看着智谱和 minimax 翻倍，心情很复杂 – V2EX 首页 注册 登录 V2EX = way to explore V2EX 是一个关于分享和探索的地方 现在注册 已注册用户请 登录 V2EX › 投资 看着智谱和 minimax 翻倍，心情很复杂 ggp1ot2 · 17 小时 10 分钟前 · 2390 次点击 最近半年在富途牛牛玩港股。 两个大模型的股票 智谱（ 2513 ） 和 minimax （ 0100 ）在最近一周内，都翻倍了 其中 [智谱] 由 2.9 号开盘的 200 元，到今天收盘 540 元，翻了一倍多 [ Minimax ] 由 2.9 号开盘的 400 元，到今天收盘 840 元，也翻了一倍多 注意 智谱 1.8 号上市，招股价 116 元 minimax 1.9 号上市，招股价 160 元 而我呢，因为是大模型算法相关工作，也就一直关注着这俩股票，关注的原因，是因为我工作真的会用到 GLM 模型 虽然会写代码，奈何不懂投资，就一直看着他俩涨，很难受，因为自己没敢买 并且到今天，minimax 再次涨了 30%，我依旧觉得，他不值这几千亿的市值。 但这是我第一次，看到两只股票涨的这么疯狂，并且我曾经在 minimax 股价 360 的时候，买入 100 股，10 分钟就变成 380 ，立刻卖掉赚了 2000 元，然后就开始希望他跌 现在就会想，如果没有卖掉，36000 已经变成 86000 了。。。 为啥这俩股票，就能一点不跌，每天 20%-30%的上涨呢？ 大家就别喷我了，我是一个新鲜韭菜。 最近还挺影响心情的，知道这是自己赚不到的钱，因为自己不敢买，也没勇气输，但是就是看着他天天涨，每天就会很关注。 第 1 条附言  · 16 小时 35 分钟前 只是除夕没事吐槽一下 我就是韭菜，不仅赚不到钱，没有能力赚到钱，还因为赚不到钱不开心😂 再补充一下，本韭菜的操作 – 滴普科技，中签一手 26 元，36 元卖了，现在股价 100 – 海致科技（就是前两天的），中签一手，27 元，暗盘 80 卖了，现在 120 元 – 紫金黄金，中签 80 ，暗盘 100 卖了，现在 300 好像 哎，这么看，本韭菜运气不错嘛。。。 就是拿不住，太穷了，有点利润就想跑，做不到卸载不看。 不对，我还买了 10 万美团，100 买的，这次倒是熬住了，但是现在价格 80 。。。 股票 涨幅 投资 39 条回复 • 2026-02-17 14:14:36 +08:00 1 kulove 17 小时 9 分钟前 via Android 感觉 MiniMax 前景可以的 2 ggp1ot2 OP 17 小时 6 分钟前 @ kulove 前景是可以，但是他现在市值 3000 亿，作为对比，京东（ 09618 ）的市值也是 3000 亿，真的对得起这个价格吗？ 3 kulove 17 小时 5 分钟前 via Android @ ggp1ot2 现在 AI 相关的都很虚的 不过我好久不炒股了 4 ggp1ot2 OP 17 小时 4 分钟前 @ kulove #3 是的，一个月从 100 涨到今天快 900 ，太疯狂了，第一次炒股，被惊到了 5 HFX3389 17 小时 0 分钟前 > 我依旧觉得，他不值这几千亿的市值。 但是市场参与者认为它值，所以它的股价就这么高 6 HFX3389 16 小时 59 分钟前 @ HFX3389 #5 说不定 deepseek v4 出了后他们俩的股价能更高呢 7 ggp1ot2 OP 16 小时 57 分钟前 @ HFX3389 #5 是的，本韭菜已经认识到了，不是他值不值，而是市场觉得值不值，但是天天 20cm ，太吓人了 8 foryou2023 16 小时 55 分钟前 这难受啥，你赚不到钱的钱难受啥，所有没有赚到手里的钱，都是能力不足，到手的鸭子飞了，那也是差点运气，运气也是实力的一部分，还是差点实力 9 renchong 16 小时 54 分钟前 是不是想如果还有下一个类似的 ai 模型股票，立即购入 10 ggp1ot2 OP 16 小时 34 分钟前 @ renchong 哈哈哈，我觉得再来一次，还是不敢买，还是赚不到钱 11 ggp1ot2 OP 16 小时 33 分钟前 @ foryou2023 怎么说呢，就是那种，你带穷人看了很奢华的东西，穷人回去就难受，即使他本来也没有这些。 这几个股票，每天就在我的自选里面，最近一周，每天都是 20%-30%的涨 其实我感觉也不是难受，就是心情很复杂 就是那种，可能换一个人，就赚很多了， 偏偏自己不行。 可能难受的就是自己没有梭哈的勇气吧 12 slowman 16 小时 29 分钟前 牛顿当年也是这样的 13 ggp1ot2 OP 16 小时 28 分钟前 @ slowman 哈哈哈非常有被安慰到 14 defaw 16 小时 26 分钟前 都没进港股通的，你真要赚大钱要巨额换汇才行，这本身就是门槛。 15 ggp1ot2 OP 16 小时 24 分钟前 @ defaw 富途牛牛放了 100w hkd 。主要用来打新 16 8byte 16 小时 15 分钟前 via iPhone 不懂不做，你拿不住是因为不懂，没什么遗憾的 17 sirenecyrus 16 小时 6 分钟前 AI 相關記憶體/NAND 升了半年 金銀也升了半年 接下來大把大把的翻倍股可能出現…. 有能力又懂的你自然會買到 不用擔心 18 ggp1ot2 OP 16 小时 4 分钟前 @ sirenecyrus 我在 8 元购买了 NVTS ，我觉得可以翻倍😂 19 mythabc 15 小时 57 分钟前 打开商汤科技股票看看历史趋势呗。 20 buru 15 小时 53 分钟前 新股票没有套牢盘， 很容易炒起来 21 ggp1ot2 OP 15 小时 42 分钟前 @ mythabc 擦，三块涨到 10 块 22 zerovoid 11 小时 58 分钟前 人赚不到认知外的钱，碰运气赚到了，迟早也得还回去，摆平心态最重要。 23 yifangtongxing28 11 小时 53 分钟前 建议你了解一下报税，趋势投机的机会 之后投资美股港股的人，税是个大问题 24 ffgrinder 11 小时 49 分钟前 请问你为什么觉得相比于几个大模型厂的估值，2000 亿港币算贵的？ 以及京东作为一个低利润率的平台经济，为什么你用他来评估 AI 的市值？ 感觉可以先从讨论“应该怎么估值”入手。 25 dfdd1811 11 小时 43 分钟前 我看了闫俊杰跟罗永浩的谈话之后就想买，可没有渠道… 26 qzydustin 9 小时 8 分钟前 不看好可以买看跌期权，或者卖看涨期权 27 tigerstudent 5 小时 59 分钟前 via Android 日经贴 28 EndlessMemory 4 小时 49 分钟前 就这么喜欢赚快钱吗 29 iorilu 3 小时 39 分钟前 港股暴涨暴跌是常事 别光看涨时候眼红 你要了解下跌的空间 我说实话, 我玩港股也有几年, 反几倍后跌 90%多的是, 没啥稀奇的 我从不

Source: V2EX | Original Link

[求助] 国内 Java 10 年，总包 几十个 个，正在日本旅游，极度纠结是否润日，求在日大佬指点 – V2EX 首页 注册 登录 V2EX = way to explore V2EX 是一个关于分享和探索的地方 现在注册 已注册用户请 登录 V2EX › 东京 [求助] 国内 Java 10 年，总包 几十个 个，正在日本旅游，极度纠结是否润日，求在日大佬指点 dlam007 · 15 小时 51 分钟前 via Android · 1251 次点击 真心求教，焦虑的不行。如果有机会想在大阪线下见面。或者大家愿意聊聊，欢迎加微信 dlam007 。 各位 V 友大家好，目前本人（ 30+，Java 10 年经验，英语 PTE 已过）正在京都/大阪一带旅游。 这几天实地走访，对日本的生活环境非常心动，但回到现实的职业规划，陷入了巨大的矛盾，想听听在日 IT 前辈们的真实建议。 先说下我的现状： 国内： 目前在某一线城市，月入 40k+，算上年终奖总包在 几十 w RMB 之间浮动。虽然收入还可以，但由于大家都懂的原因，对 35 岁后的职业稳定性极度焦虑，感觉国内是“高杠杆、高风险”的人生模式。 日本： 向往这边的职场稳定性（较少裁员、无 35 岁歧视）和生活质量，但目前调研到的 IT 平均薪资似乎在 600w-800w 日元左右（约 30w-40w RMB ），相比国内几乎是腰斩。 我的核心矛盾点： 薪资天花板： 以我 10 年 Java 的经验加英语能力，如果瞄准外资（如 Google, AWS ）或日本本土大厂（如 Mercari, Rakuten, PayPay ），拿到 1000w-1200w JPY 的概率有多大？ 如果只能拿 800w ，心理落差确实很难平衡。 生活成本与体感： 大家都说日本税高，1000w 日元到手后的实际购买力，比起国内月入 4w ，体感下降明显吗？（本人对奢侈消费无感，更看重居住环境和食品安全）。 长期价值： 这种“降薪换稳定”的买卖，在 40 岁以后回看，真的值吗？ 想请教在日的大佬们：有没有背景类似（国内大厂高薪去日本）的前辈，分享一下你们的心路历程？在现在的日本市场，10 年老兵如果想找一份平衡收入与生活的 Job ，哪些公司或渠道比较靠谱？ 如果这几天有在大阪，京都，东京的朋友愿意线下交流，我愿请喝咖啡/吃顿和牛，当面讨教。感谢各位，祝大家不卷不燥。 日本 职业 薪资 39 条回复 • 2026-02-17 13:27:12 +08:00 1 charlesmware877 15 小时 46 分钟前 日元一直在贬值，未来还会继续贬值下去，而且看起来你也没有日语基础，你现在这个条件绝对不建议润日 2 EastLord 15 小时 33 分钟前 via iPhone 1 「几天实地走访，对日本的生活环境非常心动」 展开说说，哪里心动 3 likooo125802023 15 小时 21 分钟前 你日语怎么样？ 熟练跟日本人对话不 不是说考个 N1 就完事的 不能的话洗洗睡吧 4 dlam007 OP 14 小时 59 分钟前 via Android @ EastLord 干净，安静，我对国内吵闹实在是神经衰弱 5 dlam007 OP 14 小时 58 分钟前 via Android @ likooo125802023 日语完全不行，但是雅思 4 个 6 ，口语还能好点 6 dlam007 OP 14 小时 57 分钟前 via Android @ charlesmware877 那，有什么建议吗 7 tilv37 13 小时 55 分钟前 码驻住，等后续。我估计比 lz 大几岁，和你的想法一模一样。不过我会日语，有证，但是口语有点费劲。国内外资工作。持续摇摆中。 8 kumiko 12 小时 25 分钟前 via Android 也就你们不会日语也没在日企工作过的人才会考虑润日了 9 wellsc 12 小时 19 分钟前 雅思四个六为什么不去试试澳新欧洲 10 zerovoid 11 小时 45 分钟前 阁下居住环境这么恶劣吗，年收入几十万，可以住高档小区了吧。 11 speedmancs 11 小时 25 分钟前 以你的薪资，去日本完全不值得 日本大厂像谷歌这种很难进，不是说你有 10 年 java 经验就能进的，bar 很高，即便如此薪资也不太会让你满意。 12 speedmancs 11 小时 24 分钟前 @ wellsc 四个六没法去澳洲，去澳洲雅思要求很高。 13 Anarchy 10 小时 50 分钟前 via Android 你的状况只能找日本的外企，直接海外投简历就好了，根据结果你就有答案了。 找普通日本公司的话，你至少要有承受住 400w 不崩溃的心理承受能力或者有足够自信的日语能力去拿 800w （非要量化的话等级在 c1 ，等价 n1 的 140+）。 14 forisra 10 小时 4 分钟前 除非是在日外企，要不然就日本的那些编程技术水平和工作流程绝对会把你逼疯的。 我觉得程序员润日的最合适的还就是高中毕业，大专毕业的，耐心慢慢走流程，去了日本工资不高但同样要求也不高。什么世界最好语言 PHP ，VBA 编程会一点就够了。 当然前提是你要在日本人的公司啊，华人公司约等于没润。 15 vkillwucy 7 小时 48 分钟前 via Android 这几年国际环境也要考虑下 16 timelessg 7 小时 16 分钟前 via Android 总的来说高市上台之后不是很建议，但有强烈意愿也可以试试，日本现在整体排外情绪明显，就业市场也很明显，it 相对好一些，当然也只是相对其他行业 17 Moishine 6 小时 55 分钟前 via Android @ zerovoid 一线城市几十万住什么高档小区 18 WillemNZ 6 小时 51 分钟前 降薪换来的可不只是稳定，还有更高的生活质量。如果最后决定要润就早点行动起来，千万不要拖到 35+在国内遇到职业危机再被迫行动，那个时候你会发现润这条路已经被自己的年龄卡死了。 19 Dragonish3600 6 小时 14 分钟前 朋友从国内腾讯辞职，去了日本任天堂，薪水应该是少了一半。但是她说目前除了工资其他都比较满意，主要是职场环境 生活环境好了很多。但是薪水确实低了不太爽，所以也在看跳槽机会。 20 cgpiao 6 小时 12 分钟前 日本也裁员，你这个收入，该做的是早做分散投资，财务自由，而不是来日本。 哪怕真来了日本，估计退休后还得回国，因为才交 20 多年的日本的养老金，你根本生存不了。 21 gpt5 5 小时 45 分钟前 1 不要纠结工资，先去了再说，光上网不用翻墙这一点，就约等于额外的 1000 万日元/月。 22 000sitereg 5 小时 39 分钟前 via Android 旅游感受到的和实际生活不一样 23 k9982874 4 小时 55 分钟前 via Android 不想吃苦，还不会当地语言就想移民，怎么想的 24 chihiro2014 4 小时 51 分钟前 要是 10 年 java 能进，那么国内一群人去 aws 和

Source: V2EX | Original Link

[2602.12670] SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks Computer Science > Artificial Intelligence arXiv:2602.12670 (cs) [Submitted on 13 Feb 2026] Title: SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks Authors: Xiangyi Li , Wenbo Chen , Yimin Liu , Shenghan Zheng , Xiaokun Chen , Yifeng He , Yubo Li , Bingran You , Haotian Shen , Jiankai Sun , Shuyi Wang , Qunhong Zeng , Di Wang , Xuandong Zhao , Yuanli Wang , Roey Ben Chaim , Zonglin Di , Yipeng Gao , Junwei He , Yizhuo He , Liqiang Jing , Luyang Kong , Xin Lan , Jiachen Li , Songlin Li , Yijiang Li , Yueqian Lin , Xinyi Liu , Xuanqing Liu , Haoran Lyu , Ze Ma , Bowei Wang , Runhui Wang , Tianyu Wang , Wengao Ye , Yue Zhang , Hanwen Xing , Yiqi Xue , Steven Dillmann , Han-chung Lee View a PDF of the paper titled SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks, by Xiangyi Li and 39 other authors View PDF HTML (experimental) Abstract: Agent Skills are structured packages of procedural knowledge that augment LLM agents at inference time. Despite rapid adoption, there is no standard way to measure whether they actually help. We present SkillsBench, a benchmark of 86 tasks across 11 domains paired with curated Skills and deterministic verifiers. Each task is evaluated under three conditions: no Skills, curated Skills, and self-generated Skills. We test 7 agent-model configurations over 7,308 trajectories. Curated Skills raise average pass rate by 16.2 percentage points(pp), but effects vary widely by domain (+4.5pp for Software Engineering to +51.9pp for Healthcare) and 16 of 84 tasks show negative deltas. Self-generated Skills provide no benefit on average, showing that models cannot reliably author the procedural knowledge they benefit from consuming. Focused Skills with 2–3 modules outperform comprehensive documentation, and smaller models with Skills can match larger models without them. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2602.12670 [cs.AI] (or arXiv:2602.12670v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2602.12670 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Xiangyi Li [ view email ] [v1] Fri, 13 Feb 2026 07:06:06 UTC (1,366 KB) Full-text links: Access Paper: View a PDF of the paper titled SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks, by Xiangyi Li and 39 other authors View PDF HTML (experimental) TeX Source view license Current browse context: cs.AI < prev | next > new | recent | 2026-02 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation × loading… Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps

Source: Hacker News | Original Link

Why Affordability and the Vibecession Are Real Economic Problems Mike Konczal Subscribe Sign in Why Affordability and the Vibecession Are Real Economic Problems There are many ways inflation makes people worse off even when real incomes recover, especially for essentials. Mike Konczal Feb 09, 2026 59 13 12 Share Affordability is a major concern among voters in practice. But is it a major concern in theory? This Substack sets up the economic fights for the year, subscribe to get ahead of them. Subscribe You’ve probably seen some version of Graphic 1 below, which shows real median household income from Census. After falling during the post-pandemic inflation surge, real incomes recovered. By 2024–2025, they had ended up higher than in 2019. This result holds across multiple ways of looking at this data, including hourly wage data. Yet consumer sentiment is stuck near historic lows, about as pessimistic as during the financial crisis and the depths of the Great Recession. These two facts coexist. And the politics of “affordability” has rushed into the gap between them, as politicians and advocates try to speak to people’s persistent anxiety about prices and living costs. But there has also been a subtle, yet real, pushback against this focus. That pushback usually starts from the observation that today’s affordability debate is inseparable from the 2021–2024 inflation episode. From there, two critiques follow. The first is a money-illusion story: people fail to recognize that their incomes rose alongside prices, so their distress reflects confusion rather than material harm. This argument often emphasizes that incomes at the bottom of the distribution rose faster than those at the top, producing a durable wage compression. And yet polling consistently finds that lower-income households report more dissatisfaction with inflation, not less. The second critique follows naturally. If what people really want is their old price level back, that is simply not something policymakers can deliver. Broad-based price declines tend to occur only in deep recessions, and even then only modestly. So it is said to be dangerous politics and bad economics to make affordability central. Doing so risks promising something impossible, or worse, flirting with economic collapse as a policy goal. Matthew Yglesias has made a version of this argument , affordability is “just high nominal wages” and “basically just anger at inflation,” as have others. This conversation echoes what you often hear about the so-called “vibecession.” The term was originally coined by the writer Kyla Scanlon in 2022 in a nuanced way, to describe self-reinforcing pessimism. Today, it’s more often weaponized to imply that consumer sentiment is untethered from material reality, a reflection of the circulating bad vibes. One could correctly say that these affordability problems existed in 2019 and are independent of the inflation wave. But we should take the bait. Over the past five years I’ve been hau

Source: Hacker News | Original Link

消息称英国监管机构考虑调整规则以吸引中国企业上市-36氪 账号设置 我的关注 我的收藏 申请的报道 退出登录 登录 搜索 36氪Auto 数字时氪 未来消费 智能涌现 未来城市 启动Power on 36氪出海 36氪研究院 潮生TIDE 36氪企服点评 36氪财经 职场bonus 36碳 后浪研究所 暗涌Waves 硬氪 氪睿研究院 媒体品牌 企业号 企服点评 36Kr研究院 36Kr创新咨询 企业服务 核心服务 城市之窗 政府服务 创投发布 LP源计划 VClub VClub投资机构库 投资机构职位推介 投资人认证 投资人服务 寻求报道 36氪Pro 创投氪堂 企业入驻 创业者服务 创投平台 AI测评网 首页 快讯 资讯 推荐 财经 AI 自助报道 四川 最新 创投 汽车 科技 专精特新 直播 视频 专题 活动 搜索 寻求报道 我要入驻 城市合作 02 月 17 消息称英国监管机构考虑调整规则以吸引中国企业上市 2026-02-17 11:11 分享至 打开微信“扫一扫”，打开网页后点击屏幕右上角分享按钮 英国审计监管机构正在考虑调整会计规则，以鼓励中国公司赴伦敦上市。Financial Reporting Council周一表示，将就是否允许在中国注册的公司在伦敦发行全球存托凭证（GDR）时遵循中国的会计规则征询意见。全球存托凭证（GDR）是由银行发行、代表外国公司股票的凭证。根据该拟议的规则调整，中国公司在发行GDR时，将被允许在限定的一段时间内遵循中国的审计准则。根据现行规定，在伦敦发行GDR的中国实体需要遵循英国的会计准则。（新浪财经） 原文链接 下一篇 2026春节档总票房破8亿 36氪获悉，据猫眼专业版数据，2026年春节档档期总票房破8亿。 1小时前 24小时热榜 查看更多榜单 帮长辈清手机成过年“基操”，这份2026年最新攻略请收好 红包大战变迁史：从社交到AI 过年吃“金拱门年夜饭”大逆不道？但多少打工人除夕想冲进麦当劳急头白脸搓一顿 22小时前 年轻人春节回家账单：花销超2万，机票占一半 20小时前 如何通过三个问题识破谎言 3小时前 在AI和机器人里找节目：马年春晚背后的科技大战揭秘 2小时前 揭秘春晚“两个蔡明”背后，这家机器人公司的百日奋战 1小时前 这个春节999元租个人形机器人拜年，然后呢？ 1小时前 关于36氪 城市合作 寻求报道 我要入驻 投资者关系 商务合作 关于我们 联系我们 加入我们 36氪欧洲站 36氪欧洲站 36氪欧洲站 Ai产品日报 网络谣言信息举报入口 热门推荐 热门资讯 热门产品 文章标签 快讯标签 合作伙伴 36氪APP下载 iOS & Android 本站由 阿里云 提供计算与安全服务 违法和不良信息、未成年人保护举报电话：010-89650707 举报邮箱：[email protected] 网上有害信息举报 © 2011~ 2026 北京多氪信息科技有限公司 | 京ICP备12031756号-6 | 京ICP证150143号 | 京公网安备11010502057322号 意见反馈 36氪APP 让一部分人先看到未来 36氪 鲸准 氪空间 推送和解读前沿、有料的科技创投资讯 一级市场金融信息和系统服务提供商 聚焦全球优秀创业者，项目融资率接近97%，领跑行业

Source: 36Kr | Original Link

千问：1.3亿人春节首次体验AI购物-36氪 账号设置 我的关注 我的收藏 申请的报道 退出登录 登录 搜索 36氪Auto 数字时氪 未来消费 智能涌现 未来城市 启动Power on 36氪出海 36氪研究院 潮生TIDE 36氪企服点评 36氪财经 职场bonus 36碳 后浪研究所 暗涌Waves 硬氪 氪睿研究院 媒体品牌 企业号 企服点评 36Kr研究院 36Kr创新咨询 企业服务 核心服务 城市之窗 政府服务 创投发布 LP源计划 VClub VClub投资机构库 投资机构职位推介 投资人认证 投资人服务 寻求报道 36氪Pro 创投氪堂 企业入驻 创业者服务 创投平台 AI测评网 首页 快讯 资讯 推荐 财经 AI 自助报道 四川 最新 创投 汽车 科技 专精特新 直播 视频 专题 活动 搜索 寻求报道 我要入驻 城市合作 02 月 17 千问：1.3亿人春节首次体验AI购物 2026-02-17 11:24 分享至 打开微信“扫一扫”，打开网页后点击屏幕右上角分享按钮 2月17日大年初一凌晨，千问发布的数据显示，春节活动期间，全国超过1.3亿人第一次体验AI购物，说了50亿次“千问帮我”。过去两天，用户用AI买门票的订单环比增长22倍，AI订机票等交通票务单量增长超7倍。随着春节档电影上线，用户用AI买电影票的订单环比增长372倍，其中来自三四线城市的AI订单量更是暴涨782倍。所有AI订单中，近一半订单来自县城，近400万60岁以上用户体验了AI购物。 下一篇 宇树科技王兴兴：今年宇树科技的目标出货量在1-2万台左右 宇树科技创始人王兴兴表示，此次参演的G1与H2人形机器人，其核心亮点在于全自主集群控制技术的首次亮相，实现全球第一次集群快速跑位，最快任意跑位速度可达4m/s。王兴兴预计，今年全世界人形机器人的出货量至少达到几万台，宇树科技的目标出货量在1-2万台左右。但他也指出，受限于具身智能大脑等技术瓶颈，人形机器人还处于应用早期。（财联社） 48分钟前 24小时热榜 查看更多榜单 帮长辈清手机成过年“基操”，这份2026年最新攻略请收好 红包大战变迁史：从社交到AI 过年吃“金拱门年夜饭”大逆不道？但多少打工人除夕想冲进麦当劳急头白脸搓一顿 22小时前 年轻人春节回家账单：花销超2万，机票占一半 20小时前 如何通过三个问题识破谎言 3小时前 在AI和机器人里找节目：马年春晚背后的科技大战揭秘 2小时前 揭秘春晚“两个蔡明”背后，这家机器人公司的百日奋战 1小时前 这个春节999元租个人形机器人拜年，然后呢？ 1小时前 关于36氪 城市合作 寻求报道 我要入驻 投资者关系 商务合作 关于我们 联系我们 加入我们 36氪欧洲站 36氪欧洲站 36氪欧洲站 Ai产品日报 网络谣言信息举报入口 热门推荐 热门资讯 热门产品 文章标签 快讯标签 合作伙伴 36氪APP下载 iOS & Android 本站由 阿里云 提供计算与安全服务 违法和不良信息、未成年人保护举报电话：010-89650707 举报邮箱：[email protected] 网上有害信息举报 © 2011~ 2026 北京多氪信息科技有限公司 | 京ICP备12031756号-6 | 京ICP证150143号 | 京公网安备11010502057322号 意见反馈 36氪APP 让一部分人先看到未来 36氪 鲸准 氪空间 推送和解读前沿、有料的科技创投资讯 一级市场金融信息和系统服务提供商 聚焦全球优秀创业者，项目融资率接近97%，领跑行业

Source: 36Kr | Original Link

今年已有143家企业递表港股IPO-36氪 账号设置 我的关注 我的收藏 申请的报道 退出登录 登录 搜索 36氪Auto 数字时氪 未来消费 智能涌现 未来城市 启动Power on 36氪出海 36氪研究院 潮生TIDE 36氪企服点评 36氪财经 职场bonus 36碳 后浪研究所 暗涌Waves 硬氪 氪睿研究院 媒体品牌 企业号 企服点评 36Kr研究院 36Kr创新咨询 企业服务 核心服务 城市之窗 政府服务 创投发布 LP源计划 VClub VClub投资机构库 投资机构职位推介 投资人认证 投资人服务 寻求报道 36氪Pro 创投氪堂 企业入驻 创业者服务 创投平台 AI测评网 首页 快讯 资讯 推荐 财经 AI 自助报道 四川 最新 创投 汽车 科技 专精特新 直播 视频 专题 活动 搜索 寻求报道 我要入驻 城市合作 02 月 17 今年已有143家企业递表港股IPO 2026-02-17 11:43 分享至 打开微信“扫一扫”，打开网页后点击屏幕右上角分享按钮 Wind数据显示，2026年以来，已有143家企业完成了港股IPO递表，其中120家为首次递表。截至2月16日，港股市场仍有401宗新股申请仍在处理之中，包括4家已通过聆讯的企业。德勤中国资本市场服务部预测，2026年香港新股市场或更胜去年，料约160只新股融资不少于3000亿港元。一个值得关注的趋势是，近期在港股上市的科技类企业正在受到国内外长线资金追捧。例如澜起科技17家基石投资者共认购约4.50亿美元，其中包括瑞银资管、摩根大通资管这样的全球长线基金；以安本为代表的老牌长线欧洲资金；以及中邮理财、泰康人寿等内地传统长线资金。（证券时报） 原文链接 下一篇 消息称英国监管机构考虑调整规则以吸引中国企业上市 英国审计监管机构正在考虑调整会计规则，以鼓励中国公司赴伦敦上市。Financial Reporting Council周一表示，将就是否允许在中国注册的公司在伦敦发行全球存托凭证（GDR）时遵循中国的会计规则征询意见。全球存托凭证（GDR）是由银行发行、代表外国公司股票的凭证。根据该拟议的规则调整，中国公司在发行GDR时，将被允许在限定的一段时间内遵循中国的审计准则。根据现行规定，在伦敦发行GDR的中国实体需要遵循英国的会计准则。（新浪财经） 39分钟前 24小时热榜 查看更多榜单 帮长辈清手机成过年“基操”，这份2026年最新攻略请收好 红包大战变迁史：从社交到AI 过年吃“金拱门年夜饭”大逆不道？但多少打工人除夕想冲进麦当劳急头白脸搓一顿 22小时前 年轻人春节回家账单：花销超2万，机票占一半 20小时前 如何通过三个问题识破谎言 3小时前 在AI和机器人里找节目：马年春晚背后的科技大战揭秘 2小时前 揭秘春晚“两个蔡明”背后，这家机器人公司的百日奋战 1小时前 这个春节999元租个人形机器人拜年，然后呢？ 1小时前 关于36氪 城市合作 寻求报道 我要入驻 投资者关系 商务合作 关于我们 联系我们 加入我们 36氪欧洲站 36氪欧洲站 36氪欧洲站 Ai产品日报 网络谣言信息举报入口 热门推荐 热门资讯 热门产品 文章标签 快讯标签 合作伙伴 36氪APP下载 iOS & Android 本站由 阿里云 提供计算与安全服务 违法和不良信息、未成年人保护举报电话：010-89650707 举报邮箱：[email protected] 网上有害信息举报 © 2011~ 2026 北京多氪信息科技有限公司 | 京ICP备12031756号-6 | 京ICP证150143号 | 京公网安备11010502057322号 意见反馈 36氪APP 让一部分人先看到未来 36氪 鲸准 氪空间 推送和解读前沿、有料的科技创投资讯 一级市场金融信息和系统服务提供商 聚焦全球优秀创业者，项目融资率接近97%，领跑行业

Source: 36Kr | Original Link

Enough Cream: Perfect coffee, every time. | Product Hunt Launching today Enough Cream Launching today Perfect coffee, every time. 99 followers Perfect coffee, every time. 99 followers Visit website Camera apps Never ruin your coffee with too much cream again. Enough Cream uses your phone’s camera to analyze your coffee’s color in real-time, comparing it to your saved preference and telling you exactly when to stop pouring. Perfect consistency, every time. ☕ Overview Reviews Alternatives Built with Team More Free Launch tags: iOS • Coffee Launch Team / Built With Show more Show more Launching Today # 9 Day Rank Upvote • 122 points Follow Enough Cream Add to collection Share Analytics Company Info enough-cream.app App Store Enough Cream Info Launched in 2026 Forum p/enough-cream Similar Products Artisse AI Transforming photography forever using AI. 5.0 ( 205 reviews ) Photo editing Camera apps Camo Great video made easy 4.6 ( 5 reviews ) Video editing Camera apps FieldDay Create custom vision AI apps, using just your phone. 5.0 ( 3 reviews ) Camera apps Mobile editing apps VSCO Photo editing tools and a community for creators, by creators 4.7 ( 4 reviews ) Photo editing Camera apps Swatches Accurate color picker for the real world for iOS. Camera apps Mobile editing apps View more Subscribe Sign in Webflow | AI site builder — Start fast. Build right. Start fast. Build right. Promoted What do you think? … Login to comment Evan Lee Enough Cream Maker 📌 👋 I’m Evan, a software engineer at Shopify and coffee enthusiast. The Problem: We’ve all been there: you add cream to your coffee, and suddenly it’s either too light or you’ve wasted perfectly good coffee trying to fix it. Our Solution: Enough Cream uses your phone’s camera to analyze your coffee’s color in real-time and compares it to your saved preference. The app tells you whether to add more cream, more coffee, or if it’s perfect. Key Benefits: ☕ Perfect coffee consistency every single time 📱 Simple: just point your camera at your cup 🎯 Real-time guidance prevents mistakes We’re launching on iOS and Android. We’d love your feedback! Thanks for checking us out! ☕ Upvote (3) Report Share 11d ago Nichole Elizabeth DeMerè (Eithiriel) Feeling nostalgic. This reminds me of Product Hunt culture back in the day. Upvote (3) Report Share 11h ago Van de Vouchy Hey Evan, this is such a specific problem but I totally get it. Was there a specific morning where you ruined your coffee trying to fix the cream situation and thought okay this is ridiculous, there has to be a better way? Upvote (1) Report Share 18h ago Jan Schutte How do you account for different lighting conditions? Upvote (1) Report Share 15h ago Evan Lee Enough Cream Maker @janschutte It’s a great question – we added a feature to turn on the torch to help smooth out some of the edge cases with lighting for now! We’re planning a future feature where we can detect (from camera peripherals) what the lighting conditions are like and factor that into th

Source: Product Hunt | Original Link

HookWatch: Automated webhook monitoring for indie hackers & small teams | Product Hunt Launching today HookWatch Launching today Automated webhook monitoring for indie hackers & small teams 67 followers Automated webhook monitoring for indie hackers & small teams 67 followers Visit website Observability tools Never miss a webhook again. HookWatch monitors, logs, and retries your webhooks automatically. Built for indie hackers and small teams. HookWatch monitors your webhook endpoints 24/7 and alerts you instantly when something breaks. Built for developers who need reliable monitoring without enterprise complexity. Simple setup, affordable pricing, peace of mind. Overview Reviews Alternatives Team More Interactive Free Options Launch tags: API • SaaS • Developer Tools Pro Subscription Launch Team Show more Show more Launching Today # 13 Day Rank Upvote • 96 points Follow HookWatch Add to collection Share Analytics Company Info hookwatch.dev HookWatch Info Launched in 2026 Forum p/hookwatch Subscribe Sign in Social X Webflow | AI site builder — Start fast. Build right. Start fast. Build right. Promoted What do you think? … Login to comment Mihail HookWatch Maker 📌 Hey Product Hunt! 👋 I’m thrilled to launch HookWatch today — a tool I’ve been building to solve a problem that’s been driving me crazy for years. The problem: Modern apps rely on webhooks, cron jobs, WebSocket connections, and now AI agents — but when they fail, they fail silently . No error page. No stack trace. Just… nothing happens. A Stripe payment webhook gets lost, a nightly backup script silently stops running, a WebSocket connection drops mid-session, an AI agent starts erroring out — and you don’t find out until a customer complains or data is already gone. I’ve been on the other side of that support ticket too many times. So I built HookWatch. What is HookWatch? It’s four monitoring tools under one roof: 🔗 Webhook Monitor — Track every incoming webhook in real-time. Inspect full payloads, replay failed deliveries with one click, and get automatic retries with exponential backoff. We even have a unique request buffering feature: if your server goes down, HookWatch stores incoming webhooks and replays them when you’re back online. Zero data loss. ⏰ Cron Monitor — Schedule and monitor cron jobs with human-readable syntax. Write “every day at 2am” instead of decoding 0 2 * * *. Get full execution history with stdout/stderr, automatic retries, and instant alerts when something breaks. The CLI runs jobs locally with optional cloud sync — it works 100% offline. 🌐 WebSocket Monitor — A transparent proxy for your WebSocket connections that gives you complete visibility into bidirectional traffic. Point your client to a HookWatch proxy URL instead of the original server — we forward everything transparently while capturing every message in both directions. See live connections open and close, inspect full payloads (text and binary), filter by direction (inbound/outbound), and review co

Source: Product Hunt | Original Link

PenguinBot AI: Your AI-Employee Working 24/7 | Product Hunt Launching today PenguinBot AI Launching today Your AI-Employee Working 24/7 370 followers Your AI-Employee Working 24/7 370 followers Visit website Automation tools • AI Agent Automation PenguinBot is an action-first AI that turns conversations into real work. It manages emails, schedules tasks, creates documents, and runs workflows automatically. Just tell it what you need — it plans, executes, and keeps things moving in the background. Autonomous, secure, and built to run 24/7 so you can focus on what matters. Overview Reviews Alternatives Built with Team Awards More Free Options Launch tags: Productivity • Artificial Intelligence • Tech Launch Team / Built With Show more Show more Launching Today # 5 Day Rank Upvote • 165 points Follow PenguinBot AI Add to collection Share Analytics Company Info penguinbot.org PenguinBot AI Info Launched in 2026 Forum p/penguinbot-ai Subscribe Sign in Social X Framer — Launch websites with enterprise needs at startup speeds. Launch websites with enterprise needs at startup speeds. Promoted What do you think? … Login to comment Aryan Bains PenguinBot AI Maker 📌 We built PenguinBot because most AI tools still stop at giving answers. We wanted an AI that actually does the work. PenguinBot turns simple instructions into real actions — managing emails, scheduling tasks, creating documents, and running workflows automatically in the background. The goal is simple: less prompting, more execution. This is just the beginning, and we’re building it closely with the community. I’d love your honest feedback: • What would you automate first? • What integrations should we build next? • What would make this part of your daily workflow? I’ll be here all day answering questions. Thanks for checking out PenguinBot — excited to hear what you think! Upvote (11) Report Share 2d ago YUVRAJ PenguinBot AI Maker We built PenguinBot to go further — to actually do the work . Give it a simple instruction, and it takes action: • Manages emails • Schedules tasks • Creates documents • Runs workflows quietly in the background No endless prompting. No copy-pasting between tools. Just execution. Our belief is simple: AI shouldn’t just assist you — it should operate with you. PenguinBot is still evolving, and we’re building it hand-in-hand with the community. Upvote (9) Report Share 1d ago Damanjeet Singh PenguinBot AI Maker Excited to share PenguinBot’s long-term vision : Today, PenguinBot helps users automate workflows and connect across the tools they already use. Our future scope is to become a true cross-platform AI workspace companion: Native apps on Android, iOS, Windows, and macOS for seamless access anywhere Smarter multi-step automations that can plan, execute, and adapt in real time Voice-first interactions for hands-free productivity Better team collaboration with shared context and sync across devices A growing integration ecosystem so PenguinBot fits into every workflow,

Source: Product Hunt | Original Link

华尔街见闻

Source: Wall Street CN | Original Link

No content available.

Source: Hacker News | Original Link

Dolphin Emulator – Rise of the Triforce Download Blog Screenshots FAQ Guides Compatibility Community Discord Forum Wiki GitHub English Asturianu Bahasa Melayu Català Česky Cymraeg Dansk Deutsch Español Ελληνικά Français Galego Italiano Magyar Nederlands Norsk (Bokmål) Polski Português Português (Brasil) Русский Svenska Türkçe 中文(中国) 日本語 한국어 Blog 2026 February 16 Rise of the Triforce Rise of the Triforce Written by JMC47 , MayImilae , OatmealDome on Feb. 16, 2026 / Last update on Feb. 16, 2026 / Short link / Forum thread During the rapid technological advancements of the early 1990s, the video game industry was on the cusp of a massive addition – another dimension . With console shenanigans like the Super FX chip giving players a taste of 3D, hype was at an all-time high. But the games released for home consoles were nothing compared to what arcade developers were capable of doing. By employing gigantic budgets and cutting-edge hardware, the arcade gave players a chance to see the future, today. This is filler text to try and hack around a problem on the website. You shouldn’t be seeing this. If you are, please report this bug. In early 1994, this was 3D graphics for the home. 15 FPS, low resolution (256×192) , 6500 polygons per second, practically no textures, and no texture filtering. And this required extra hardware in the cartridge to achieve. Meanwhile, this was in arcades. 60 FPS , ~4x the pixel count (496×384), 300,000 polygons per second, and filled with perspective-correct and trilinear-filtered textures. Video Credit: arronmunroe But the future eventually arrived with the launch of the 5th generation of consoles . All of a sudden, the revolutionary 3D hardware features that were once exclusive to arcades were now available in home consoles. Without next-generation hype pushing players into the arcade, powerful but expensive arcade machines were no longer sustainable to develop. The industry adjusted by moving toward more cost effective solutions, with many turning to the inexpensive, already proven 3D-capable hardware available in 5th gen home consoles. Rather than turning around the decline of the arcade, the cheaper hardware may have helped accelerate it. There were fewer unique experiences to pull players into the arcade, and previous hit exclusives were now seeing high quality home console ports that allowed them to be enjoyed without munching quarters. When the 6th generation arrived with the Dreamcast and the PlayStation 2, many arcade stalwarts waved the white flag and started to shift their arcade divisions to home console projects, with mixed success . Sega was among those hit hardest by this era. They produced some of the greatest arcade thrills of the 1990s and enjoyed massive success in the home console market with the Sega Genesis/Mega Drive. But a string of mistakes and miscalculations combined with the slumping arcade industry sent them to the brink of bankruptcy. By 2002, the Dreamcast had been soundly defeated by the lau

Source: Hacker News | Original Link

AI is destroying Open Source, and it’s not even good yet – Jeff Geerling AI is destroying Open Source, and it’s not even good yet Feb 16, 2026 Over the weekend Ars Technica retracted an article because the AI a writer used hallucinated quotes from an open source library maintainer. The irony here is the maintainer in question, Scott Shambaugh, was harassed by someone’s AI agent over not merging it’s AI slop code. It’s likely the bot was running through someone’s local ‘agentic AI’ instance (likely using OpenClaw). The guy who built OpenClaw was just hired by OpenAI to “work on bringing agents to everyone.” You’ll have to forgive me if I’m not enthusastic about that. Video This blog post is a lightly-edited transcript of the video I published to YouTube today. Scroll past the video embed if you’re like me, and you’d rather read the text 🙂 Impacts on Open Source Last month, even before OpenClaw’s release, curl maintainer Daniel Stenberg dropped bug bounties because AI slop resulted in actual useful vulnerability reports going from 15% of all submissions down to 5%. And that’s not the worst of it—the authors of these bug reports seem to have a more entitled attitude: These “helpers” try too hard to twist whatever they find into something horribly bad and a critical vulnerability, but they rarely actively contribute to actually improve curl. They can go to extreme efforts to argue and insist on their specific current finding, but not to write a fix or work with the team on improving curl long-term etc. I don’t think we need more of that. These agentic AI users don’t care about curl. They don’t care about Daniel or other open source maintainers. They just want to grab quick cash bounties using their private AI army. I manage over 300 open source projects , and while many are more niche than curl or matplotlib, I’ve seen my own increase in AI slop PRs. It’s gotten so bad, GitHub added a feature to disable Pull Requests entirely . Pull Requests are the fundamental thing that made GitHub popular. And now we’ll see that feature closed off in more and more repos. AI slop generation is getting easier, but it’s not getting smarter. From what I’ve seen, models have hit a plateau where code generation is pretty good 1 … But it’s not improving like it did the past few years. The problem is the humans who review the code—who are responsible for the useful software that keeps our systems going—don’t have infinite resources (unlike AI companies). Some people suggest AI could take over code review too, but that’s not the answer. If you’re running a personal weather dashboard or building a toy server for your Homelab, fine. But I wouldn’t run my production apps—that actually make money or could cause harm if they break—on unreviewed AI code. If this was a problem already, OpenClaw’s release, and this hiring by OpenAI to democratize agentic AI further, will only make it worse. Right now the AI craze feels the same as the crypto and NFT boom , with the same signs o

Source: Hacker News | Original Link

Forest Diary – Reuben P. Box, US Forest Service US Forest Service Diaries Daily work diaries of Reuben P. Box , US Forest Service Ranger for the North Butte Protection Unit of the Lassen National Forest , stationed at Stirling City, California. These diaries document forest management, fire suppression, law enforcement, road construction, and daily life in the northern California mountains from 1927 to 1945. These family records were scanned and digitized by Lance Orner . Handwriting transcription by Mistral OCR . Text summaries and indexes built by Anthropic Claude . Hosting by DreamHost . In partnership with Working Toast, LLC and Stirling City Historical Society . For more information, contact Lance at [email protected] . Highlights: March 1927 — Diaries begin at Stirling City; transferred to Klamath National Forest February 1928 — Returned to Stirling City, North Butte Protection Unit April 20, 1931 — Stirling City town fire July 22, 1931 — Mud Creek Fire, largest fire of 1931, 10-day operation October 1932 — Federal arson arrests and Grand Jury testimony in Sacramento January 1938 — Transferred to Hat Creek District December 7, 1941 — Pearl Harbor attack, forest watches set up March 31, 1945 — Retired from the Forest Service 217 Months 4656 Days 7488 Pages 413 People 70 Places 50 Events 1927 March 10 pages April 31 pages May 33 pages June 32 pages July 30 pages August 26 pages September 28 pages October 32 pages November 22 pages December 14 pages 1928 January 23 pages February 25 pages March 26 pages April 31 pages May 36 pages June 30 pages July 40 pages August 32 pages September 27 pages October 28 pages November 29 pages December 15 pages 1929 January 21 pages February 28 pages March 34 pages April 32 pages May 33 pages June 31 pages July 33 pages August 33 pages September 33 pages October 31 pages November 33 pages December 25 pages 1930 January 31 pages February 28 pages March 5 pages April 28 pages May 31 pages June 31 pages July 31 pages August 33 pages September 30 pages October 28 pages November 27 pages December 30 pages 1931 January 29 pages February 25 pages March 31 pages April 35 pages May 43 pages June 41 pages July 50 pages August 48 pages September 47 pages October 45 pages November 33 pages December 23 pages 1932 January 43 pages February 33 pages March 41 pages April 38 pages May 37 pages June 37 pages July 43 pages August 42 pages September 40 pages October 36 pages November 41 pages December 28 pages 1933 January 34 pages February 20 pages March 15 pages April 38 pages May 40 pages June 51 pages July 47 pages August 40 pages September 45 pages October 42 pages November 39 pages December 35 pages 1934 January 37 pages February 33 pages March 46 pages April 36 pages May 39 pages June 42 pages July 46 pages August 40 pages September 39 pages October 42 pages November 40 pages December 26 pages 1935 January 32 pages February 36 pages March 34 pages April 38 pages May 39 pages June 47 pages July 47 pages August 47 pages Septem

Source: Hacker News | Original Link

GitHub – zachlatta/freeflow: Free and open source alternative to Wispr Flow / Superwhisper / Monologue / etc Skip to content You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert zachlatta / freeflow Public Notifications You must be signed in to change notification settings Fork 1 Star 84 Free and open source alternative to Wispr Flow / Superwhisper / Monologue / etc License MIT license 84 stars 1 fork Branches Tags Activity Star Notifications You must be signed in to change notification settings zachlatta/freeflow main Branches Tags Go to file Code Open more actions menu Folders and files Name Name Last commit message Last commit date Latest commit History 54 Commits 54 Commits .github/ workflows .github/ workflows Resources Resources Sources Sources .gitignore .gitignore FreeFlow.entitlements FreeFlow.entitlements Info.plist Info.plist LICENSE LICENSE Makefile Makefile README.md README.md View all files Repository files navigation FreeFlow Free and open source alternative to Wispr Flow , Superwhisper , and Monologue . ⬇ Download FreeFlow.dmg Works on all Macs (Apple Silicon + Intel) I like the concept of apps like Wispr Flow , Superwhisper , and Monologue that use AI to add accurate and easy-to-use transcription to your computer, but they all charge fees of ~$10/month when the underlying AI models are free to use or cost pennies. So over the weekend I vibe-coded my own free version! It’s called FreeFlow. Here’s how it works: Download the app from above or click here Get a free Groq API key from groq.com Press and hold Fn anytime to start recording and have whatever you say pasted into the current text field One of the cool features is that it’s context aware. If you’re replying to an email, it’ll read the names of the people you’re replying to and make sure to spell their names correctly. Same with if you’re dictating into a terminal or another app. This is the same thing as Monologue’s “Deep Context” feature. An added bonus is that there’s no FreeFlow server, so no data is stored or retained – making it more privacy friendly than the SaaS apps. The only information that leaves your computer are the API calls to Groq’s transcription and LLM API (LLM is for post-processing the transcription to adapt to context). FAQ Why does this use Groq instead of a local transcription model? I love this idea, and originally planned to build FreeFlow using local models, but to have post-processing (that’s where you get correctly spelled names when replying to emails / etc), you need to have a local LLM too. If you do that, the total pipeline takes too long for the UX to be good (5-10 seconds per transcription instead of <1s). I also had concerns around battery life. Some day! License Licensed under the MIT license. About Free and open source alternative to Wispr Flow / Superwh

Source: Hacker News | Original Link

南京联通进小黑屋了 – V2EX 首页 注册 登录 V2EX = way to explore V2EX 是一个关于分享和探索的地方 现在注册 已注册用户请 登录 V2EX › 宽带症候群 南京联通进小黑屋了 sinycn1 · 22 小时 33 分钟前 · 1787 次点击 昨晚打开某测试网站，发现我的 IP 变成了 116.147.102.9, 一看就是小黑屋 IP ，再次查询自己 PPPOE 发现公网也没了，变成了 100.108.115.25 ，同时 IPv6 消失了，我想了下家里没有跑 PCDN ，只有迅雷下载多一些， 目前的症状： 1.常规测速站点速度正常（白名单） 2.迅雷下载速度起不来了原来可以 40-50MB/s 现在只能 2-3MB/s 3.连接数变少了，下载的时候网络就卡了 4.出国速度没啥影响 由于我没啥特殊公网应用，目前先观察下，看后期是否会主动拉出小黑屋 第 1 条附言  · 13 小时 26 分钟前 没找客服，先找装维师傅，年后再处理了 南京 联通 小黑屋 28 条回复 • 2026-02-17 02:00:41 +08:00 1 thetbw 20 小时 38 分钟前 看来大家都是一起的，我还纳闷我最近的 ddns 咋出问题，域名访问不了了，一看也是公网 ip 没了。不过现在也没啥影响了，我找了个便宜的 nat 小鸡做中继，价格一个月 10 来块也能接受，比跟运营商斗智斗勇好多了。 2 lyz2754509784 17 小时 19 分钟前 via Android 家宽应该是可以投诉出来的，我的家宽出来了，坐标在常州，企业宽带的话放弃吧 半个月前进去的，到现在还没出来 3 naver1 17 小时 2 分钟前 “小黑屋”有点轻了，叫“联通黑狱” 4 love2328 17 小时 1 分钟前 1 想知道怎么判断是小黑屋 ip 5 mooyo 16 小时 34 分钟前 经典 116.147. 6 js9528 15 小时 35 分钟前 via iPhone @ lyz2754509784 现在也没用投诉也是说宽带测速正常，ip 地址随机分配 7 lyz2754509784 14 小时 57 分钟前 @ js9528 我是投诉的早，还好后面没再进去。。估计再进去也是这样子的，现在我的商宽投诉就喊装维来测速，测完速就走了，给他看 iperf3 或者日常使用他说他不懂要上报公司，公司那边就说测速没问题就是宽带没问题。。至于日常为什么限速是他们服务商对我限速不是联通对我限速。。。然后 ip 为什么异地说是江苏省内随机分配的。。真的逆天 8 lyz2754509784 13 小时 46 分钟前 @ love2328 江苏联通的话目前黑屋有：116.147.x 全段，122.192.185.x-122.192.200.x ，122.194 段（比较不常见，偶尔可以播到），157.0.203.x ，ip 查询属地为徐州、南通、南京、镇江。ipv6 表现为/64 的前缀地址，常见开头为 2408:823c:2e02 ，iperf3 打流可以发现限速大约在 3m-5m ，部分段严重限速可以限速到 1-2m ，同时表现为使用迅雷、steam 等下载时虽未跑满带宽但是此时网络几乎无法使用，todesk 、向日葵、uu 远程等软件几乎全天不可使用或仅在兼容模式的模糊画质下可用。出国的话速度不受影响但是基于 udp 的 hy2 、tuic 等节点用不了，然后基于 tcp 的部分节点会绕路，美西直连节点延迟由 210ms 左右增加至 290ms-320ms 区间，特别是晚上用网高峰期会网络卡顿 9 crushark 12 小时 9 分钟前 改 dns 试下，我原来进小黑屋就改 dns 放出来的 10 okzy520 12 小时 2 分钟前 找装维用处不大 同南京联通拉锯半年了 11 janeyee110 11 小时 47 分钟前 1 之前一直觉得只有电信才有比较多的限制，现在从越来越多的案例看来，三大运营商都每一个好东西。 12 barnettluo1994 11 小时 35 分钟前 没有用的，这个段一旦进去了 基本很难出来 除非你关系很硬。。。。 比如我。。。。。 我五个宽带账号全部是白名单。 其实这些黑名单段是中间加了个隧道。 13 barnettluo1994 11 小时 34 分钟前 @ okzy520 请装维上门检查你到底有没有不合规的用途。 展示一下自己的拓扑，解释一下到底为什么要有这么大上行。 14 barnettluo1994 11 小时 33 分钟前 @ lyz2754509784 没有加白的话， 基本后面还会再进去的 15 lyz2754509784 10 小时 55 分钟前 @ barnettluo1994 #14 有没有加白的方法，装维上门就一个劲测速，我给他展示拓扑和上传用在哪他也不听，我再三要求下他拍了张照片说传给公司了，不知道公司那边有没有收到了。。反正也没有回复，10010 那边也只会回电说装维上门测速的回复是宽带测速正常建议排查内网的设备…再往下扯就说是什么服务提供商对我的限速了，我质疑说怎么会同一天限速的他们又说不出来说反正不是联通的问题。。。 16 barnettluo1994 10 小时 54 分钟前 @ lyz2754509784 没有。 来的是装维基本就是加白无望， 我的都还是省网络一把手来看的，最后给加白的。 师傅那边啥都不懂，如果看到点不该看到的东西给你写成什么盒子之类的，更不会解封的 17 lyz2754509784 10 小时 52 分钟前 @ barnettluo1994 #16 目前除了找装维，找我的客户经理（这是我目前直接有联系方式的），有没有可行的可以通过他们联系到可以做有效申请的人物 做一个转达之类的也好 18 barnettluo1994 10 小时 51 分钟前 @ lyz2754509784 你最起码要能做到家里能有等保 2 级或者三级 级别的防护，他们才会愿意考虑解封的。。。。毕竟出事了 他们还要担责。 19 barnettluo1994 10 小时 42 分钟前 @ janeyee110 电信和联通采用的是一个设备一个规则 20 lyz2754509784 10 小时 39 分钟前 @ barnettluo1994 #18 没有那么专业的 目前就前置一个企业级的 IK-A700X 做拨号和流控分流，配一个飞塔的防火墙，部分重点的 http 页面再套了个雷池的 waf ，核心的 ssh 之类的敏感内容都是走的 tailscale 配自建的 derp 中转相当于内网 vpn 21 barnettluo1994 10 小时 19 分钟前 @ lyz2754509784 不要用爱快。。。。。 爱快也不给解封的。。。。 22 okzy520 9 小时 53 分钟前 @ barnettluo1994 来过了 展示过了 半个月都没坚持到 就又进小黑屋了 23 okzy520 9 小时 52 分钟前 @ barnettluo1994 运营商和爱快有仇？ 24 barnettluo1994 6 小时 57 分钟前 via iPhon

Source: V2EX | Original Link

H-1B Exposed | Banking Sector Visa Sponsorship Investigation An Independent Investigation The talent shortage is a myth. This is labor arbitrage. Three of America’s largest banks filed 7,474 Labor Condition Applications for foreign workers. Public records expose the playbook: file at scale, pay the legal minimum, and lock workers in with green card sponsorship. See the Data View Companies → Scroll Total LCA Filings 7,474 | Approval Rate 98-99% | Green Card Sponsors 5,142 | Near Wage Floor 31.8% | Commodity IT Roles 69.8% | Companies Tracked 3 | Years of Data FY2017-2025 | Total LCA Filings 7,474 | Approval Rate 98-99% | Green Card Sponsors 5,142 | Near Wage Floor 31.8% | Commodity IT Roles 69.8% | Companies Tracked 3 | Years of Data FY2017-2025 | The Numbers What the public records show. Every data point here comes from federal filings these companies made themselves — the same data they hope nobody reads. 7,474 Total LCA Filings FY2020–2026, three banks 5,142 Green Card Sponsorships PERM applications filed 69.8% Commodity IT Work Software devs, analysts, QA — not specialized talent 98–99% USCIS Approval Rate Rubber stamp. Virtually zero oversight. 31.8% Near Wage Floor Nearly 1 in 3 filings pay barely above the legal minimum 3.5% AmEx H-1B Workforce % And growing — highest ratio of the three banks The Playbook The talent shortage is a myth. This is labor arbitrage. 01 File at Scale Thousands of Labor Condition Applications filed per year. Volume is the strategy. 02 Pay the Floor Nearly a third of filings cluster right at the prevailing wage minimum. Compliance minimums, not competitive offers. 03 Route to Low-Cost Cities Workers sent to locations where prevailing wage obligations are lowest. Geographic arbitrage baked into the system. 04 Lock Them In Sponsor green cards. The worker can’t leave, can’t negotiate, can’t push back. Their immigration status is the leash. The Companies Three banks. Thousands of filings. Click through for the full data breakdown on each institution. 01 JPM JPMorgan Chase The largest U.S. bank by assets and one of the biggest H-1B sponsors in financial services. LCA Filings 3,534 Avg Salary $162,588 99.5 % APPROVED View Data → 02 AXP American Express A global financial services corporation known for credit cards and payment processing. LCA Filings 920 Avg Salary $151,466 99.1 % APPROVED View Data → 03 COF Capital One A major bank holding company specializing in credit cards, auto loans, and banking, with a large tech workforce. LCA Filings 728 Avg Salary $196,150 99.9 % APPROVED View Data → Methodology All sourced. All public. This entire investigation runs on public federal data — filings these companies submitted to the Department of Labor, USCIS, the SEC, and the Bureau of Labor Statistics. They created the paper trail. We read it. DOL Dept. of Labor OFLC Every LCA filing, prevailing wage determination, and worksite location. USCIS USCIS Employer Data Hub H-1B petition approvals, denials, and outcomes by employer and

Source: Hacker News | Original Link

Run NanoClaw in Docker Shell Sandboxes | Docker Running NanoClaw in a Docker Shell Sandbox Posted Feb 16, 2026 Oleg Selajev Ever wanted to run a personal AI assistant that monitors your WhatsApp messages 24/7, but worried about giving it access to your entire system? Docker Sandboxes’ new shell sandbox type is the perfect solution. In this post, I’ll show you how to run NanoClaw , a lightweight Claude-powered WhatsApp assistant, inside a secure, isolated Docker sandbox. What is the Shell Sandbox? Docker Sandboxes provides pre-configured environments for running AI coding agents like Claude Code, Gemini CLI, and others. But what if you want to run a different agent or tool that isn’t built-in? That’s where the shell sandbox comes in. It’s a minimal sandbox that drops you into an interactive bash shell inside an isolated microVM. No pre-installed agent, no opinions — just a clean Ubuntu environment with Node.js, Python, git, and common dev tools. You install whatever you need. Why Run NanoClaw in a Sandbox? NanoClaw already runs its agents in containers, so it’s security-conscious by design. But running the entire NanoClaw process inside a Docker sandbox adds another layer: Filesystem isolation – NanoClaw can only see the workspace directory you mount, not your home directory Credential management – API keys are injected via Docker’s proxy, never stored inside the sandbox Clean environment – No conflicts with your host’s Node.js version or global packages Disposability – Nuke it and start fresh anytime with docker sandbox rm Prerequisites Docker Desktop installed and running Docker Sandboxes CLI ( docker sandbox command available) (v.0.12.0 available in the nightly build as of Feb 13) An Anthropic API key in an env variable Setting It Up Create the sandbox Pick a directory on your host that will be mounted as the workspace inside the sandbox. This is the only part of your filesystem the sandbox can see: mkdir -p ~/nanoclaw-workspace docker sandbox create –name nanoclaw shell ~/nanoclaw-workspace Connect to it docker sandbox run nanoclaw You’re now inside the sandbox – an Ubuntu shell running in an isolated VM. Everything from here on happens inside the sandbox. Install Claude Code The shell sandbox comes with Node.js 20 pre-installed, so we can install Claude Code directly via npm: npm install -g @anthropic-ai/claude-code Configure the API key This is the one extra step needed in a shell sandbox. The built-in claude sandbox type does this automatically, but since we’re in a plain shell, we need to tell Claude Code to get its API key from Docker’s credential proxy: mkdir -p ~/.claude && cat > ~/.claude/settings.json << 'EOF' { "apiKeyHelper": "echo proxy-managed", "defaultMode": "bypassPermissions", "bypassPermissionsModeAccepted": true } EOF What this does: apiKeyHelper tells Claude Code to run echo proxy-managed to get its API key. The sandbox’s network proxy intercepts outgoing API calls and swaps this sentinel value for your real Anthropic key,

Source: Hacker News | Original Link

‎Wildex Identify Plants/Animals App – App Store Wildex Identify Plants/Animals Identify Animals, Plants, Bugs Only for iPhone Free · Designed for iPhone Share Ages 4+ Years Category Utilities Developer DREAMPRESS LTD Language EN English Size 196.3 MB iPhone Wildex – Discover Wildlife Around You Turn every walk into a treasure hunt. Wildex uses your camera to instantly identify any plant or animal, then adds it to your personal collection—like Pokémon Go, but everything is real. [Collect] • Point your camera at any living thing—get an instant ID with species info • Every species has a rarity tier based on how common it is in your area • Build a beautiful visual collection of everything you’ve found • That weird bird at the park? Now it’s yours forever [Discover Your Area] • See which species have been found near you—and which haven’t • Hidden legendary species are waiting in your neighborhood • Your local park has dozens of species you’ve walked past without noticing • Track where and when you found each one on your personal map [Compete] • Climb local and global leaderboards • Complete quests for bonus XP • Rare finds = massive points • See how your collection stacks up against friends • Share finds with one tap [Learn] • Every species comes with facts and info • Actually remember what you saw on that hike 100% free. Works anywhere there’s nature—your backyard counts! Lots of love, The Wildex Team more Ratings & Reviews This app hasn’t received enough ratings or reviews to display an overview. What’s New Version History MASSIVE UPDATE Improved AI model Improved library, card, collection and leaderboard screens Danger ratings and rankings Introducing: Wildboy – your’e wilderness guide! 2.0 1d ago Minor tweaks and bug fixes!! Happy collecting! 1.1 08/13/2025 1.0 08/12/2025 MASSIVE UPDATE Improved AI model Improved library, card, collection and leaderboard screens Danger ratings and rankings Introducing: Wildboy – your’e wilderness guide! more Version 2.0 1d ago App Privacy App Privacy The developer, DREAMPRESS LTD , indicated that the app’s privacy practices may include handling of data as described below. This information has not been verified by Apple. For more information, see the developer’s privacy policy . To help you better understand the developer’s responses, see Privacy Definitions and Examples . Privacy practices may vary, for example, based on the features you use or your age. Learn More Data Used to Track You The following data may be used to track you across apps and websites owned by other companies: Identifiers Usage Data Data Linked to You The following data, which may be collected and linked to your identity, may be used for the following purposes: Third-Party Advertising Identifiers Device ID Usage Data Advertising Data Analytics Usage Data Product Interaction App Functionality Location Precise Location Coarse Location Contact Info Email Address Name User Content Photos or Videos Identifiers User ID Data Not Linked to You The fol

Source: Hacker News | Original Link