、按你的方式运行。该实验室目前未透露具体产品细节，但已引发行业关注。

宝玉@dotey · 1天前67

OpenDoor 裁掉了其在印度的整个离岸团队（200多人），并正以美国本土规模更小的 AI 原生团队取而代之。 AI 未来可能也会影响外包行业的运作模式。

译OpenDoor 裁掉了其在印度的整个离岸团队（200多人），并正以美国本土规模更小的 AI 原生团队取而代之。 AI 未来可能也会影响外包行业的运作模式。

Alibaba Cloud@alibaba_cloud · 1天前26

Mexico | The Rhythm of Creation and Wisdom, Reimagined by AI From the cradle of ancestral wisdom and the time-honored mastery drawn from the earth’s bounty, to the vibrant rhythms of mariachi and the intricate woven patterns of Zapotec art—this video, created by Qwen and Wan, uses AI to reimagine time and culture. We journey from ancient sanctuaries to modern hubs of innovation, illustrating how AI bridges the wisdom of the past with the possibilities of the future, charting infinite new paths. Ready to bring your own ideas to life? Explore the power of our models on Model Studio and start creating today. 🔗 https://int.alibabacloud.com/m/1000414330/ #AlibabaCloud #Wan #Qwen #ModelStudio #Mexico #CreativeAI #AInnovation

译阿里云发布一段由Qwen和Wan模型生成的AI视频，以墨西哥文化为主题，融合祖先智慧、传统工艺、Mariachi音乐和Zapotec编织艺术，从古老圣地到现代创新枢纽，展现AI如何连接过去与未来。视频旨在展示模型的创意能力，并引导用户通过Model Studio平台使用相关模型进行创作。

Alibaba Cloud@alibaba_cloud · 1天前21

UAE | The Fusion of Heritage and Innovation, Reimagined by AI From the endless desert horizons and the flowing artistry of Arabic calligraphy to the timeless warmth of traditional hospitality and the glowing modern skyline of Dubai, this video—created by Qwen and Wan—uses AI to connect these iconic symbols. We trace the journey from ancient astronomical wisdom and cultural landmarks to the futuristic possibilities of tomorrow, showcasing how the UAE balances its rich traditions with boundless innovation. Ready to bring your own ideas to life? Explore the power of our models on Model Studio and start creating today. 🔗 https://int.alibabacloud.com/m/1000414322/ #AlibabaCloud #Wan #Qwen #ModelStudio #UAE #Dubai #TechForGood #CreativeAI #ArabicCulture #AInnovation

译阿里云发布AI生成视频，由通义千问（Qwen）和万相（Wan）模型共同创作。视频以阿联酋为主题，融合沙漠、阿拉伯书法、传统待客之道与迪拜现代天际线等元素，从古代天文智慧到未来科技，展现文化遗产与创新的平衡。官方推荐使用Model Studio平台调用模型进行创作。

AYi@AYi_AInotes · 1天前60

昨天Claude Fable 5发布以后， 压力给到了Open AI， OpenAI 正在考虑大幅降价， 以争取从劲敌 Anthropic 手中赢得更多用户， 感觉有点难追啊，全球大模型铁王座大概是Claude稳坐了

译昨天Claude Fable 5发布以后， 压力给到了Open AI， OpenAI 正在考虑大幅降价， 以争取从劲敌 Anthropic 手中赢得更多用户， 感觉有点难追啊，全球大模型铁王座大概是Claude稳坐了

Alibaba Cloud@alibaba_cloud · 1天前31

Day 2 from the Singapore 🇸🇬 Three things at #SuperAI2026 today: 👜 AI totebag printing at the Alibaba_Cloud x #Qwen booth 🛠️ Alibaba Cloud hands-on workshops 🎤 Catch Andy (Regional GM ASEAN) on the 11:50 AM main stage discussing "The Visual AI Stack." See you there!

译来自新加坡的第2天 🇸🇬 今天在#SuperAI2026的三件事： 👜 在阿里云 x #Qwen 展台打印AI帆布袋 🛠️ 阿里云动手实践工作坊 🎤 上午11:50在主舞台听Andy（东盟区域总经理）探讨“视觉AI堆栈”。 到时候见！

宝玉@dotey · 1天前57

DeepSeek 招 Agent Harness 研究员，能是世界范围内第一次招聘“Harness研究员” https://app.mokahr.com/su/mCyA8 ● 招聘岗位：Agent Harness 研究员（实习/全职） ● 工作性质：全职 | 实习 ● 工作地点：浙江·杭州市 / 北京市 ================================ 【团队使命】 Model + Harness = Agent 我们正在把 DeepSeek 的前沿模型能力，转化为领先的 Agent 产品。这其中除模型本身以外的所有工作，都属于 Harness 的范畴。 你将加入 Harness 团队，与研究员、工程师、产品经理紧密协作，探索 Harness 领域的研究前沿，定义 DeepSeek 对 Harness 的理解。 ================================ 【主要职责】 - 前沿创新：与 Harness 团队的研究员与工程师深度沟通、紧密合作，共同定义和实现 Harness 领域基于模型能力的前沿创新，包括但不限于上下文管理、长期记忆、Subagent 与 Multi-Agent、自进化 Agent 等领域。 - 深度适配：与模型训练团队的研究员与工程师深度沟通与合作，实现模型与 Harness 的共同进化，从 Harness 的角度实现 DeepSeek 的 Harness 与模型的深度适配。 - 评测构建：提出 Harness 领域的基准测试与评测方法，构建评测基准数据和制定数据标注策略，从 Harness 的角度研究并优化 Agent 在各领域的智能水平。 - 真实迭代：以真实世界的任务作为 Harness 研究的重要反馈源，设计相关数据与实验，持续迭代 Agent 能力在真实使用场景下的表现。 - 用户优化：基于团队收集到的用户反馈，从 Harness 的角度研究并优化为最广大用户解决真实场景问题的能力。 ================================ 【任职要求】 - 科研背景：硕士学历及以上，2年以上计算机科学或相关领域的科研经验，水平过硬，眼界广阔，有科研品味；在计算机科学领域具有含金量的论文发表（特别优秀候选人可放宽学历及年限要求）。 - 独立推进：面对问题能够独立分析并提出自己的 idea，具备从 0 到 1 推动研究的能力；能够快速将想法转化为可运行的原型，具备高效的实验迭代能力。 - 全栈开发：熟练使用 AI Agent 工具进行软件开发，在软件开发领域具有极强的学习能力；能够在 AI 辅助下，在没有直接经验的领域（如语言、技术、框架等）进行研究目的的编程工作。 - 重度用户：是 Agent 产品的高强度用户，对 Agent Harness 的开发和研究有极大的热情，对模型行为有品味有判断力；深度使用过代码类及通用类 Agent 产品，并将相关产品的使用融入到自己的工作和生活中。 - 知识储备：熟悉 LLM 以及 Agent 基本机制及其技术原理（包括 LLM API、KV Cache、Agent Loop、Tool Use、Reasoning、Planning、Skills、MCP、Memory、Subagent、Multi-Agent 等）；对 Prompt Engineering、Context Engineering、Harness Engineering 等课题有深入的理解。 - 沟通能力：具备良好的中文沟通能力。 ================================ 【加分项】 - AI 领域相关科研经验，或 AI 行业的研究员任职经验。 - 拥有个人开源作品，或对开源社区有深度贡献。 - 在计算机科学领域、编程领域、或数据科学领域获得过具有含金量的比赛奖项。 - 其它超乎常人的与此工作相关的才能。

译DeepSeek 开放 Agent Harness 研究员岗位（实习/全职），工作地点杭州/北京。团队使命为 “Model + Harness = Agent”，旨在将模型能力转化为 Agent 产品。职责涵盖：上下文管理、长期记忆、Subagent 与 Multi-Agent、自进化 Agent 等前沿创新；实现模型与 Harness 的深度适配；构建 Harness 领域评测基准；基于真实任务与用户反馈迭代 Agent 能力。要求硕士以上学历、2年科研经验、全栈开发能力、Agent 产品重度用户，熟悉 LLM 及 Agent 机制（KV Cache、Tool Use、Reasoning、MCP 等）。

MiniMax (official)@MiniMax_AI · 1天前50

Tired of your token costs? M3 is free on @PBDTokenRouter through june 17th With every other MiniMax model 50% off 🔥 That means building with 1M context, and $0 in / $0 out Build something today, before this opportunity ends.

译厌倦了高昂的 token 成本？ M3 在 @PBDTokenRouter 上免费至 6 月 17 日 其他所有 MiniMax 模型五折优惠🔥 这意味着以 1M 上下文窗口构建，输入/输出费用均为 $0 趁机会结束前，今天就动手搭建点什么吧。

Greg Brockman@gdb · 1天前61

Use your Oracle cloud commitment for OpenAI products: https://openai.com/index/openai-on-oracle-cloud/

译将你的Oracle云承诺用于OpenAI产品：https://openai.com/index/openai-on-oracle-cloud/

郭明錤｜Ming-Chi Kuo@mingchikuo · 2天前65

關於台積電的次世代先進封裝 CoPoS 的幾個關鍵（省略可查詢到的技術細節）： 1. 預計 2H28 量產，目標提升 9.5 倍光罩尺寸以上的封裝之量產經濟性，Nvidia 的 AI 晶片 Feynman 可能將首度採用。 2. 根據產業調查，兩個不同的地方會用到玻璃（尺寸 mm）： → 310 x 310 的臨時玻璃載具（glass carrier） → 250 x 250（測試）/ 510 x 515（量產）的玻璃面板，加工後切割為玻璃核心載板（glass core substrate） 3. 玻璃核心載板的架構主要分成三層：玻璃作為核心層，上下以 ABF（ABF-GCP）增層包覆。玻璃加工的挑戰，像是TGV（through glass via）、填銅 / 金屬化（metallization）等，指的都是這個階段。 4. CoPoS 常見的錯誤論述： → ❌ 錯誤 1：採用玻璃中介層（interposer）。⭕️ 應修正為：玻璃非中介層，其互連角色由晶片側 RDL 與玻璃核心載板側 TGV / ABF 增層分別承接。 → ❌ 錯誤 2：玻璃取代 ABF。⭕️ 應修正為：如前述的玻璃核心載板架構，玻璃與 ABF 並存。 → ❌ 錯誤 3：晶片放在玻璃上。⭕️ 應修正為：晶片貼附於玻璃核心載板的 ABF 增層表面。 5. CoPoS 將延續並強化台積電先進封裝的優勢，預期讓該優勢能見度可達約2032年。

译台积电CoPoS预计2028年下半年量产，目标提升9.5倍光罩尺寸以上封装的经济性，Nvidia AI芯片Feynman或首度采用。架构采用玻璃核心载板：玻璃为核心层，上下以ABF（ABF-GCP）增层包覆。临时玻璃载具尺寸310×310 mm，玻璃面板测试阶段250×250 mm、量产阶段510×515 mm。澄清常见误解：玻璃非中介层，互连由芯片侧RDL与玻璃核心载板侧TGV/ABF增层分别承接；玻璃与ABF并存而非取代；芯片贴附于ABF增层表面。CoPoS将延续台积电先进封装优势，能见度可达约2032年。

郭明錤｜Ming-Chi Kuo@mingchikuo · 2天前58

Key takeaways on TSMC's next-generation advanced packaging, CoPoS (publicly available technical details omitted): 1. CoPoS is currently expected to enter mass production in 2H28. It is designed to improve the economics of ultra-large packages above the 9.5x reticle-size class, with NVIDIA’s Feynman AI chip a potential first adopter. 2. According to industry checks, glass is used in two distinct places (dimensions in mm): → 310 x 310 temporary glass carriers → 250 x 250 (pilot) / 510 x 515 (mass production) glass panels, processed and later cut into individual glass core substrates 3. The glass core substrate is essentially a three-layer structure: a glass core sandwiched between ABF (ABF-GCP) build-up layers on both sides. The widely discussed glass processing challenges, such as TGV formation and copper filling / metallization, are tied to this part of the stack. 4. Common misconceptions about CoPoS: → ❌ Misconception 1: CoPoS uses a glass interposer. ⭕️ Correction: The glass is not an interposer. The interconnect role is instead handled by the chip-side RDL, plus the TGV/Cu interconnects and ABF build-up layers in the glass-core substrate stack. → ❌ Misconception 2: Glass replaces ABF. ⭕️ Correction: As the substrate architecture above shows, glass and ABF coexist. → ❌ Misconception 3: Chips sit directly on glass. ⭕️ Correction: Chips are attached to the ABF build-up surface of the glass core substrate. 5. CoPoS should extend and reinforce TSMC’s leadership in advanced packaging, potentially giving that advantage visibility through around 2032.

译郭明錤分析，台积电CoPoS预计2028下半年量产，面向9.5倍光罩尺寸以上超大封装，NVIDIA Feynman AI芯片或率先采用。玻璃用于两个位置：310×310mm临时载板，以及250×250mm（试产）/510×515mm（量产）玻璃面板加工成玻璃核心基板。该基板为三层结构——玻璃芯两侧叠加ABF增层，TGV成孔与铜填充等挑战集中于此处。澄清常见误解：玻璃非中介层，互连由RDL、TGV/Cu及ABF共同承担；玻璃与ABF共存而非替代；芯片贴装在ABF增层表面。CoPoS有望延续台积电先进封装领先优势至2032年左右。

Rohan Paul@rohanpaul_ai · 2天前44

Anthropic’s Dario Amodei has only 1 direct report, his chief of staff. The rest of Anthropic’s executive system flows through Dario’s sister, Anthropic President Daniela Amodei, who handles daily operations and reports to the board. For some comparison, OpenAI CEO Sam Altman has around half a dozen direct reports, while Nvidia Corp. CEO Jensen Huang has 60 people reporting to him. --- From "Bloomberg Originals" YouTube channel, (link in comment)

译Anthropic 的 Dario Amodei 只有 1 名直接下属，即他的幕僚长。 其余高管系统通过 Dario 的妹妹、Anthropic 总裁 Daniela Amodei 运作，她负责日常运营并向董事会汇报。 作为对比，OpenAI CEO Sam Altman 约有半打直接下属，而 Nvidia Corp. CEO Jensen Huang 有 60 人向他汇报。 --- 来自“Bloomberg Originals”YouTube 频道（链接见评论）

Tibo@thsottiaux · 2天前65

Welcome Clint and Michael! Incredibly excited to see what we do together to contribute to the cybersecurity field and accelerate defenders across the globe. It's time to build.

译Clint Gibler和Michael Aiello加入OpenAI领导网络安全。Clint此前在Semgrep打造了全球最流行的开源安全代码扫描工具。他提出未来方向：通过安全代码生成和简化检测—验证—修复流程，系统性消除漏洞类别（韧性设计）；构建模型和工具为防御者提供“超能力”；保护开源软件，已投入数百万美元修复浏览器、操作系统等核心库漏洞；与社区和合作伙伴共同守护关键基础设施。具体包括大规模漏洞发现修复、广泛提供顶级模型、创建安全技能与剧本、构建防御者编排平台等。

Rohan Paul@rohanpaul_ai · 2天前55

Sam Altman is reportedly warning staff that recursive self-improvement (RSI) could delay its IPO. Altman said a major RSI breakthrough could justify delaying the listing, as some work may be easier while being a private company. Public-market pressure for Revenue/profit could be huge during a phase where model capability may change faster than quarterly reporting can explain. --- cryptobriefing. com/openai-ipo-delay-recursive-self-improvement/

译Sam Altman 据报告正警告员工，递归自我改进（RSI）可能会延迟其 IPO。 Altman 表示，一次重大的 RSI 突破可能合理推迟上市，因为一些工作在作为私营公司时可能更容易进行。 在模型能力变化可能比季度报告解释更快的阶段，公开市场对收入/利润的压力可能巨大。