Episode: China to Cover More Cancer and Rare Disease Drugs Duration: approximately 8 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, let me ask you a simple question. If a life-saving drug costs 3000 yuan a month, and insurance suddenly makes it cost only 300, would you say that is good news? zh:欢迎回到 Learn English with Podcasts！Sarah，我来问你一个简单的问题。如果一种救命药每月要花3000元，而医保突然让它只要300元，你会说这是好消息吗？ [Sarah]: That is not just good news, that is life-changing news. For many families, that difference decides whether they can keep taking the medicine or not. zh:那不仅仅是好消息，那是能改变生活的好消息。对很多家庭来说，这个差额决定了他们能否继续吃药。 [Mike]: Exactly. And that is what happened in China this week. On September 4, at a big press conference in Beijing about the 15th Five-Year Plan, the National Healthcare Security Administration made a clear promise. zh:没错。而这正是本周在中国发生的事。9月4日，在北京一场关于“十五五”规划的新闻发布会上，国家医保局作出了一个明确的承诺。 [Sarah]: The 15th Five-Year Plan — that is China's plan for the next 5 years, right? What did they promise? zh:“十五五”规划——就是中国未来5年的计划，对吧？他们承诺了什么？ [Mike]: Yes. Vice Director Li Tao said the government will adjust the basic medical insurance drug list every year. And they will put more innovative drugs into the insurance — especially for cancer, chronic diseases, rare diseases, and children's diseases. zh:是的。副局长李滔说，政府将每年调整基本医保药品目录。并且将把更多创新药纳入医保——尤其是针对肿瘤、慢性病、罕见病和儿童疾病的药。 [Sarah]: Innovative drugs — you mean new drugs that are very effective but usually very expensive? zh:创新药——你是说那些非常有效但通常非常贵的新药吗？ [Mike]: Exactly. The newest science, like targeted cancer therapies or rare disease treatments. In the past, many of these drugs were not covered, so families had to pay the full price themselves. zh:完全正确。最新的科技，比如靶向抗癌疗法或罕见病治疗药。过去，很多这类药不在报销范围，家庭必须自己付全款。 [Sarah]: So what is the situation now? How many drugs does China's basic insurance already cover? zh:那现在情况如何？中国的医保已经覆盖了多少种药？ [Mike]: The current list, which started on January 1, 2026, includes 3253 Western and Chinese patent medicines, plus 892 traditional Chinese medicine pieces. And over the last 8 years, the government has added 199 innovative drugs through negotiations. zh:现行的目录从2026年1月1日开始执行，包括3253种西药和中成药，外加892种中药饮片。而在过去8年里，政府通过谈判已经新增了199种创新药。 [Sarah]: 199 new drugs in 8 years — that is a lot. What was new just this year? zh:8年新增199种新药——这非常多。今年新增了哪些？ [Mike]: This year, 114 new drugs were added, and 50 of them were Class 1 new drugs — that means completely new drugs developed in China. For example, a breast cancer drug called sacituzumab, a pancreatic cancer drug, a diabetes drug called tirzepatide, and a cholesterol drug that you only need twice a year. zh:今年新增了114种药，其中50种是1类新药——就是在中国全新研发的药。比如一种叫芦康沙妥珠单抗的乳腺癌药、一种胰腺癌药、一种叫替尔泊肽的糖尿病药，还有一种一年只需打两针的降胆固醇药。 [Sarah]: Wait, twice a year for cholesterol? That sounds much easier than taking pills every day. zh:等等，一年两针就能降胆固醇？这听起来比每天吃药容易多了。 [Mike]: It is. And there is also a thalassemia drug called deferiprone. In places like Guangxi, patients used to pay almost 3000 yuan a month for it. After negotiation and insurance, they now pay only about one tenth. zh:确实。还有一种治疗地中海贫血的药叫去铁酮。像在广西，患者过去每月要付近3000元。经过谈判和医保报销后，现在只需付大约十分之一。 [Sarah]: So from 3000 to 300 — that brings us back to your first question. That one tenth is exactly the difference between giving up and continuing treatment. zh:所以从3000到300——这就回到了你一开始的问题。这十分之一，正是放弃治疗和坚持治疗的区别。 [Mike]: Yes. And on September 4, Li Tao said they want more stories like that. They will also make the reimbursement list for medical services and medical consumables the same across the whole country. So the difference between regions will be smaller. zh:是的。而在9月4日，李滔说他们想要更多这样的故事。他们还将推动医疗服务项目和医用耗材报销目录在全国统一。这样地区之间的差异就会更小。 [Sarah]: You mean right now, a treatment might be covered in Shanghai but not in a smaller city? zh:你是说现在，一项治疗在上海可能报销，但在小城市就不行？ [Mike]: Exactly. By making a national unified list, people can get fairer coverage no matter where they live. That is the second part of the promise. zh:没错。通过制定全国统一的目录，无论住在哪里，人们都能得到更公平的保障。这是承诺的第二部分。 [Sarah]: This all sounds great, but how does the government make expensive drugs affordable? Do they just pay whatever the company asks? zh:这些听起来都很棒，但政府如何让昂贵的药变便宜的？难道公司要多少就给多少吗？ [Mike]: No, that is the smart part. Every year, the government does a negotiation. They bring together about 1.3 billion insured people as one big buyer. And they say to the drug company: if you lower your price, we will put your drug in the insurance and millions of people can use it. zh:不是，这是最巧妙的部分。每年，政府都会进行谈判。他们把约13亿参保人集合成一个巨大的买方。然后对药企说：如果你降价，我们就把你的药放进医保，让数百万人都能用上。 [Sarah]: Ah, so it is like group shopping. Lower price, but many more customers. Did it work? zh:啊，所以就像团购。价格更低，但客户多得多。这奏效了吗？ [Mike]: It worked very well. As of February 2026, the fund has paid 504.8 billion yuan for negotiated drugs since 2018, which drove total sales to 740 billion yuan and helped 1.17 billion visits. And the industry also grew — profits up 11.3% a year, research spending up 23% a year. zh:非常奏效。截至2026年2月，自2018年以来基金已为谈判药品支付5048亿元，带动总销售7400亿元，惠及11.7亿人次。而且行业也在增长——利润年均增长11.3%，研发投入年均增长23%。 [Sarah]: So both patients and companies win. But what about the most expensive drugs that are still too costly even after negotiation? Like CAR-T therapy for cancer? zh:所以患者和企业双赢。但那些即使谈判后依然太贵的药怎么办？比如治疗癌症的CAR-T疗法？ [Mike]: That is why China created a second list — the Commercial Health Insurance Innovative Drug List. The first version started on January 1, 2026 with 19 drugs. These are high-value drugs for cancer, Alzheimer's, and rare diseases in children. zh:这就是为什么中国创建了第二份清单——商业健康保险创新药目录。第一版于2026年1月1日启动，有19种药。这些是针对癌症、阿尔茨海默病和儿童罕见病的高价值药。 [Sarah]: So these 19 are not in basic insurance yet, but commercial insurance can recommend covering them? zh:所以这19种还没进基本医保，但商业保险可以推荐报销它们？ [Mike]: Yes. And for hospitals, using drugs from this list has three special rules. They do not count toward the hospital's self-pay rate, they are not counted as replaceable drugs in centralized procurement, and the payment is not part of the normal disease-based payment. So hospitals are not afraid to use them. zh:是的。而且对医院来说，使用这份目录里的药有三个特殊政策。它们不计入医院的自费率指标，不纳入集采可替代品种监测，也不计入常规的按病种付费。所以医院不怕使用它们。 [Sarah]: That solves the old problem — sometimes a drug is approved but hospitals do not want to use it because of cost controls. zh:这就解决了老问题——有时候一种药获批了，但医院因为控费而不愿使用。 [Mike]: Exactly. As of May 2026, those 19 drugs were already available in 1486 medical institutions, double the number at the start of the year. And more than 100 commercial insurance products now include them. zh:没错。截至2026年5月，这19种药已在1486家医疗机构配备，是年初的两倍多。而且已有100多个商业保险产品覆盖了它们。 [Sarah]: And what is next? When will we know which new drugs are added? zh:那接下来呢？我们什么时候能知道哪些新药会被加入？ [Mike]: Right now, actually. In June, the government received 818 applications for 674 generic drugs. And the next negotiation is starting on September 5 in Beijing. For the first time, they will negotiate the basic list and the commercial list together, at the same time. The results will be announced in November. zh:就是现在。6月，政府收到了674个通用名共818份申报。而下一轮谈判将于9月5日在北京开始。第一次，他们将同时对基本目录和商保目录进行谈判。结果将在11月公布。 [Sarah]: So in just two months, we will know which cancer, rare disease, and children's drugs become much cheaper for families. zh:所以短短两个月后，我们就会知道哪些癌症、罕见病和儿童用药会为家庭便宜很多。 [Mike]: Yes. And Li Tao also said they want to push unified national standards so the promise is real everywhere — from Beijing to a small county hospital in Guangxi. zh:是的。李滔还说他们想推动全国统一的标准，让承诺在各地都落到实处——从北京到广西的小县医院。 [Sarah]: You know, my favorite part of this story is a small detail. For basic drugs, the government also added four domestic Class 1 innovative drugs to the National Essential Drug List on September 1 — including a depression drug and a cholesterol drug. That list tells every public hospital: you should have these. zh:你知道吗，我最喜欢这个故事里的一个小细节。9月1日，政府还把四种国产1类创新药加入了国家基本药物目录——包括一种治疗抑郁的药和一种降胆固醇药。这份目录告诉每一家公立医院：你们应该配备这些药。 [Mike]: Right. So it is not just about paying. It is about making sure the drug is actually in the hospital when you need it. zh:没错。所以不仅仅是支付问题。还要确保当你需要时，药真的就在医院里。 [Sarah]: It reminds me of what my aunt says: you hope you never need the most expensive drug in the pharmacy, but you sleep better knowing it is there and you can afford it if you do need it. zh:这让我想起我阿姨说的话：你希望永远不需要药房里最贵的那种药，但知道它就在那里、需要时你能负担得起，你会睡得更安稳。 [Mike]: That is a perfect way to put it. Insurance does not make us sick less often, but it makes getting better less scary. zh:说得太好了。医保不会让我们少生病，但它让治好病变得不那么可怕。 [Sarah]: Thanks for listening to Learn English with Podcasts today. Have you ever seen insurance help your family with a big medical cost? Tell us your story. zh:感谢今天收听 Learn English with Podcasts。你有没有见过医保帮助你的家人支付一大笔医疗费用？告诉我们你的故事。 [Mike]: And we will see you next time. Maybe by November, some of those 818 applications will have become good news for someone you know. zh:我们下期再见。也许到11月，这818份申请中的一些就会成为你认识的某人的好消息。

Episode: Norway Goes Electric: 97.6% New Cars Are EVs Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, imagine you walk into a car showroom and almost every new car is electric. Would you believe it? zh:欢迎回到 Learn English with Podcasts！Sarah，想象一下你走进一家汽车展厅，几乎每一辆新车都是电动车，你会相信吗？ [Sarah]: Almost every car? That sounds like a science fiction movie. How many are we talking about? zh:几乎每一辆？那听起来像科幻电影。我们在说多少比例？ [Mike]: In Norway, in July, 97.6% of all new cars were pure electric. Out of 9609 new cars, 9379 were electric. zh:在挪威，今年7月，97.6%的新车是纯电动车。在9609辆新车中，有9379辆是电动车。 [Sarah]: Wow, 97.6%? So how many gas cars were left? zh:哇，97.6%？那还剩下多少燃油车？ [Mike]: Only 15 gasoline cars and 2 diesel cars. Just 17 fuel cars in the whole country for the entire month. zh:只有15辆汽油车和2辆柴油车。整个国家整整一个月只卖出17辆燃油车。 [Sarah]: 17? In a whole country? That is crazy. What about hybrids? zh:17辆？在一个国家里？太疯狂了。那混动车呢？ [Mike]: Also very few. Only 96 regular hybrids and 117 plug-in hybrids. So together, fuel and hybrid cars were less than 3% of the market. zh:也非常少。只有96辆普通混动和117辆插电混动。所以加起来，燃油车和混动车不到市场的3%。 [Sarah]: That means Norway is the first country in the world where new gas cars almost disappeared. But is this normal in Europe? zh:这意味着挪威成了全球第一个新燃油车几乎消失的国家。但这在欧洲算正常吗？ [Mike]: Not at all. In Spain, only 10% of new cars are electric. In Italy, it is just 5.9%. Norway is completely different. zh:完全不正常。在西班牙，只有10%的新车是电动车。在意大利，只有5.9%。挪威完全不同。 [Sarah]: So how did Norway do it? This did not happen overnight, right? zh:那挪威是怎么做到的？这不是一夜之间发生的吧？ [Mike]: Exactly. 7 years ago, in 2019, Norway was already number one in Europe. Back then 58% of new cars were electric or plug-in hybrid. And now it is 97.6%. zh:没错。7年前，也就是2019年，挪威就已经是欧洲第一了。当时58%的新车是纯电或插电混动。而现在是97.6%。 [Sarah]: That is a huge jump. What is their secret? Did they ban gas cars? zh:这是一个巨大的飞跃。他们的秘诀是什么？他们禁止燃油车了吗？ [Mike]: No ban. They used money to change behavior. For many years, electric cars in Norway had no VAT and no registration tax. So an electric car was often cheaper than the same size gasoline car. zh:没有禁令。他们用经济手段改变行为。多年来，挪威的电动车免增值税和注册税。所以一辆电动车常常比同级别的燃油车更便宜。 [Sarah]: Wait, an electric car cheaper than a gas car? That is the opposite of most countries. zh:等等，电动车比燃油车便宜？这和大多数国家相反。 [Mike]: Yes, that is why it worked. And there were more perks. Cheaper tolls, free or cheap parking, you could even drive in the bus lane, and there were charging stations everywhere. zh:是的，这就是它奏效的原因。还有更多福利。过路费更便宜，停车免费或优惠，甚至可以走公交专用道，而且到处都有充电桩。 [Sarah]: Ah, so they made electric life really easy, and gas life a little harder. Did it cost the government a lot? zh:啊，所以他们让电动车的生活变得非常便利，让燃油车的生活稍微难一点。这让政府花了很多钱吧？ [Mike]: It did. In 2021 alone, Norway lost 39.4 billion kroner in taxes because of these electric car perks. zh:确实花了很多。仅在2021年，挪威就因为这些电动车优惠损失了394亿克朗的税收。 [Sarah]: 39.4 billion? That is a lot of money. Can they keep doing this forever? zh:394亿？那是一大笔钱。他们能永远这样补贴下去吗？ [Mike]: They are slowly changing it. Right now, the VAT-free limit is 500,000 kroner. The new plan is to lower it to 300,000 kroner, and by 2027, the free VAT will end completely. zh:他们正在慢慢调整。现在免增值税的上限是50万克朗。新计划是降到30万克朗，到2027年完全取消免税。 [Sarah]: And what about gas cars? zh:那燃油车呢？ [Mike]: Taxes on fuel cars will keep going up. The idea is simple — keep pushing people toward electric. zh:燃油车的税会继续上涨。思路很简单——继续把人们推向电动车。 [Sarah]: Okay, so people are buying electric. But what brands are they buying? I thought Tesla owned Norway. zh:好吧，所以人们在买电动车。但他们买什么品牌？我以为特斯拉统治着挪威。 [Mike]: For many years, Tesla was number one in Oslo. Oslo has the highest density of electric cars in the world. But now, things are more competitive. zh:多年来，特斯拉在奥斯陆是第一。奥斯陆是全球电动车密度最高的城市。但现在，竞争更激烈了。 [Sarah]: So who was number one in July? zh:那7月谁是第一？ [Mike]: Toyota. With 1260 cars. More than double the same month last year. Volkswagen was second with 985 cars. zh:丰田，卖了1260辆。比去年同月翻了一倍多。大众第二，卖了985辆。 [Sarah]: Toyota first? I thought Toyota was famous for hybrids. zh:丰田第一？我以为丰田以混动闻名。 [Mike]: It is. But in Norway, even Toyota now sells mostly electric. And here is the big surprise — number three is XPENG, a Chinese company, with 843 cars. zh:确实是。但在挪威，即使是丰田现在也主要卖电动车。而最大的惊喜是——第三名是小鹏，一家中国公司，卖了843辆。 [Sarah]: A Chinese brand in the top 3 in Norway? That is new. zh:一个中国品牌在挪威排进前三？这很新鲜。 [Mike]: It shows how much the market has changed. A few years ago, people in Europe barely knew Chinese car brands. Now one is beating many old European brands in Norway. zh:这说明市场变化有多大。几年前，欧洲人几乎不认识中国汽车品牌。现在有一个已经在挪威超过了很多老牌欧洲品牌。 [Sarah]: It makes sense. If you make good, affordable electric cars, people will buy them, no matter where the brand is from. zh:有道理。如果你造出好用、价格合理的电动车，人们就会买，不管品牌来自哪里。 [Mike]: That is the lesson from Norway. They did not just build better cars. They built a better system. Cheaper prices, easy charging, real benefits every day. zh:这就是挪威的启示。他们不只是造出了更好的车。他们建立了更好的体系。更便宜的价格、便捷的充电、每天都能感受到的实惠。 [Sarah]: I love that. And you know what is funny? Norway is famous for oil. It is one of the biggest oil producers in the world. And now it is the first country to stop buying gas cars. zh:我很喜欢这一点。你知道什么很有趣吗？挪威以石油闻名，是全球最大的石油生产国之一。而现在它是第一个不再购买燃油车的国家。 [Mike]: Right! The country that got rich from oil is the first to leave oil cars behind. That is a perfect twist. zh:没错！靠石油致富的国家，第一个告别燃油车。这是个完美的反转。 [Sarah]: So maybe the future is not about who has the most oil under the ground, but who builds the best charging station on the street. zh:所以也许未来不在于谁的地下石油最多，而在于谁在街上建了最好的充电站。 [Mike]: Exactly. And for our listeners — if 97.6% of a whole country can switch, maybe switching is not as scary as we think. zh:没错。对我们的听众来说——如果一个国家97.6%的人都能切换，也许改变并没有我们想的那么可怕。 [Sarah]: Thanks for listening to Learn English with Podcasts today! What do you drive — gas or electric? Tell us what you think, and will your country be the next Norway? zh:感谢今天收听 Learn English with Podcasts！你开的是油车还是电车？告诉我们你的想法，你的国家会成为下一个挪威吗？ [Mike]: See you next time. And keep watching the road — it is getting more electric every month. zh:下期见。继续关注马路——它每个月都变得更电动。

Episode: The Wolf Boy of Spain: 12 Years in the Wild Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, what if you were raised by wolves? zh:欢迎回到 Learn English with Podcasts！Sarah，如果你是被狼群养大的，会怎样？ [Sarah]: Raised by wolves? Like in a fairy tale? That sounds wild — and a little scary. zh:被狼养大？像童话里那样？听起来很野，也有点吓人。 [Mike]: It is not a fairy tale. On August 15, a man in Spain died at 80. His name was Marcos Rodriguez Pantoja. People called him the Wolf Boy of Spain. zh:这不是童话。8月15日，西班牙一位80岁的老人去世了。他叫马科斯·罗德里格斯·潘托哈。人们叫他西班牙狼孩。 [Sarah]: The Wolf Boy? Did he really live with wolves? zh:狼孩？他真的和狼一起生活过吗？ [Mike]: He did. For 12 years. From age 7 to age 19. Alone in the Sierra Morena mountains in southern Spain. zh:真的，整整12年。从7岁到19岁。独自在西班牙南部的塞拉莫雷纳山脉里。 [Sarah]: Wait, 12 years? How did a 7-year-old survive alone in the mountains? zh:等等，12年？一个7岁的孩子怎么在山里独自活下来的？ [Mike]: His early life was very hard. His mother died when he was 3. His father married again, but his stepmother was cruel to him. His family was too big and too poor, so they sent him to the mountains to herd goats with an old man. zh:他的童年非常艰难。3岁时母亲去世，父亲再婚，但继母对他很残忍。家里人口太多又很穷，所以把他送到山上跟一位老人一起放羊。 [Sarah]: So he was just a little goat herder. What happened to the old man? zh:所以他只是个小牧羊人。那位老人后来怎么样了？ [Mike]: The old man died when Marcos was 7. Suddenly the boy was completely alone. No family, no house, no human voice. zh:老人去世时，马科斯才7岁。男孩突然完全孤身一人。没有家人，没有房子，没有人的声音。 [Sarah]: That is heartbreaking. A 7-year-old, alone in a wild forest. What did he do? zh:太让人心碎了。一个7岁的孩子，独自在荒野森林里。他怎么办的？ [Mike]: At first he just watched animals. He watched what they ate — wild fruits, roots, leaves. He copied them. That was how he stayed alive. zh:一开始他只是观察动物。看它们吃什么——野果、根茎、树叶。他模仿它们。就是这样活了下来。 [Sarah]: He learned from animals. That is amazing. But when did the wolves come in? zh:他向动物学习。这太厉害了。但狼是什么时候出现的？ [Mike]: He said a mother wolf found him. She had just fed her cubs and then she threw a piece of meat toward him. zh:他说是一只母狼发现了他。它刚喂完自己的幼崽，就朝他扔了一块肉。 [Sarah]: A wolf giving meat to a human child? Was he not scared? zh:狼给人类的小孩送肉？他不害怕吗？ [Mike]: He was terrified. He thought the wolf would attack him. He did not dare to touch the meat. But the wolf just pushed the meat toward him with her nose. zh:他吓坏了。以为狼要攻击他，不敢碰那块肉。但母狼只是用鼻子把肉推到他面前。 [Sarah]: Oh, like a gentle invitation. zh:哦，像是一个温柔的邀请。 [Mike]: Exactly. He picked up the meat and ate it. He was sure the wolf would bite him. But instead, she licked him. And from that moment, he became part of the family. zh:没错。他捡起肉吃了，以为狼会咬他。但它却舔了他。从那一刻起，他就成了这个家庭的一员。 [Sarah]: Wow. So the wolf adopted him. Did he live with the whole pack? zh:哇，所以母狼收养了他。他是跟整个狼群一起生活吗？ [Mike]: Yes, with the whole pack. He learned to live like a wolf. He also had another strange friend — a snake. He fed it goat milk, and the snake followed him everywhere. zh:是的，跟整个狼群一起。他学会了像狼一样生活。他还有另一个奇怪的朋友——一条蛇。他喂它羊奶，蛇就到处跟着他。 [Sarah]: A boy, a wolf pack, and a pet snake. That sounds like a movie. zh:一个男孩、一群狼、还有一条宠物蛇。听起来像电影。 [Mike]: It actually became a movie. In 2010, his story was made into a film called Entrelobos, or Among Wolves. But the real story after that is much sadder. zh:真的被拍成了电影。2010年，他的故事被拍成电影《与狼共伍》。但之后的真实故事要悲伤得多。 [Sarah]: What happened? Did people find him? zh:发生了什么？人们发现他了吗？ [Mike]: In 1965, the Spanish Civil Guard was searching the mountains. A young man crawled out of the forest. He was 19, naked, barefoot, walking on all fours. His feet were covered with hard, thick skin. And he could not speak. zh:1965年，西班牙国民警卫队在山里搜寻。一个年轻人从丛林里爬了出来。他19岁，赤身裸体，光着脚，四肢着地。脚上长满了厚厚的老茧。而且他不会说话。 [Sarah]: 19 years old and still on all fours? That is hard to imagine. zh:19岁了还四肢爬行？很难想象。 [Mike]: The guards forced him down the mountain. He howled and bit like a wolf. They had to tie his hands and feet and cover his mouth. zh:警卫队员强行把他从山上拖下来。他像狼一样嚎叫和撕咬。他们不得不绑住他的手脚、堵住他的嘴。 [Sarah]: That must have been terrifying for him. He did not know he was being saved. He thought he was being captured. zh:对他来说一定很可怕。他不知道自己是被营救，还以为是被抓捕。 [Mike]: Later he said, returning to human society was the most frightening experience of his life. Worse than being alone in the forest. zh:后来他说，重返人类社会是他一生中最可怕的经历。比独自在森林里还要可怕。 [Sarah]: Where did they take him? zh:他们把他带到哪里了？ [Mike]: To an orphanage. Nuns taught him to walk upright, to sit at a table, to use a knife and fork. He said walking on two legs felt strange. On all fours felt more comfortable. zh:送到一家孤儿院。修女们教他直立行走、坐在桌前吃饭、如何使用刀叉。他说用两条腿走路感觉很奇怪。四肢着地反而更舒服。 [Sarah]: I can understand that. He had lived as a wolf for 12 years. That was his normal. zh:我能理解。他像狼一样生活了12年。那就是他的常态。 [Mike]: Slowly, he learned to speak again. And he could finally tell his story. He said those 12 years with the wolves were the happiest years of his life. zh:慢慢地，他重新学会了说话。终于能讲述自己的故事。他说，和狼群在一起的12年是他一生中最快乐的时光。 [Sarah]: Happier than life with humans? That says a lot about his early family. zh:比和人类在一起还快乐？这说明他早年的家庭经历了很多痛苦。 [Mike]: He tried to go back. Years later, he went to the mountains to find his pack. But the cave where he lived had been turned into a house. And the wolves did not accept him anymore. zh:他试过回去。几年后，他回到山里去找他的狼群。但他住过的洞穴已经被改成了别墅。而且狼群也不再接纳他。 [Sarah]: Oh no. He lost both worlds. He was no longer a wolf, but he never fully felt human either. zh:哦不。他失去了两个世界。他不再是狼，但也从未完全觉得自己是人类。 [Mike]: That is the part that stays with me. The forest gave him food and family when humans gave him nothing. And when humans took him back, the forest closed its door. zh:这正是最让我难忘的部分。当人类什么都没给他时，森林给了他食物和家庭。而当人类把他带回去后，森林却关上了门。 [Sarah]: So what does this story teach us? What is really wild — the forest or human society? zh:所以这个故事教会我们什么？到底什么才是真正荒野的——是森林，还是人类社会？ [Mike]: Maybe the wolves knew something we forget. Family is not about blood. It is about who pushes the meat toward you, and who licks you instead of biting you. zh:也许狼懂得一些我们遗忘的事。家庭不在于血缘。而在于谁会把肉推向你，谁会选择舔你而不是咬你。 [Sarah]: That is beautiful. And also a little funny — I am thinking about my own life. My cat pushes food off the table, not toward me. zh:真美。也有点好笑——我想到了自己的生活。我的猫只会把食物从桌上推下去，而不是推给我。 [Mike]: Haha, right. Your cat is not adopting you. But Marcos's story makes you think. Next time life feels noisy and cold, maybe happiness is simpler — a piece of meat shared, a tongue that says you belong. zh:哈哈，对。你的猫可不会收养你。但马科斯的故事让人思考。下次当生活感到嘈杂寒冷时，也许快乐很简单——一块分享的肉，一个告诉你你属于这里的舔舐。 [Sarah]: Thanks for listening to Learn English with Podcasts today. If this story touched you, share it with a friend. And tell us — what does home mean to you? zh:感谢今天收听 Learn English with Podcasts。如果这个故事触动了你，就分享给朋友吧。也告诉我们——对你来说，家意味着什么？ [Mike]: See you next time. And be kind — you never know who needs a little meat pushed their way. zh:下期见。记得善良一点——你永远不知道谁正需要有人把一块肉推到他面前。

Episode: BYD x PaXini: Robots That Can Feel in Factories Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, quick question — can your robot feel? zh:欢迎回到 Learn English with Podcasts！Sarah，问你个小问题——你的机器人会有触觉吗？ [Sarah]: My robot? Mike, I do not even have a robot at home. Just a very slow vacuum. zh:我的机器人？Mike，我家里连机器人都没有，只有一个很慢的吸尘器。 [Mike]: Right, most factory robots are like that vacuum. They can see and move, but they cannot feel. zh:对，大多数工厂机器人就像那个吸尘器一样。它们能看、能动，但不会感知触觉。 [Sarah]: That is true. They can lift a heavy box, but they do not know if the box is slipping. zh:确实，它们能搬起重箱子，却不知道箱子是不是正在打滑。 [Mike]: Exactly. And that is why a new deal in China caught my eye. On August 25, 2026, BYD and a company called PaXini Tech signed a strategic agreement. zh:没错，这就是为什么中国的一笔新合作吸引了我的注意。2026年8月25日，比亚迪和一家叫帕西尼感知科技的公司签署了战略合作协议。 [Sarah]: PaXini? I have heard of BYD, the big electric car maker. But PaXini sounds new. zh:帕西尼？我听说过比亚迪，那家很大的电动车公司。但帕西尼听起来很新。 [Mike]: PaXini Tech is a startup that makes touch for robots. Their specialty is tactile sensors and humanoid robots. They build the fingertips of robots. zh:帕西尼是一家给机器人制造触觉的初创公司。他们的专长是触觉传感器和人形机器人。他们在打造机器人的指尖。 [Sarah]: So BYD builds cars, PaXini builds robot skin. Why team up? zh:所以比亚迪造车，帕西尼造机器人皮肤。为什么要联手？ [Mike]: The signing was at BYD's global headquarters. Big names were there. PaXini's founder and CEO Xu Jincheng, co-founder Nie Xiangru, and from BYD, vice president Luo Zhongliang and board secretary Li Qian. zh:签约就在比亚迪全球总部举行。大人物都到场了。帕西尼创始人兼CEO许晋诚、联合创始人聂相如，还有比亚迪方面的副总裁罗忠良和董事会秘书李黔。 [Sarah]: Wow, senior leaders from both sides. Who actually signed the paper? zh:哇，双方高层都来了。是谁代表签字呢？ [Mike]: Zhao Weibing, head of BYD's Future Lab, and Luo Xiaoheng, chief strategy officer of PaXini. So it was serious. zh:比亚迪未来实验室主任赵伟冰和帕西尼首席战略官罗霄恒代表签字。所以这是一次很正式的合作。 [Sarah]: Okay, so it is official. But what will they actually do together? zh:好吧，合作已经官宣了。但他们具体要一起做什么呢？ [Mike]: They will work together on two things: data collection and robots for real factories. BYD will give PaXini something very valuable — real factory life. zh:他们将在两件事上合作：数据采集和面向真实工厂的机器人。比亚迪会给帕西尼一样非常宝贵的东西——真实的工厂生活。 [Sarah]: What do you mean by real factory life? Like, a car assembly line? zh:你说的真实工厂生活是什么意思？比如，汽车装配线？ [Mike]: Exactly. Building a car is very complex. You have thousands of steps, tiny parts, strong press machines, long supply lines. BYD has huge factories and a giant supply chain to test robots in. zh:没错。造车非常复杂。有成千上万道工序、细小的零件、重型冲压设备、长长的供应链。比亚迪有庞大的工厂和供应链，可以让机器人在真实环境中接受检验。 [Sarah]: I see. So it is not a lab. It is a noisy, busy, real car factory. If a robot can work there, it can work anywhere. zh:明白了。所以不是实验室，而是嘈杂、忙碌的真实汽车工厂。如果机器人能在那里工作，就能在任何地方工作。 [Mike]: Right. BYD calls it a real industrial testing ground. And there is a second gift: data. Every grip, every slip, every push from the factory will help train the AI model. zh:对。比亚迪称之为真实的工业验证土壤。还有第二份礼物：数据。工厂里的每一次抓取、每一次打滑、每一次推压，都会用来训练AI大模型。 [Sarah]: So the factory teaches the robot brain. More real data means the brain gets smarter and more general. zh:所以工厂在教机器人大脑。越多真实数据，大脑就越聪明、越通用。 [Mike]: You got it. That is how BYD will help improve the model's general ability and decision making. For robots to work at large scale, they need this training. zh:你说对了。这就是比亚迪帮助提升模型泛化能力和决策能力的方式。要让机器人实现规模化落地，就需要这样的训练。 [Sarah]: And what does PaXini bring? The touch? zh:那帕西尼带来什么呢？触觉吗？ [Mike]: Yes. PaXini brings its own technology for tactile sensors, multi-dimensional sensing, and full-modal embodied data collection. In simple words, they give robots nerve endings. zh:对。帕西尼带来自主的触觉传感器、多维感知和全模态具身数据采集技术。简单说，他们给机器人装上了神经末梢。 [Sarah]: Nerve endings — I like that image. So the robot can finally feel pressure, texture, even a tiny vibration? zh:神经末梢——我喜欢这个比喻。所以机器人终于能感受到压力、纹理，甚至轻微的震动了？ [Mike]: Exactly. Think of a human hand. You can close your eyes and still feel if an egg is raw or cooked, or if a screw is tight enough. PaXini wants robots to have that same feeling. zh:没错。想想人的手。你可以闭上眼睛，依然能感觉到鸡蛋是生的还是熟的，或是螺丝有没有拧紧。帕西尼想让机器人拥有同样的感觉。 [Sarah]: That is a game changer. Without touch, a robot just squeezes with fixed power. With touch, it can be gentle or firm when needed. zh:这可是颠覆性的。没有触觉，机器人只能用固定力度去挤压。有了触觉，它就能在需要时或轻或重地调整。 [Mike]: Right. And there is an interesting backstory. BYD was already an investor in PaXini. Now they move from investor to strategic partner. That is a big step up. zh:对。还有个有趣的背景。比亚迪本来就是帕西尼的股东。现在从股东变成了战略合作伙伴。这是一次大升级。 [Sarah]: So first they said, we believe in you, here is money. Now they say, let's build together, here is our factory. zh:所以一开始他们说，我们相信你们，给你们资金。现在他们说，我们一起干，把我们的工厂也交给你。 [Mike]: Exactly. It shows a clear path. A top electric car company is now moving deep into embodied AI and core robot sensing. The line between car maker and robot maker is getting blurry. zh:完全正确。这展现出清晰的路径。一家头部电动车企业正深入具身智能和机器人核心感知环节。车企和机器人公司之间的界限正变得模糊。 [Sarah]: That is not just one company's story. If it works, it could be a new sample for all of China's smart manufacturing. zh:这不只是一家公司的故事。如果成功，它可能成为中国智能制造升级的新样本。 [Mike]: The article said exactly that. As robot technology grows, this deal could become a new example of how Chinese factories upgrade with AI. zh:文章就是这么说的。随着机器人技术进步，这次合作有望为中国智能制造升级提供新的样本。 [Sarah]: I am thinking about a simple example. In a BYD factory, a robot has to pick up a soft rubber tube. Without touch, it might crush it. With PaXini's sensors, it feels the softness and holds it just right. zh:我想到一个简单的例子。在比亚迪工厂，机器人要拿起一根软橡胶管。没有触觉，它可能会压扁管子。有了帕西尼的传感器，它能感受到柔软度，力度刚刚好地拿住。 [Mike]: Perfect example. Or tightening a bolt inside a car door. Eyes cannot see inside. But fingertips can feel when it is tight. zh:太好的例子了。或者在车门里拧螺栓。眼睛看不见里面，但指尖能感觉到什么时候拧紧了。 [Sarah]: Okay, but here is my funny thought. If robots can finally feel, will they start complaining? Like, boss, this metal sheet is too hot! zh:好吧，但我有个好笑的想法。如果机器人终于会感知的，会不会开始抱怨？比如，老板，这块钢板太烫了！ [Mike]: Haha, maybe! And maybe that is the real test. When a robot says, hey, go easy, this part is fragile — then you know it truly has touch. zh:哈哈，也许吧！也许那才是真正的考验。当机器人说，嘿，轻点，这个零件很脆——你就知道它真的拥有触觉了。 [Sarah]: I love that picture. A former car company teaching a robot hand to feel a car part. It feels like BYD is not just building cars anymore. It is building hands. zh:我喜欢这个画面。一家曾经的车企正在教机器人之手去感知汽车零件。感觉比亚迪不再只是造车了，它在造一双手。 [Mike]: And hands that learn inside a real factory, with real data, get really good, really fast. That is the bet both companies are making. zh:而在真实工厂里、用真实数据学习的手，会非常快地变得非常灵巧。这就是两家公司共同的赌注。 [Sarah]: So next time we see a BYD car on the road, we can think, the same tech that built the car might soon build the robots that build the next car. zh:所以下次在路上看到比亚迪的车，我们可以想，造这辆车的技术，很快可能就会用来打造下一代造车的机器人。 [Mike]: Nice loop. Thanks for joining us on Learn English with Podcasts today! If you liked this story about robots that can feel, share it with a friend who still thinks robots are just cold metal arms. zh:漂亮的闭环。感谢今天收听 Learn English with Podcasts！如果你喜欢这个关于会感知的机器人的故事，就分享给还觉得机器人只是冰冷金属手臂的朋友吧。 [Sarah]: And tell us, what job would you give a robot that can truly feel? See you next time! zh:也告诉我们，你会给一个真正会感知的机器人安排什么工作？下期见！

Episode: BYD Hits 10,000 Stations: Flash Charging Everywhere Duration: approximately 9 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to "Learn English with Podcasts"! Sarah, I have a question for you. If I told you there is a company with ten thousand charging stations across China, would you believe me? zh:欢迎回到"Learn English with Podcasts"！Sarah，我有个问题问你。如果我告诉你有一家公司在中国建了一万个充电站，你会相信吗？ [Sarah]: Ten thousand? That sounds like a lot. Is that even possible? zh:一万个？听起来很多。这真的可能吗？ [Mike]: It is real. BYD just hit ten thousand flash charging stations. They reached this number on August 28th, 2026. zh:这是真的。比亚迪刚刚达到了一万个闪充站。他们在2026年8月28日达到了这个数字。 [Sarah]: Wow. And this is not their first big news about flash charging, right? We talked about the German award before. zh:哇。这不是他们关于闪充的第一个大新闻了，对吧？我们之前聊过那个德国奖项。 [Mike]: Right. Episode 32 covered the Paul Pietsch Award. But today is different. This time, it is about pure scale. Ten thousand stations. Three national records at once. zh:对。第32期聊过保罗·皮奇奖。但今天不一样。这次是纯粹的规模。一万个充电站。一次性创下三项全国纪录。 [Sarah]: Three records? Tell me more. zh:三项纪录？说来听听。 [Mike]: First, BYD now has the most charging stations built by any Chinese car company. Second, they were the fastest to reach ten thousand. And third, their network covers more regions than any other. zh:第一，比亚迪现在拥有中国车企自建数量最多的充电站。第二，他们是最快达到一万个的。第三，他们的网络覆盖的地域最广。 [Sarah]: Okay, that is impressive. But ten thousand stations across China, what does that actually mean for drivers? Can you really find one easily? zh:好吧，那确实很厉害。但一万个充电站分布在中国各地，对司机来说到底意味着什么？真的能方便地找到吗？ [Mike]: Great question. They cover 332 cities. In first and second tier cities, there is one hundred percent coverage. In cities like Shenzhen, Shanghai, and Hangzhou, there is a flash charging station every three kilometers. zh:问得好。他们覆盖了332个城市。在一二线城市，覆盖率是百分之百。在深圳、上海、杭州这样的城市，每三公里就有一个闪充站。 [Sarah]: Every three kilometers. That is like having a convenience store on every block. You never have to worry about finding one. zh:每三公里。那就像是每个街区都有便利店。你根本不用担心找不到。 [Mike]: Exactly. And even in smaller cities, tier three and four, coverage is also one hundred percent. Stations are within four kilometers in urban areas. zh:没错。即使是三四线小城市，覆盖率也是百分之百。城区内充电站都在四公里以内。 [Sarah]: What about rural areas? Or really far places? Like, can you charge in the middle of nowhere? zh:那农村地区呢？或者特别偏远的地方？比如，你在荒郊野外也能充电吗？ [Mike]: That is the crazy part. BYD stations are everywhere. From Hulunbuir in Inner Mongolia in the north, all the way to Sanya in Hainan in the south. From Shuangyashan in Heilongjiang in the east, to Kashgar in Xinjiang in the west. zh:最疯狂的部分来了。比亚迪充电站无处不在。北至内蒙古呼伦贝尔，南到海南三亚。东达黑龙江双鸭山，西到新疆喀什。 [Sarah]: Wait, even in Kashgar? That is on the border. And what about high mountains? Or really hot places? zh:等等，连喀什都有？那可是边境地区。那高山呢？或者特别热的地方呢？ [Mike]: They have stations at Everest, the highest mountain in the world. In Turpan, one of the hottest places in China. Even in Horgos, the farthest border crossing. zh:他们在珠穆朗玛峰——世界最高的山——都有充电站。在吐鲁番——中国最热的地方之一。甚至在霍尔果斯——最远的边境口岸。 [Sarah]: That is unbelievable. You can drive from the coldest part of China to the hottest part, and BYD has you covered the whole way. zh:太不可思议了。你可以从中国最冷的地方开到最热的地方，比亚迪全程为你保驾护航。 [Mike]: And the speed is real too. They say five minutes to get a good charge, nine minutes for a full charge. Even in minus thirty degrees, it only takes three extra minutes. zh:充电速度也是实打实的。他们说5分钟就能充好，9分钟充满。即使在零下30度，也只多花3分钟。 [Sarah]: Three extra minutes in freezing cold? That is barely noticeable. That solves the cold weather problem. zh:在极寒天气下只多3分钟？几乎感觉不到。这解决了低温充电的问题。 [Mike]: It does. And here is a fun fact. In just six months, BYD stations charged over 210 million kilowatt hours of electricity. That is a huge number. zh:确实如此。这里有个有趣的数据。仅在6个月内，比亚迪充电站充电量超过2.1亿度电。这是个巨大的数字。 [Sarah]: Two hundred ten million. Okay, but here is what I really want to know. Who is actually using these stations? Just BYD drivers? zh:2.1亿度。好吧，但我真正想知道的是，谁在用这些充电站？只有比亚迪的车主吗？ [Mike]: That is the best part. Total users have passed 1.83 million. And nearly one third of them are other brands. Not BYD. zh:这是最精彩的部分。总用户量突破183万。其中接近三分之一是其他品牌车主。不是比亚迪的车。 [Sarah]: Wait, so people who do not even own a BYD are using their charging stations? zh:等等，所以连不开比亚迪的人都在用他们的充电站？ [Mike]: Yes. BYD is not just building for themselves. They are building for everyone. That is why they call it "good technology for all." zh:对。比亚迪不只是为自己建。他们是为所有人建。所以他们称之为"好技术人人可享"。 [Sarah]: That is smart business. You build the network, and every electric car driver becomes your customer. Even if they bought a different brand. zh:这是聪明的商业策略。你建好网络，每个电动车司机都成了你的客户。即使他们买了别的品牌。 [Mike]: And they are not doing it alone. BYD has partnered with Sinopec, PetroChina, Shell, and others. They are building stations together at gas stations and shopping centers. zh:而且他们不是单打独斗。BYD与中石化、中石油、壳牌等达成了合作。他们在加油站和购物中心一起建充电站。 [Sarah]: That makes sense. Gas stations already have the space and the traffic. Turning them into charging hubs is a natural move. zh:有道理。加油站本来就有空间和车流量。把它们变成充电枢纽是自然而然的事。 [Mike]: They also work with Alipay and Gaode Maps for easy payment and navigation. So you can find a station, charge, and pay without any hassle. zh:他们还和支付宝、高德地图合作，实现便捷支付和导航。所以你可以找到充电站、充电、付款，全程无忧。 [Sarah]: You know what this reminds me of? It reminds me of when smartphones first came out. Everyone needed a charging network. BYD is building that network for electric cars. zh:你知道这让我想到什么吗？让我想到智能手机刚出来的时候。每个人都需要充电网络。比亚迪正在为电动车建那个网络。 [Mike]: That is a great comparison. And BYD is not stopping at ten thousand. They plan to have twenty thousand stations in China by the end of this year. zh:这个比喻很棒。而且比亚迪不会停在一万个。他们计划今年年底在中国建到两万个充电站。 [Sarah]: Twenty thousand? That is doubling in just a few months. zh:两万个？那几个月内就翻倍了。 [Mike]: And overseas, they are building six thousand more. Stations are already running in Europe and the Americas. zh:在海外，他们还要再建六千个。欧洲和美洲已经有充电站在运营了。 [Sarah]: So this is not just a China story anymore. It is going global. zh:所以这不再只是中国的故事了。它正在走向全球。 [Mike]: BYD has three world firsts now. First to stop making fuel cars. First to produce ten million new energy vehicles. And first to use solar, storage, and charging all in one system. zh:BYD现在有三个全球第一。第一个宣布停产燃油车。第一个实现千万级新能源整车下线。第一个规模化落地光储充一体技术。 [Sarah]: You know what I find funny? A few years ago, people thought Chinese electric cars were cheap copies. Now BYD is building the infrastructure that the whole world needs. zh:你知道我觉得什么好笑吗？几年前，人们觉得中国电动车是便宜的仿制品。现在BYD正在建设全世界都需要的基础设施。 [Mike]: Things change fast. And that is the story of ten thousand stations. It is not just about charging. It is about building trust. zh:变化真快。这就是一万个充电站的故事。不只是充电。而是建立信任。 [Sarah]: Trust that you can drive anywhere, charge anywhere, and never get stuck. zh:相信你可以开到任何地方、在任何地方充电、永远不会被困住。 [Mike]: Next time you see a BYD charging station, remember. There are nine thousand nine hundred ninety nine more somewhere in China. zh:下次你看到比亚迪充电站，记住。在中国的某个地方，还有九千九百九十九个。 [Sarah]: And soon there will be twenty thousand. That is all for today on "Learn English with Podcasts"! See you next time. zh:很快就会有两万个了。今天的"Learn English with Podcasts"就到这里！下期见。

Episode: Hunyuan Hy4 Preview: Open-Source Top Tier Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, imagine an AI that could do almost your whole workday - write code, check your invoices, even build a small video game - and this week you can try it for free. zh:欢迎回到 Learn English with Podcasts！Sarah，想象一个 AI 几乎能搞定你一整天的工作——写代码、核对发票、甚至做个小游戏——而且这周还能免费试用。 [Sarah]: Free? That sounds too good. Which AI are we talking about? zh:免费？听着太好了吧。我们说的是哪个 AI？ [Mike]: Tencent just released and open-sourced a new model called Hunyuan Hy4 preview. And they say it is now among the very best open-source models in the world. zh:腾讯刚发布并开源了一个新模型，叫混元 Hy4 preview。而且他们说，它现在已经是全球最好的开源模型之一。 [Sarah]: Wait, Tencent? The company behind WeChat? They made a top AI too? zh:等等，腾讯？微信那家公司？他们也做出了顶尖 AI？ [Mike]: Yes. And here is the first big number: the model has 770 billion total parameters, but only 49 billion are active each time you use it. zh:是的。第一个大数字是：模型总参数 7700 亿，但你每次使用时只有 490 亿被激活。 [Sarah]: Only 49 of 770? So it is like a huge company where most people are on standby, and a small team does the actual job? zh:770 亿里只有 49 亿？那它像一家巨型企业，大部分人待命，只有一小队人真正干活？ [Mike]: Perfect analogy. That design is called Mixture of Experts. You keep a giant library of knowledge, but only wake up the parts you need. It saves cost and speed. zh:这个比喻太贴切了。这种设计叫专家混合。你保留一个巨大的知识库，但只唤醒需要的那部分。省成本又提速。 [Sarah]: Smart. And what else is impressive? zh:聪明。还有什么厉害的？ [Mike]: The memory. It can read up to 1 million tokens in one go. That is roughly a whole book, or your entire year of chat logs. zh:它的记忆力。一次能读多达 100 万 Token。差不多是一整本书，或者你一整年的聊天记录。 [Sarah]: A whole book at once? My notebook would be full after two pages. zh:一次读完一整本书？我的笔记本两页就满了。 [Mike]: And because it is open-source, developers can download the weights, study how it works, and run it on their own servers. zh:而且因为它开源，开发者可以下载权重、研究它的原理，还能部署在自己的服务器上。 [Sarah]: So it is not locked inside one app. The whole world can build on it? zh:所以它不被锁在某个 App 里。全世界都能在它基础上再创造？ [Mike]: That is what open-source means. Tencent shared it on Hugging Face, GitHub, and ModelScope. zh:这就是开源的含义。腾讯把它放到了 Hugging Face、GitHub 和 ModelScope 上。 [Sarah]: Okay, but is it actually good, or just open? zh:好，但它真有本事，还是只是开源而已？ [Mike]: Tencent built it for work, not small talk. They tested it in a blind test with 163 internal experts across 203 engineering tasks. zh:腾讯是冲着干活造它的，不是闲聊。他们做了盲测，请了 163 位内部专家，覆盖 203 个工程任务。 [Sarah]: Blind test meaning the experts did not know which model they were grading? zh:盲测就是说专家不知道自己在评哪个模型？ [Mike]: Exactly. And Hy4 preview scored 2.99 out of 4, just above GLM 5.3 at 2.92 and Kimi K3 at 2.94. zh:没错。Hy4 preview 拿到 4 分里的 2.99，略高于 GLM 5.3 的 2.92 和 Kimi K3 的 2.94。 [Sarah]: So it barely edged them out. Close race at the top. zh:所以它只是险胜。顶尖这场很胶着。 [Mike]: Very close. Now for real examples. In office work, it read 72 invoices and three company rule books, then decided which reimbursements were allowed. zh:非常接近。说点实际例子。办公场景里，它读了 72 张发票和三本公司规章，然后判断哪些报销能过。 [Sarah]: Three rule books? That alone would take me an afternoon. zh:三本规章？光这个我就得耗一个下午。 [Mike]: And in game development, it connected to the Unreal 5 engine through something called MCP, and built a playable shooting game just by talking to it. zh:游戏开发里，它通过一个叫 MCP 的东西接入了 Unreal 5 引擎，光靠对话就做出了一款能玩射击游戏。 [Sarah]: A whole game from conversation? No coding by hand? zh:整款游戏靠对话生成？不用手写代码？ [Mike]: That is the claim. The developer just kept chatting to improve it. And in science, it ran a molecular simulation of 32,512 atoms at 54.9 milliseconds per step. zh:这是它的说法。开发者只要不断对话来完善。科学方面，它跑了一个 32,512 个原子的分子模拟，每步 54.9 毫秒。 [Sarah]: Each step takes just a few milliseconds? That sounds fast for something so small. zh:每步只要几毫秒？那么小的体系还这么快？ [Mike]: They say it is 2.0 times faster than before, and one top GPU could hold 300,000 atoms. Useful for drug and material research. zh:他们说比以前快 2.0 倍，一张高端 GPU 能装下 30 万个原子。对药物和材料研究很有用。 [Sarah]: Okay, but here is the part that gave me chills. This model helped improve itself? zh:好，但最让我起鸡皮疙瘩的是这点。这个模型还帮着自己改进自己？ [Mike]: Yes. Hy4 preview joined its own development - it suggested training methods, data plans, even optimized the computer code that runs it. End-to-end speed went up 31.8 percent. zh:是的。Hy4 preview 参与了自己的研发——它提议训练方法、数据方案，甚至优化了运行它的底层代码。端到端速度提升了 31.8%。 [Sarah]: A model that tunes its own engine. That is like a car that redesigns its own engine while driving. zh:一个会调自己引擎的模型。就像一辆车一边开一边重新设计自己的引擎。 [Mike]: And Tencent has been rebuilding its AI infrastructure since February, shipping a big update about every two months. They release a preview first, then a final version. zh:而且腾讯从 2 月起重建了 AI 基础设施，大约每两个月发一个大版本。他们先发预览版，再发正式版。 [Sarah]: So the preview is like a beta you get to use early? zh:所以预览版就像你提前用上的测试版？ [Mike]: Exactly. And to collect feedback, WorkBuddy and CodeBuddy are free to try for two weeks. If you want, you can build something with it today. zh:没错。为了收集反馈，WorkBuddy 和 CodeBuddy 限时两周免费。想试的话，今天就能用它做个东西。 [Sarah]: I like that. Listeners, if you tried Hy4 preview, what would you build first - a game, a report, or your own little assistant? zh:这个我喜欢。听众们，如果你试了 Hy4 preview，会先做什么——游戏、报告，还是你自己的小助手？ [Mike]: Thanks for listening to Learn English with Podcasts. The next time you hear a model is big, ask not how big, but how much of it actually shows up to work. zh:感谢收听 Learn English with Podcasts。下次听说某个模型很大时，别问它多大，问问它到底有多少真的来上班了。 [Sarah]: See you next time! zh:下次见！

Episode: Zhipu Tang Jie: Scaling Beyond Parameters Duration: approximately 8 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, if a company told you their new AI has 1 trillion parameters, would you say wow, it must be the best? zh:欢迎回到 Learn English with Podcasts！Sarah，如果一家公司告诉你他们的新 AI 有 1 万亿参数，你会说哇，它一定是最好的吗？ [Sarah]: A few years ago, I would say yes. Bigger means smarter, right? Like a bigger brain? zh:几年前我会说是的。更大意味着更聪明，对吧？就像更大的大脑？ [Mike]: That is exactly what Zhipu founder Tang Jie wants to challenge. He just posted his Thoughts About Scaling Law, and his first line is simple - Scaling, but not just parameters. zh:这正是智谱创始人唐杰想挑战的观念。他刚发布了关于 Scaling Law 的思考，开头就很简单——要 Scaling，但不只是参数。 [Sarah]: So scaling is still true, but we are scaling the wrong thing? zh:所以 Scaling 依然成立，但我们一直在扩大错误的东西？ [Mike]: We are scaling an incomplete picture. Tang Jie says reporting only the parameter number tells you almost nothing today. You also need to ask: how much data, where you spend your compute, and how the model will run every day. zh:我们看到的是不完整的图景。唐杰说，今天只报参数量几乎说明不了什么。你还得问：用了多少数据、把算力花在哪，以及模型每天是怎么运行的。 [Sarah]: Okay, give me the history. When did we start thinking bigger is better? zh:好，给我讲讲历史。我们什么时候开始觉得越大越好？ [Mike]: It started in 2020. Researchers led by Kaplan tested many model sizes and found a rule. When you have more compute, you should grow parameters faster than data. About 2.7 to 1. zh:从 2020 年开始。Kaplan 带领的研究者测试了很多模型尺寸，发现一条规律：算力更多时，参数应该比数据涨得更快，大约是 2.7 比 1。 [Sarah]: And the industry listened? zh:然后行业就照做了？ [Mike]: Completely. That rule set the race. GPT-3 came with 175B parameters. Then Gopher, then MT-NLG with 530B. Everyone started chasing 1 trillion. zh:完全照做。这条规则定下了竞赛方向。GPT-3 带着 175B 参数登场，然后是 Gopher，再到 530B 的 MT-NLG。大家都开始追逐 1 万亿。 [Sarah]: I remember those headlines. Every launch was look, we are bigger! zh:我记得那些标题。每次发布都是看，我们更大了！ [Mike]: And then in 2022, DeepMind did the test again. But bigger. They trained over 400 models, from 70M to 16B parameters, on 5B to 500B tokens. zh:然后在 2022 年，DeepMind 又把这个实验重做了一遍。但规模更大。他们训练了超过 400 个模型，参数从 70M 到 16B，数据从 50 亿到 5000 亿 Token。 [Sarah]: 400 models just to check the math? That is serious. zh:为了验算数学就训了 400 个模型？真拼。 [Mike]: And they found we were wasting resources. Models had too many parameters and not enough data. So they built Chinchilla. zh:然后他们发现我们一直在浪费资源。模型参数塞得太多，吃进去的数据却不够。所以他们造了 Chinchilla。 [Sarah]: Little Chinchilla versus big Gopher? I know this story. zh:小小的 Chinchilla 对大大的 Gopher？我知道这个故事。 [Mike]: Exactly. Chinchilla is only 70B parameters, 4 times smaller than Gopher at 280B. But it was trained on 1.4T tokens, 4 times more data. With the same training compute, the smaller, better-fed animal won. It beat Gopher, GPT-3, Jurassic-1, and MT-NLG. zh:没错。Chinchilla 只有 70B 参数，比 280B 的 Gopher 小 4 倍。但它用了 1.4T Token 训练，多 4 倍数据。用同样的训练算力，更小但吃得更饱的选手赢了。它击败了 Gopher、GPT-3、Jurassic-1 和 MT-NLG。 [Sarah]: So the new rule became 20 tokens per parameter, and params and data should grow together? zh:所以新规则就成了每个参数对应 20 个 Token，参数和数据应该一起长？ [Mike]: Yes. Chinchilla Scaling Law. But Tang Jie says even that was not the final answer. zh:对，这就是 Chinchilla Scaling Law。但唐杰说，那也不是最终答案。 [Sarah]: Why? Did we find another missing piece? zh:为什么？我们又发现了缺失的一块？ [Mike]: Because training is only done once, but running the model happens billions of times a day. A 1 trillion parameter model needs huge compute every single time you ask it something. Chinchilla did not count that daily bill. zh:因为训练只做一次，但运行模型每天要发生数十亿次。1 万亿参数的模型每次你提问都要消耗巨大算力。Chinchilla 没算这笔日常账单。 [Sarah]: Ah, the inference cost. The price after you buy the car. zh:啊，推理成本。买车之后的使用成本。 [Mike]: Perfect metaphor. Some studies looked at the full life. If a model will handle about 1B requests, a smaller model trained longer can be cheaper overall. They pushed training to 10,000 tokens per parameter and quality still kept improving. zh:这个比喻完美。一些研究看了全生命周期。如果一个模型要处理约 10 亿次请求，一个更小但训练更久的模型总成本可能更低。他们把训练强度一路推到每个参数 1 万个 Token，质量还在提升。 [Sarah]: Can you give me real numbers? zh:能给我真实数字吗？ [Mike]: Sure. Llama 2 7B was trained on about 290 tokens per parameter. Gemma 2 9B reached about 889 tokens per parameter. Way beyond the old 20. zh:可以。Llama 2 7B 每个参数约对应 290 个训练 Token，Gemma 2 9B 更达到约 889 个。远超当年的 20。 [Sarah]: So from fix the size, to feed it more data, to count the daily running cost. The best place to spend compute keeps moving. zh:所以从扩大参数，到补足数据，再到计算日常运行成本。最值得花钱的地方一直在变。 [Mike]: Exactly. And for today's MoE models, it moves again. Tang Jie splits two ideas: total params and active params. zh:没错。而对于今天的 MoE 模型，它又变了。唐杰把两个概念分开：总参数和激活参数。 [Sarah]: Like a warehouse and the workers inside? zh:像仓库和里面的工人？ [Mike]: He uses a similar picture. Total params is the warehouse - how much knowledge you can store. Active params and effective depth is how much power you use in one go, how far you can think in one chain. zh:他用了类似的比喻。总参数是仓库——能装多少知识。激活参数和有效深度是单次能调动多少能力，能把一条推理链走多远。 [Sarah]: So you need both. Lots of books is not enough, you must read them well. zh:所以两者都需要。藏书多不够，还得会读。 [Mike]: And bug hunting shows the difference. Remembering many CVEs means you saw many cases. But to find a new bug, you must follow the code for 20 steps - from strange code to how it breaks, to how to test it - without losing the thread. zh:而抓漏洞最能看出区别。记住很多 CVE 意味着你见过很多案例。但要发现一个新漏洞，你得把代码跟上 20 步——从异常代码到触发条件再到验证方法——全程不能掉线。 [Sarah]: Twenty steps without getting lost. That is not memory, that is focus. zh:20 步不掉线。那不是记忆力，是专注力。 [Mike]: So Tang Jie says you cannot judge that long-chain skill by total size alone. And that brings us to GLM-5.3. zh:所以唐杰说，你不能只用总规模去判断这种长链能力。这就说到了 GLM-5.3。 [Sarah]: The new Zhipu model everyone is talking about? zh:就是大家在聊的智谱新模型？ [Mike]: Yes. For GLM-5.3, Zhipu kept the base, the architecture, total params and active params all the same as GLM-5.2. No bigger warehouse. For the past month, they put new compute into long-horizon environments and reinforcement learning. zh:是的。做 GLM-5.3 时，智谱保留了 GLM-5.2 的基座和架构，总参数和激活参数都没变。没有把仓库变大。过去一个月，他们把新增算力都投进了长程任务环境和强化学习。 [Sarah]: So same body, better training after school? zh:所以身子没变，放学后加练？ [Mike]: Great way to say it. The gains came from post-training. Official numbers: Terminal-Bench 3.0 went from 4.6 to 28.3, DeepSWE from 46.2 to 66.9, and Agents' Last Exam from 23.8 to 28.5. zh:说得太好了。提升来自后训练。官方数字：Terminal-Bench 3.0 从 4.6 升到 28.3，DeepSWE 从 46.2 涨到 66.9，Agents' Last Exam 也从 23.8 升到 28.5。 [Sarah]: That jump on Terminal-Bench is huge! zh:Terminal-Bench 那个涨幅也太大了！ [Mike]: It is. Now, these are official tests, outside teams will need to check again when weights are open. But for Tang Jie, it proves a point - scaling has many knobs. zh:确实很大。当然这些目前主要是官方测试，外部复现还得等权重开放。但对唐杰来说，这已经证明一点——Scaling 有很多旋钮。 [Sarah]: Many knobs, you do not have to turn them all at once. zh:很多旋钮，不用一次全拧。 [Mike]: Exactly. GLM-5.3 turned the post-training knob because that knob had the most room left. Next time, Zhipu might turn pre-training or mid-training or compute per forward pass. zh:没错。GLM-5.3 拧了后训练这个旋钮，因为当时这里剩余空间最大。下一次，智谱可能再去拧预训练、中训练或单次前向算力。 [Sarah]: So the lesson is not bigger is dead, but single-number bigger is dead. zh:所以教训不是越大越没用，而是只看一个数字的越大已经过时了。 [Mike]: Beautiful summary. The game is moving from who has the biggest library to who can finish a real job - work for hours or days like a digital employee, without dropping the ball. zh:总结得漂亮。竞赛正从谁的藏书最大，转向谁能把一份真实工作做完——像数字员工那样连续工作几小时、几天，不掉线、不跑偏。 [Sarah]: If the model only answers one question, size still sounds cool. If it must work all day for you, you care if it can stay on track. zh:如果模型只答一道题，参数大小听起来还很酷。如果它得为你干一整天，你更关心它能不能一直靠谱。 [Mike]: And Tang Jie closes with a calm line: 1 trillion params has not disappeared. But from now on, it may not be the only number worth putting in the headline. zh:唐杰最后也平静地收了尾：万亿参数没有消失。只是从现在开始，它可能不再是唯一值得写进标题的那个数字。 [Sarah]: That hits me. I used to check the number like a price tag. Now I will ask, and how long can it actually work? zh:这句话点到我了。我以前像看价签一样看参数。现在我会问，它到底能干多久？ [Mike]: And that is the quiet joke of this story. For three years we counted params like height. Turns out AI is not a basketball team. The tallest player does not always win the marathon. zh:这也是这个故事里安静的幽默。三年里我们像量身高一样数参数。结果发现 AI 不是篮球队，最高的不一定赢得了马拉松。 [Sarah]: I love that. Okay, listeners, what would you check now before you pick a model - the warehouse size or how far it can run? Tell us! zh:太喜欢这个比喻了。好，听众们，现在让你选模型，你会先看仓库大小，还是看它能跑多远？告诉我们吧！ [Mike]: Thanks for listening to Learn English with Podcasts. Try asking your next AI not how big it is, but how many steps it can stay with you. zh:感谢收听 Learn English with Podcasts。下次问你的 AI 时，别问它多大，问问它能陪你走多少步。 [Sarah]: See you next time! zh:下次见！

Episode: GLM-5.3 Flash: Same Brain, 1/40 the Price Duration: approximately 9 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, quick riddle - what costs 40 times less but thinks just as well? zh:欢迎回到 Learn English with Podcasts！Sarah，猜个谜语——什么东西便宜 40 倍，脑子却一样好使？ [Sarah]: Is this a sale at the supermarket? Or a new phone plan? zh:是超市大促销？还是新的手机套餐？ [Mike]: Better. It is a brand-new AI from China. Zhipu AI just released and open-sourced GLM-5.3 Flash. 320B total size, but only 18B active at a time. And its brain is frontier-level. zh:更厉害。是一颗来自中国的全新 AI。智谱刚发布并开源了 GLM-5.3 Flash。总参数 320B，但每次只激活 18B。脑子却是前沿水平。 [Sarah]: 320B-A18B - that sounds like a giant library where you only open one shelf at a time. zh:320B-A18B——听起来像一个巨大图书馆，但每次只打开其中一排书架。 [Mike]: Perfect picture. And here is the number that made everyone stop scrolling. On the Artificial Analysis Intelligence Index, a very respected global test, it scored 57. zh:这个比喻完美。而让所有人停下刷手机的数字是这个：在非常权威的全球测试 Artificial Analysis Intelligence Index 上，它得了 57 分。 [Sarah]: 57 means... good? Great? Help me feel it. zh:57 分算……好？非常好？让我感受一下。 [Mike]: 57 is frontier. Exactly the same as Anthropic's most popular model, Claude Opus 4.8. On Z.ai's own Code Bench, which tests real coding feel, it also ties with Opus 4.8. zh:57 就是前沿。和 Anthropic 最受欢迎的模型 Claude Opus 4.8 完全同分。在 Z.ai 自己的 Code Bench 体感测试里，编程表现也和 Opus 4.8 打平。 [Sarah]: Same brain, then. So what is the price part of your riddle? zh:那就是脑子一样。所以你谜语里的价格部分呢？ [Mike]: GLM-5.3 Flash costs 1/10 of the full GLM-5.3. With the current limited-time discount, it is 1/20. And compared to Opus 4.8, it is 1/40. zh:GLM-5.3 Flash 的定价是完整版 GLM-5.3 的 1/10。现在限时折扣内是 1/20。和 Opus 4.8 比，是 1/40。 [Sarah]: Wait, let me do the math. Same 57 points, 1/40 the price? That is not a discount, that is a different universe. zh:等等，我算一下。同样的 57 分，1/40 的价格？这不是打折，这是换了个宇宙。 [Mike]: Their slogan says it: same intelligence, 1/40 the price. Frontier AI, finally cheap enough that you do not have to save it. zh:他们的口号就是这么说的：同样智力，1/40 价格。前沿智能，第一次不用省着用。 [Sarah]: Okay, I love a good slogan, but how do we know it is real? Companies can claim any number. zh:口号我喜欢，但怎么知道是真的？公司可以随便说数字。 [Mike]: They did a blind test. Before the launch, they released it anonymously as Ox-Alpha - in Chinese, the community called it Niu Lai. On OpenCode and OpenRouter, no name, no logo. zh:他们做了盲测。发布前，他们匿名把它放出来，代号 Ox-Alpha——中文社区叫它牛来。在 OpenCode 和 OpenRouter 上，无名字无 Logo。 [Sarah]: And people just... started using it without knowing what it was? zh:然后大家就……在不知道它是谁的情况下开始用？ [Mike]: And loved it. It quickly became the most popular model of the week and broke call records on both platforms. Only later they said, surprise, that was us. zh:而且很喜欢。它很快成了当周最受欢迎的模型，创下双平台调用量纪录。后来他们才说，惊喜一下，那就是我们。 [Sarah]: Okay, that is confident. But you said same frontier brain for less money. Where does the savings come from? Smaller model? zh:好，这很自信。但你说同样的前沿脑子花更少的钱。省钱从哪来？把模型变小？ [Mike]: The opposite trick. Total size is similar to the older GLM-4.5 - 320B vs 355B. But look inside. Active parameters dropped from 32B to 18B. Layers nearly halved, from 92 to 45. All trained on a new 30T token multimodal dataset. zh:恰恰相反的技巧。总参数和上一代 GLM-4.5 差不多——320B 对 355B。但看里面。激活参数从 32B 降到 18B。层数几乎减半，从 92 层到 45 层。全部用全新的 30T token 多模态数据训练。 [Sarah]: Fewer active parts, fewer layers, but stronger result? That sounds like losing weight and getting stronger at the same time. zh:激活更少、层数更少，结果却更强？这像减重同时变壮，一起发生。 [Mike]: Exactly. And for long contexts, they built a new attention system. It mixes two ideas: linear attention for nearby words, and sparse attention for far-away context. zh:没错。而为了处理长上下文，他们造了一套新的注意力系统。混合两种思路：线性注意力管附近的词，稀疏注意力管远处的上下文。 [Sarah]: Like having both a flashlight for the page in front of you, and a search light for the whole library? zh:就像既有照亮眼前这一页的手电，又有扫完整座图书馆的探照灯？ [Mike]: Beautiful. To save even more, they added IndexPool. It squeezes 4 memory vectors into 1 for the 1M context index. Much less memory, much less delay. zh:太形象了。为了再省一点，他们加了 IndexPool。它把 1M 上下文索引里的 4 个缓存向量压成 1 个。内存和延迟都大幅下降。 [Sarah]: Numbers please. How much did they actually save? zh:给我数字。到底省了多少？ [Mike]: Compared to the full GLM-5.3, attention compute is down 3.01 times, and KV cache size is down 4.44 times per layer. Among all the models they compared, including DeepSeek-V4-Flash and Kimi-K3, this one has the lowest compute per token. zh:和完整版 GLM-5.3 比，每层注意力计算量降了 3.01 倍，KV 缓存大小降了 4.44 倍。在他们对比的所有模型里，包括 DeepSeek-V4-Flash 和 Kimi-K3，它的单 token 计算量最低。 [Sarah]: But you said lowest compute, so what is still not perfect? zh:但你说计算量最低，那还有哪里不完美？ [Mike]: Honest detail: its KV cache is still a bit bigger than Kimi-K3 and DeepSeek-V4-Flash. The team says that is their next job. zh:诚实的细节：它的 KV 缓存仍然比 Kimi-K3 和 DeepSeek-V4-Flash 略大。团队说那是下一步要优化的。 [Sarah]: I like that honesty. Now, you said this is the first native multimodal in the GLM-5 family. What does multimodal mean here? It can see? zh:我喜欢这种坦诚。对了，你说这是 GLM-5 系列首个原生多模态。说的多模态在这里是什么意思？它会看？ [Mike]: It can see, and it uses seeing to code better. They call it Visual Coding. For frontend, games, 3D, the final result is something you see and touch. So the model learns when to look at its own output and fix it. zh:它会看，而且用看来看得更会写代码。他们叫视觉编码。做前端、游戏、3D 时，最终产物是你能看到、能交互的东西。所以模型学会何时去看自己的输出并修正。 [Sarah]: So it is not just write code, run, hope it works. It is write, look, fix, like a human designer? zh:所以不是写完代码跑一下就祈祷成功。而是写、看、改，像人类设计师那样？ [Mike]: Exactly. They built a data pipeline where the model has to interact, check its own screen, and improve. With reinforcement learning from real user flows, it even judges if a button looks right, not just if the code runs. zh:没错。他们搭了一条数据流水线，让模型必须去交互、检视自己的画面、再迭代改进。加上基于真实用户流程的强化学习，它甚至能判断按钮好不好看，而不只是代码能不能跑。 [Sarah]: Give me a wild example. What did it actually build? zh:给我一个疯狂的例子。它到底做出了什么？ [Mike]: My favorite. With no outside images or models, GLM-5.3 Flash ran alone for 16 hours in Blender and built a 400-square-meter professional chef's home and test kitchen. Every piece of furniture, light, and material placed consistently from any angle. zh:我最喜欢这个。在没有任何外部素材的情况下，GLM-5.3 Flash 在 Blender 里独自跑了 16 个小时，搭出了一套约 400 平方米的专业主厨自宅和测试厨房。每一件家具、每一束光、每一种材质，从任何角度看都一致。 [Sarah]: 400 square meters, 16 hours, no help? That is my apartment times four, built while I slept. zh:400 平方米，16 小时，无协助？那是我家四倍大，在我睡觉时就盖好了。 [Mike]: And inside ZCode, it can now work across code, browser, and computer screen together. Many bugs only show up after you render or click around. zh:而且在 ZCode 里，它现在能在代码、浏览器和电脑界面之间协同工作。很多问题只有渲染出来、点一点才暴露。 [Sarah]: Okay, coding is impressive. But most people at work just want a good PPT, not a 3D kitchen. zh:编程很厉害，但大多数上班族只想要一份好看的 PPT，而不是 3D 厨房。 [Mike]: It does that too. With visual understanding, it can compare its own PPTX, PDF, DOCX, and XLSX output to what it expected, and fix the beauty. Plus they trained it for finance and law. zh:那个也行。靠视觉理解，它能把自己输出的 PPTX、PDF、DOCX 和 XLSX 与预期对比并优化美观。还专门为金融和法律做了训练。 [Sarah]: Finance and law? Like real reports? zh:金融和法律？像真正的报告那种？ [Mike]: Yes. In finance: from research with sources to report writing to modeling, all with traceable references. In law: it can review cost and liability clauses, mark them like a real lawyer, and draft letters and contracts ready to send. zh:对。金融那边：从带来源的研究、到报告生成、再到建模分析，全流程可追溯。法律这边：能审费用和责任条款，按律师习惯批注留痕，还能起草函件和合同，格式规范到可直接交付。 [Sarah]: Ready to send? No more my 2 a.m. formatting nightmare? zh:可直接交付？再也不用凌晨两点调格式了？ [Mike]: That is the promise. Now the last twist, and it is a big one. Remember Ox-Alpha broke records? All that traffic ran on domestic Chinese chips. zh:这就是承诺。还有最后一个大反转，也很大。还记得牛来创下纪录吗？那些流量全部跑在国产芯片上。 [Sarah]: Wait, all those calls were served by Chinese chips? Not the usual big brand? zh:等等，那些调用全是国产芯片扛住的？不是常用的大品牌？ [Mike]: A full domestic cluster, connected by their own high-speed network. Single chips have less memory and bandwidth, especially for 1M context. So they built a special engine on SGLang. zh:一整套国产芯片集群，用自研高速网络连起来。单张芯片内存和带宽相对有限，特别是要撑 1M 上下文。所以他们基于 SGLang 造了专用推理引擎。 [Sarah]: And who built that engine? A huge team working for months? zh:谁造的这个引擎？一支大团队干了几个月？ [Mike]: Here is the fun circle. A big part of the engine was built faster with help from an infra agent powered by GLM-5.3 itself. The model helped optimize the system that now runs the model. zh:好玩的闭环来了。引擎很大一部分是靠 GLM-5.3 自己驱动的 infra agent 加速搭起来的。模型帮忙优化了现在承载模型的系统。 [Sarah]: The model improved the system that carries the model. Like a student who builds a better desk to study faster? zh:模型改进了承载自己的系统。像学生为了学得更快先造了一张更好的书桌？ [Mike]: Exactly. They used tricks like intra-node tensor parallel for linear attention, ReplaySSM, W8A8 quantization, mixed INT8/FP8/BF16 cache, and Layer Split. At the cluster level, they split Encode, Prefill, and Decode into separate pools that scale alone. zh:没错。他们用了很多技巧：线性注意力的节点内张量并行、ReplaySSM、W8A8 量化、INT8/FP8/BF16 混合缓存、Layer Split。集群层面还把编码、预填充和解码拆成独立可扩缩的工作池。 [Sarah]: Okay, translate that to human: did it work? zh:好，翻译成人话：管用吗？ [Mike]: Compared to the first baseline on the same hardware, end-to-end speed went up 3 times. And cost per token is now similar to mainstream NVIDIA GPUs. zh:和同一硬件上的初始基线比，端到端性能提升了 3 倍。单 token 成本已和主流英伟达 GPU 相当。 [Sarah]: So domestic chips can now do frontier work at a similar price. That changes the story from who has the best chips to who uses them best. zh:所以国产芯片现在也能以差不多的价格扛前沿任务。故事就从谁有最好的芯片，变成了谁把芯片用得最好。 [Mike]: And it is open now. API on BigModel and Z.ai, chat on chat.z.ai, plus ZCode and AutoClaw. With GLM Coding Plan giving 10,000 free cards a day. zh:而且现在已开源开放。BigModel 和 Z.ai 上可调 API，chat.z.ai 可直接体验，还有 ZCode 和 AutoClaw。GLM Coding Plan 每天还发 10,000 张体验卡。 [Sarah]: You know what sticks with me? Not the 57, not the 1/40. It is the 16-hour kitchen. A model that can stare at its own work, say hmm, that light looks wrong, and fix it - that feels... human. zh:你知道最让我记住的是什么吗？不是 57 分，也不是 1/40。是那间搭了 16 小时的厨房。一个会盯着自己的作品说嗯这灯光不对然后改掉的模型——感觉……很人类。 [Mike]: And the quiet joke behind it all - weeks ago, thousands of people fell in love with a nameless cow. Niu Lai. They had no idea they were already living in the cheaper future. zh:而背后那个安静的玩笑是——几周前，成千上万人爱上了一头无名牛。牛来。他们完全没意识到，自己已经活在了更便宜的未来里。 [Sarah]: Same brain, tiny bill, and a cow that tricked everyone. I think our listeners will remember that. Thanks for listening to Learn English with Podcasts - see you next time! zh:同样的脑子，更小的账单，还有一头骗过所有人的牛。我想听众会记住这个。感谢收听 Learn English with Podcasts——下次见！ [Mike]: Try GLM-5.3 Flash and tell us what you would build - a kitchen, a deck, or your dream app. See you next time! zh:试试 GLM-5.3 Flash，告诉我们你会做什么——一间厨房、一份演示文稿，还是你梦想中的 App。下期见！

Episode: OpenAI Jalapeño: The Chip That Beat NVIDIA Duration: approximately 9 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, here's a strange question to start - what do a Japanese pork cutlet, an Indian curry, a bowl of chickpeas, and a chili pepper have in common? zh:欢迎回到 Learn English with Podcasts！Sarah，开场先来个奇怪的问题——日式炸猪排、印度咖喱、一碗鹰嘴豆和一根辣椒，有什么共同点？ [Sarah]: Hmm... somebody planned a very international lunch? zh:嗯……有人计划了一顿非常国际化的午餐？ [Mike]: Funny you say that - those are real product names inside OpenAI's data centers. The tray holding CPUs is called Katsu. The chip tray is Vindaloo. The network switch is Chana. zh:你说巧了——这些是 OpenAI 数据中心里真实的产品名。装 CPU 的托盘叫 Katsu，装芯片的托盘叫 Vindaloo，交换机叫 Chana。 [Sarah]: Wait, seriously? Engineers named computer parts after dinner? zh:等等，认真的吗？工程师用晚餐的名字给电脑零件命名？ [Mike]: And the star of this week - OpenAI's first chip designed completely in-house - is named after the pepper. It's called Jalapeño. zh:而本周的主角——OpenAI 第一颗完全自研的芯片——是用那个辣椒命名的。它叫 Jalapeño。 [Sarah]: Okay, a pepper chip. So what did it do? Because I'm guessing you didn't come here with food news. zh:好，一颗辣椒芯片。那它干了什么？我猜你不会专程来报美食新闻。 [Mike]: It picked a fight with the king. An independent testing lab called SemiAnalysis put Jalapeño side by side with NVIDIA's strongest AI chips - GB200 and GB300. And the pepper beat them. On every chart. zh:它向王者下了战书。一家叫 SemiAnalysis 的独立测试实验室，把 Jalapeño 和英伟达最强的 AI 芯片 GB200、GB300 放在一起对比。结果辣椒赢了。每张图上都赢。 [Sarah]: Every chart? Give me a number I can feel. zh:每张图都赢？给我一个有感觉的数字。 [Mike]: My favorite one: on a model called GPT-OSS 120B, Jalapeño pushed out 1459 tokens per second. Tokens are the little pieces of words that AI generates one by one. zh:我最喜欢的：在一个叫 GPT-OSS 120B 的模型上，Jalapeño 每秒输出 1459 个 token。token 就是 AI 一个个生成的小文字碎片。 [Mike]: NVIDIA's best managed about 535 on the same job. Not even half. zh:同样的活儿，英伟达最强只跑出大约 535 个。连一半都不到。 [Sarah]: 1459 versus 535? That's not winning a race, that's lapping people. zh:1459 对 535？这不是赢比赛，这是套圈了。 [Mike]: And analysts went wild. Dylan Patel, a famous semiconductor analyst, wrote that not just Blackwell got surpassed - even NVIDIA's newest Rubin was beaten. Even Sam Altman commented, and his whole statement was: we built a chip that is absurdly fast. zh:分析师们也炸锅了。著名半导体分析师 Dylan Patel 写道：被超越的不只是 Blackwell，连英伟达最新的 Rubin 也被击败了。连奥特曼都出来表态，他的原话就一句：我们造了一颗快得离谱的芯片。 [Sarah]: Confident. Okay, but speed per second is one thing. When my food delivery says thirty minutes, what I care about is when the doorbell rings. Total time. zh:够自信。不过每秒速度是一回事。外卖说三十分钟送到时，我在乎的是门铃什么时候响。总时间。 [Mike]: Same story there. One task: Jalapeño finished in 1.65 seconds. GB300 needed almost 6 seconds - about 3.6 times slower. zh:总时间也是同样的故事。同一个任务：Jalapeño 用 1.65 秒完成，GB300 要将近 6 秒——慢了约 3.6 倍。 [Sarah]: Four extra seconds sounds harmless for a chat, honestly. zh:说实话，多 4 秒在聊天里好像没什么伤害。 [Mike]: For chatting, sure. But think about AI agents - programs that run long jobs step by step, like booking flights or filling in forms. Dozens of steps, each one a little slow... the delays multiply. Slow chips make painfully slow agents. zh:聊天确实没事。但想想 AI agent——那些一步步执行长任务的程序，比如订机票、填表格。几十步，每步慢一点……延迟会相乘。慢芯片造出慢到让人痛苦的 agent。 [Sarah]: Ah, that's why everyone suddenly cares about milliseconds. Okay, there's also a wild number going around online - something about 104 times? zh:啊，所以大家才突然在意毫秒。对了，网上还流传一个疯狂数字——好像是 104 倍？ [Mike]: 104.3 times, and it needs context. Engineers pushed GB300 up to its own maximum decoding speed - around 169 tokens per second. At exactly that point, Jalapeño's throughput was 104.3 times higher. zh:104.3 倍，这个数字需要上下文。工程师把 GB300 推到它自己的极限解码速度——大约每秒 169 个 token。恰恰在那个点上，Jalapeño 的吞吐量是它的 104.3 倍。 [Sarah]: So where the champion collapses out of breath, the newcomer is still cruising? zh:所以冠军喘不上气的地方，新人还在匀速巡航？ [Mike]: Perfect image. To be fair, at each chip's best setting the gap is smaller - between 1.5 and 1.9 times across three models. Still stunning for a first attempt. zh:画面感满分。公平地说，各自跑在最佳设置时差距小一些——三个模型上是 1.5 到 1.9 倍。但对第一次尝试来说依然惊人。 [Sarah]: What about electricity? Chips this fast usually drink power like crazy. zh:那耗电呢？这么快的芯片通常是电老虎吧。 [Mike]: Another shock. Jalapeño runs at 700 watts. GB300 needs 1400. Half the electricity, double the work. zh:又一个震撼点。Jalapeño 跑在 700 瓦，GB300 需要 1400 瓦。一半的电，双倍的活。 [Sarah]: Even my hair dryer feels embarrassed now. zh:连我的吹风机现在都觉得丢脸。 [Mike]: Ha! Count cooling and everything in, and running one Jalapeño costs about 1.56 dollars per hour - almost identical to an older H100, and far below NVIDIA's newest Rubin at 3.61 dollars. zh:哈！把散热全都算进去，跑一颗 Jalapeño 每小时成本约 1.56 美元——跟老将 H100 几乎一样，远低于英伟达最新 Rubin 的 3.61 美元。 [Sarah]: Cheaper per hour than my streaming subscriptions... okay wait. Is the pepper even showing its final form here? zh:每小时比我的视频会员还便宜……等等。这颗辣椒这是最终形态了吗？ [Mike]: Not even close! They ran these tests WITHOUT speculative decoding - a clever guessing trick that speeds up AI output. Experts estimate that trick alone could cut the cost per token by more than two thirds. zh:差远了！他们是在没开投机解码的情况下测的——一个加速 AI 输出的聪明猜词技巧。专家估算，光这一招就能把每 token 成本再砍掉三分之二以上。 [Sarah]: Hold on. Rewind. This is OpenAI's FIRST chip ever. Chip companies usually spend years learning this craft. How is this possible? zh:等一下。倒带。这是 OpenAI 有史以来第一颗芯片。芯片公司通常要花多年修炼这门手艺。这怎么可能？ [Mike]: Two reasons. First, raw speed: design to factory took nine months. The industry norm is 18 to 36 months. zh:两个原因。第一，纯粹的速度：从设计到工厂只花了九个月。行业常规是 18 到 36 个月。 [Sarah]: Nine months! Some online shopping deliveries take longer! zh:九个月！有些网购快递都比这慢！ [Mike]: Reason two is the wild one. OpenAI brought an AI into the design room - their strongest model, called GPT-Astra. Probably the same model everyone keeps whispering about as GPT-6. zh:第二个原因才是真离谱。OpenAI 把一个 AI 请进了设计室——他们最强的模型，叫 GPT-Astra。很可能就是大家一直悄悄传的 GPT-6。 [Mike]: It explored design options, compressed the whole build-measure-test loop, and fine-tuned tiny circuits. One math unit shrank by 8%, another by 10% - while getting faster and cooler. zh:它帮忙探索设计方案，压缩整个"搭建-测量-验证"循环，还微调了微型电路。一个运算单元缩小了 8%，另一个缩小了 10%——同时速度更快、发热更低。 [Sarah]: So an AI shrank circuits for a chip whose entire job is running AI? zh:所以是一个 AI 为一颗芯片缩小了电路，而这颗芯片的全部工作就是跑 AI？ [Mike]: Yes. AI designs the chip. The chip runs smarter AI. Smarter AI designs the next chip. Round and round. Engineers call it a flywheel - heavy at first, then almost impossible to stop. zh:没错。AI 设计芯片，芯片跑更聪明的 AI，更聪明的 AI 再设计下一代芯片。一圈又一圈。工程师管这叫飞轮——起步很沉，转起来之后几乎停不下来。 [Mike]: One more detail shows how extreme it is. The main computing area measures about 840 square millimeters. The lithography machine's biggest possible canvas is 858. They used nearly all of it. zh:还有个细节能看出它有多极端。主计算区域约 840 平方毫米。光刻机能画的最大画布是 858 平方毫米。几乎全用上了。 [Sarah]: Like packing your suitcase until the airport scale reads 22.9 kilograms. zh:就像把行李箱打包到机场秤显示 22.9 公斤为止。 [Mike]: Exactly. And the next version is already in production, promising about 25% better performance per watt. zh:就是这样。而且下一版已经在生产了，预计每瓦性能还会再高约 25%。 [Sarah]: Impressive engineering. But NVIDIA has more than chips, right? Years ago every engineer learned their software... CUDA? zh:工程上确实漂亮。但英伟达手里不止芯片吧？多年前每个工程师都学过他们的软件……CUDA？ [Mike]: Right - CUDA has been the famous moat. A moat is the deep water ring around a castle. Nearly twenty years of tools and habits sat around NVIDIA's chips. Switching hardware meant relearning everything from zero. zh:对——CUDA 就是那条著名的护城河。护城河就是城堡周围那圈深水。围绕英伟达芯片积累了将近二十年的工具和习惯。换硬件意味着一切从零重学。 [Sarah]: So did the pepper swim across? zh:那这颗辣椒游过来了吗？ [Mike]: SemiAnalysis wrote a line nobody can stop quoting - the CUDA moat may be dead. Their reasoning: NVIDIA's hardware isn't weak. Rubin actually reached the factory one month before Jalapeño. What lost the race was how fast the software around the chip grew up. zh:SemiAnalysis 写了一句大家疯狂引用的话——CUDA 护城河可能已经死了。他们的推理是：英伟达硬件并不弱。Rubin 其实比 Jalapeño 早一个月进工厂。输掉比赛的是芯片周边软件长大的速度。 [Sarah]: And starting from zero helped OpenAI there, right? No old code dragging behind. zh:而从零开始反而帮了 OpenAI，对吧？没有旧代码拖后腿。 [Mike]: A blank page lets you draw anything. Now, ready for the ironic part? zh:白纸才能随便画。现在，准备好听最讽刺的部分了吗？ [Sarah]: Always. zh:随时奉陪。 [Mike]: Those clever OpenAI models that helped design the pepper? While they worked on it, they were running on NVIDIA GPUs. zh:帮着设计这颗辣椒的那些聪明 OpenAI 模型？它们干活的时候，跑的就是英伟达 GPU。 [Sarah]: Wait. So NVIDIA's machines raised their own challenger? Like training the apprentice who comes back to take over your shop? zh:等等。所以英伟达的机器养大了自己的挑战者？就像培训了一个学成回来接管你店面的学徒？ [Mike]: And they're not stopping at one pepper. Last October, OpenAI signed a deal with Broadcom for 10 gigawatts of custom chips - roughly ten nuclear reactors running flat out. zh:而且他们不会止步于一颗辣椒。去年 10 月，OpenAI 跟博通签了 10 吉瓦定制芯片的大单——差不多相当于十座核电站满负荷发电。 [Sarah]: Ten reactors' worth of chips. And Jalapeño is only generation one? zh:十座核电站级别的芯片。而 Jalapeño 只是第一代？ [Mike]: Generation two is deep in development, generation three taking shape, mass production climbing through 2027. Though there's one shadow: their data center chief, Chris Malone - who oversaw the huge Stargate project - just left the company. zh:第二代在深度开发，第三代正在成形，量产会在 2027 年爬坡。不过有一道影子：他们的数据中心掌门人 Chris Malone——负责庞大的星际之门项目——刚刚离职。 [Sarah]: Right before a possible IPO, too. Risky timing. But honestly, Mike, the picture I can't shake is much quieter than all this drama. zh:还赶在可能的 IPO 前夜。这个时机挺危险。不过说实话，Mike，我脑子里挥之不去的画面比这些大戏安静多了。 [Mike]: Which picture? zh:哪个画面？ [Sarah]: Somewhere in a lab sits a chip named after a snack pepper, resting in trays named after dinner - and it quietly beat the most valuable chip company on Earth. zh:某个实验室里坐着一颗以佐餐辣椒命名的芯片，躺在以晚餐命名的托盘里——而它安静地击败了地球上最值钱的芯片公司。 [Mike]: And today NVIDIA still wears the crown. But notice something - the challenger didn't queue at the front gate. It's already inside the server room, fans spinning. zh:今天英伟达仍戴着王冠。但注意一件事——挑战者没有在大门口排队。它已经在机房里面了，风扇正转着呢。 [Sarah]: Spicy times ahead. That's all for today! If you enjoyed this episode, share it with a friend who follows tech news. zh:接下来的日子够辣的。今天就到这里！如果你喜欢这期节目，分享给关注科技新闻的朋友吧。 [Mike]: See you next time on Learn English with Podcasts! zh:下期 Learn English with Podcasts 再见！

Episode: Pocket, Car, Home: Xiaomi XRING O3, O100, D100 Duration: approximately 10 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, here's a question to warm up your brain - what do your phone, your car, and your smart speaker have in common? zh:欢迎回到 Learn English with Podcasts！Sarah，先来个问题热热身——你的手机、汽车和智能音箱有什么共同点？ [Sarah]: Easy. They all ignore me when I talk to them. zh:简单。它们在我跟它们说话的时候都不理我。 [Mike]: Ha! Okay, technically true. But the real answer is - they all need a brain. A little computer inside that does the thinking. zh:哈！好吧，严格来说没错。但真正的答案是——它们都需要一个大脑。身体里负责思考的小电脑。 [Sarah]: True. And those brains usually come from big chip companies like Qualcomm. zh:没错。而且这些大脑通常来自高通这样的大芯片公司。 [Mike]: Right, most companies buy them. But yesterday, Xiaomi introduced three brand-new chips it designed all by itself. One for phones, one for AI speed, and one for cars. zh:对，大多数公司都是买来的。但就在昨天，小米一口气发布了三颗完全自研的全新芯片。一颗给手机，一颗给 AI 提速，还有一颗给汽车。 [Sarah]: Three at once? Designing chips is famously hard. That's like a restaurant growing its own vegetables, baking its own bread, AND raising its own cows. zh:一次三颗？做芯片是出了名的难。这就像一家餐厅自己种菜、自己烤面包，还要自己养牛。 [Mike]: And it's not their first rodeo. Last May they released the XRING O1 - the first 3nm flagship chip designed in mainland China. It went straight into their top phone and two tablets. zh:而且这不是他们第一次干了。去年 5 月他们发布了玄戒 O1——中国大陆首款自研 3nm 旗舰 SoC，直接装进了自家顶级手机和两款平板。 [Sarah]: How long does something like that take? Genuinely curious. zh:这种事要花多长时间？真心好奇。 [Mike]: Over five years of work, more than 21 billion yuan in research spending, and a team of nearly 3000 engineers. zh:五年多的投入，超过 210 亿元的研发经费，还有一支将近 3000 人的工程师团队。 [Sarah]: Five years before anyone could even hold a product. That's patience most companies don't have. zh:五年之后才有人真正拿到产品。这份耐心大多数公司都没有。 [Mike]: Now, about the new family. The star is called XRING O3, and it will appear first inside the Xiaomi 18 Fold - their upcoming folding phone. zh:现在说新家族。主角叫玄戒 O3，它将首发于即将推出的折叠屏手机小米 18 Fold。 [Sarah]: What makes this one special? Give me the wow number. zh:这颗特别在哪里？给我一个惊叹数字。 [Mike]: 24 billion transistors, on a piece of silicon about the size of your thumbnail. zh:240 亿个晶体管，塞在一块差不多指甲盖大小的硅片上。 [Sarah]: Transistors being... the tiny switches that do all the actual thinking? zh:晶体管就是……那些真正负责思考的微型开关？ [Mike]: Exactly. Think of each one as a light switch. More switches means smarter chip. This one has 26% more switches than last year's model, and it scores 5.22 million on AnTuTu - a popular speed test for phones. zh:没错。把每一个都想象成一个电灯开关。开关越多，芯片越聪明。这一代比上一年多了 26%，在安兔兔——一个常用的手机跑分软件——上拿到了 522 万分。 [Sarah]: Impressive. But honestly, my current phone already feels fast. Why is everyone suddenly so obsessed with phone chips? zh:厉害。但说实话，我现在的手机已经觉得挺快了。为什么大家突然都对手机芯片这么执着？ [Mike]: One word - AI. The new AI features run directly on your phone, not in some faraway data center. And that completely changes what a phone chip needs to be good at. zh:一个词——AI。新的 AI 功能直接在你的手机上运行，而不是在遥远的机房里。这彻底改变了手机芯片需要擅长的事情。 [Sarah]: Ah, so the chip isn't just running apps anymore. It's running brains. zh:啊，所以芯片不再只是跑应用了。它是在跑大脑。 [Mike]: Beautifully said. Example: the graphics section got almost twice as strong while using 64% less power than last year. In this industry, one generation usually brings maybe 20% to 30% improvement. zh:说得好。举个例子：图形模块性能几乎翻倍，功耗却比上一年降低了 64%。在这个行业，一代升级通常也就带来 20% 到 30% 的提升。 [Sarah]: Twice the muscle on less electricity? That's breaking the usual rules of progress. zh:更省电还肌肉翻倍？这打破了通常的进步规律啊。 [Mike]: Even memory got special treatment. This is the first phone chip that supports LPDDR6, a new super-fast type of memory. But Xiaomi also attacked waiting time - how long the chip just sits there, twiddling its thumbs, waiting for data to arrive. zh:连内存也得到了特殊照顾。这是第一款支持 LPDDR6 新式高速内存的手机芯片。不过小米还向"等待时间"开刀了——也就是芯片闲坐在那里干等数据送达的时间。 [Sarah]: Twiddling its thumbs - love that image. So how do you make data arrive faster? zh:"闲得搓手手"——这个画面我喜欢。那怎么让数据更快到达呢？ [Mike]: Three tricks. First, old systems kept translating between different protocols - like translating Chinese into English, then German, then Russian. Xiaomi made everyone speak one language instead. zh:三个招数。第一，老系统要在不同协议之间不停翻译——像把中文译成英文、再译成德文、再译成俄文。小米干脆让大家都说同一种语言。 [Mike]: Second, they built a dedicated express lane between two important parts of the chip. That alone saved 15 nanoseconds per trip. zh:第二，他们在芯片里两个重要部件之间修了一条专属高速通道。光这一项，每趟就省下 15 纳秒。 [Sarah]: Nanoseconds. A billionth of a second each. Are we seriously counting those? zh:纳秒。每纳秒是十亿分之一秒。我们真要计较到这个程度吗？ [Mike]: We are. Your phone does billions of these trips every second. Save a little each time, and suddenly videos load faster and batteries last longer. zh:真的要。你的手机每秒要做几十亿次这样的数据往返。每次省一点点，加起来就是视频加载更快、电池更耐用。 [Sarah]: Okay, you said AI changed everything. Where exactly does the AI part live in this chip? zh:好，你刚才说 AI 改变了一切。那 AI 部分到底住在这颗芯片的哪里？ [Mike]: There's a special engine called an NPU. Picture a math genius sitting next to a room full of general office workers. Regular tasks go to the workers; heavy matrix math goes to the genius. This one delivers 200 TOPS of computing power, tuned specially for large language models. zh:里面有个叫 NPU 的专用引擎。想象一位数学天才坐在一屋子普通职员旁边。日常任务交给职员，繁重的矩阵运算交给天才。它能输出 200 TOPS 算力，专门为大语言模型调校。 [Sarah]: A genius on the team. Every office needs one. zh:团队里有个天才。每个办公室都需要一位。 [Mike]: And here's my favorite detail - the chip team and the AI model team actually sat together and co-designed things. They pack the model like a suitcase. Same clothes, tighter folding. zh:还有我最喜欢的一个细节——芯片团队和 AI 模型团队真的坐到一起联合设计。他们把模型像行李箱一样打包。同样的衣服，叠得更紧。 [Sarah]: Suitcase packing - finally an analogy I can reuse at the airport. zh:行李箱打包——终于有一个我能在机场复用的比喻了。 [Mike]: And the numbers came out great. Answers start 40% faster, replies flow 45% faster, and it uses 26% less power than typical top chips. zh:而且数字非常漂亮。出第一个答案快了 40%，回复生成快了 45%，功耗比典型旗舰芯片低 26%。 [Sarah]: So the smart stuff happens right in your hand. No internet needed. zh:所以聪明的部分就发生在你手心里。不需要网络。 [Mike]: And not just the main brain. They added small AI engines everywhere - camera, screen, even the microphone system for online meetings. zh:而且不只是主大脑。他们把小型 AI 引擎加到了各个角落——相机、屏幕，连线上会议用的麦克风系统都有。 [Sarah]: Got it. Phone chip covered. What about the other two? zh:明白了。手机芯片讲完了。另外两颗呢？ [Mike]: Number two is my favorite weirdo. It's called XRING O100, and it exists to attack something engineers dramatically call the memory wall. zh:第二颗是我最喜欢的怪咖。它叫玄戒 O100，存在的意义就是攻克工程师们口中那个听起来很戏剧化的"内存墙"。 [Sarah]: Memory wall? Like when you walk into a room and forget why you came? zh:内存墙？就像你走进一个房间却忘了自己为什么来？ [Mike]: Ha - close enough, but silicon version. Picture a kitchen. The chef - that's the processor - stands in the living room. The fridge - that's the memory - sits out on the balcony. Every single ingredient gets carried over by hand, back and forth, all day long. zh:哈——差不多，不过这是硅片版。想象一个厨房。厨师——也就是处理器——站在客厅里。冰箱——也就是内存——放在阳台上。每一种食材都得靠人手来回搬，搬一整天。 [Sarah]: So half the energy goes to walking, not cooking. zh:所以一半力气花在走路上了，不是做饭。 [Mike]: Exactly. And the obvious fix? zh:就是这个道理。那显而易见的解决办法是什么？ [Sarah]: Move the fridge into the kitchen! zh:把冰箱搬进厨房啊！ [Mike]: They went further - they stacked the memory directly UNDER the processor, floor after floor, like a sandwich tower. Data now travels vertically, millimeters instead of centimeters. It's the world's first chip built this way at 6nm. zh:他们做得更彻底——把内存直接堆到处理器正下方，一层接一层，像一座三明治高塔。数据现在垂直流动，走毫米而不是厘米。这是全球第一颗用 6nm 工艺这样造出来的芯片。 [Mike]: Result: bandwidth of 1.22 terabytes per second. Your fancy flagship phone manages around 76.8 gigabytes. This thing is 16 times faster. zh:结果：带宽达到每秒 1.22TB。你那台高级旗舰手机大约只有 76.8GB。这东西快了 16 倍。 [Sarah]: Sixteen times! And what does it do with all that speed? zh:16 倍！那这些速度用来干什么？ [Mike]: Feeding giant AI models locally. Paired with the O3, it can push out up to 330 tokens per second. Tokens are the small pieces AI uses to build words. zh:在本地喂饱巨型 AI 模型。和 O3 搭配后，每秒最多能输出 330 个 token。token 就是 AI 组词用的小碎片。 [Sarah]: So the machine answers before you finish blinking. Okay, chip number three? zh:所以你还没眨完眼它就把答案吐出来了。好，第三颗芯片呢？ [Mike]: Number three rides in your car. The D100 is China's first 3nm chip for smart driving, strong enough to run a model with 200 billion parameters inside the vehicle itself. zh:第三颗坐在你的车里。D100 是国内首款 3nm 智驾芯片，强大到可以直接在车内运行一个 2000 亿参数的模型。 [Sarah]: The car does its own thinking - no cloud required at highway speed. zh:车自己动脑——高速公路上完全不依赖云端。 [Mike]: They previewed it inside a desktop prototype called AI Cube. And my favorite silly detail? Its metal shell has exactly 33,874 tiny holes drilled for cooling. zh:他们提前把它装进一台叫 AI Cube 的桌面原型机展示。我最喜欢的傻细节是什么？它的金属外壳上精确地钻了 33874 个散热小孔。 [Sarah]: Somebody counted every single hole. Give that person a raise. zh:真有人把每个孔数了一遍。给那位同事加薪吧。 [Sarah]: You know what strikes me though? Most people still picture Xiaomi as a phone company. zh:不过你知道让我感触最深的是什么吗？大多数人还是把小米当成一家手机公司。 [Mike]: Meanwhile the chips are quietly becoming the whole story. Phone, car, living room - one family of brains that all speak the same language. Because here's the thing: the cleverest AI model is useless if the silicon underneath can't keep up. zh:而芯片正在悄悄变成整个故事本身。手机、汽车、客厅——一家人大脑说着同一种语言。因为关键在这儿：如果底下的硅片跟不上，再聪明的 AI 模型也没用。 [Sarah]: From pocket to garage to sofa, all running the same recipe. Meanwhile, I still lose my keys twice a week. Maybe they'll design a chip for that next. zh:从口袋到车库再到沙发，跑的都是同一套配方。而我呢，一周还是要丢两次钥匙。也许他们下一颗就该为这个设计芯片了。 [Mike]: Give them five years and another 21 billion yuan - those keys will never escape again. That's all for today! If you enjoyed this episode, share it with a friend who loves gadgets. zh:再给他们五年时间和 210 亿元——钥匙从此再也逃不掉了。今天就到这里！如果你喜欢这期节目，分享给爱数码的朋友吧。 [Sarah]: See you next time on Learn English with Podcasts! zh:下期 Learn English with Podcasts 再见！ [Mike]: See you next time! zh:下期见！

Episode: Wan3.0 Goes Live: Stable, Real, Textured Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, quick question - how long can most AI video models film in one single go? zh:欢迎回到 Learn English with Podcasts！Sarah，问个问题——大多数 AI 视频模型一次最多能生成多长的视频？ [Sarah]: A few seconds, usually five or ten. Long enough for a looping clip, not long enough for a story. zh:一般就几秒钟，五秒或十秒。够做一段循环动画，但不够讲一个故事。 [Mike]: Right, and that changed today. Alibaba's video model Wan3.0 is now officially live, and it can generate 30 seconds in one take. zh:没错，今天这事儿变了。阿里巴巴的视频模型 Wan3.0 正式上线了，单次就能生成 30 秒的视频。 [Sarah]: 30 seconds in one take? That's basically a full scene with a beginning and an end. zh:一次生成 30 秒？那基本就是一个有开头有结尾的完整场景了。 [Mike]: Exactly. It was in public testing since August 6, and today is the official launch day. And there's a second surprise - you can feed it documents. zh:对。它从 8 月 6 日就开始公测了，今天是正式上线。还有第二个惊喜——你可以喂文档给它。 [Sarah]: Documents? Like Word files and spreadsheets? zh:文档？就像 Word 文件和表格那种？ [Mike]: Yes - doc, xls, ppt, pdf, even markdown files. Upload them, and the model builds a video from the content. zh:对——doc、xls、ppt、pdf，甚至 markdown 文件都行。上传之后，模型会根据内容做出一段视频。 [Sarah]: Wait, so I could upload my meeting notes and get a video version? That's a strange new superpower. zh:等等，那我上传会议记录就能得到一个视频版？这是个奇怪的新超能力。 [Mike]: Welcome to 2026. But the word people keep repeating about this model is "real". Real faces, real light, real physics. zh:欢迎来到 2026。不过大家提到这个模型时反复说的一个词是"真实"。真实的脸、真实的光、真实的物理效果。 [Sarah]: Okay, prove it. What did people actually make with it? zh:好，证明给我看。人们用它到底做出了什么？ [Mike]: One creator made a short film called The Bar That Sells Memories - a bar where customers buy memories. Great premise already. zh:有一位创作者拍了一部短片，叫《出售回忆的酒吧》——一家顾客可以购买回忆的酒吧。这个设定本身就很棒。 [Sarah]: Oh, I want to watch that already. What about the quality? zh:哦，我已经想看了。画质怎么样？ [Mike]: The hero's face stays the same from a wide shot to a very close close-up. That's the hardest part for AI, because faces usually drift. zh:主角的脸从中景推到大特写都保持一致。这是 AI 最难的部分，因为脸通常会变来变去。 [Sarah]: I've seen that drift. By shot five, your character looks like a cousin of shot one. zh:我见过那种漂移。到第五个镜头，角色看起来像第一个镜头的表亲。 [Mike]: Ha, exactly. And the bar itself kept one color mood - warm amber lights, wooden furniture, and glass bottles reflecting the whole room. zh:哈，就是。而且酒吧本身保持了统一的色彩氛围——温暖的琥珀色灯光、木质家具，玻璃瓶还反射着整个店内的景象。 [Sarah]: Glass reflecting the room? Even professional crews struggle with that. zh:玻璃反射店内景象？连专业摄制组都觉得难。 [Mike]: Now the business side. An AI review team called Flova kept saying one word: stable. Not just pretty frames - stable output you can plan real work around. zh:再说说商业这边。一个叫 Flova 的 AI 评测团队反复说一个词：稳定。不只是画面好看，而是稳定到你敢围绕它安排真正的生产。 [Sarah]: For companies, boring words like stable are actually exciting words. zh:对公司来说，"稳定"这种无聊的词其实才让人兴奋。 [Mike]: In their test clip, one single shot had eye reflections, moving hair, flowing white fabric, and shiny water - four things AI usually breaks. zh:在他们的测试片段里，一个镜头同时出现了眼球反射、飘动的头发、白纱的运动和水面光泽——这四样通常是 AI 容易翻车的。 [Sarah]: Four hard things in one shot? And the skin didn't look plastic? zh:一个镜头四个难点？皮肤也没有塑料感？ [Mike]: No plastic at all. It even held a cold grey-blue color tone all the way through. zh:完全没有塑料感。它甚至还全程守住了冷灰蓝的色调。 [Sarah]: Did anyone outside China test it? zh:有海外的团队测试吗？ [Mike]: Yes. Koyal.ai tested six animation categories, and Wan3.0 ranked number one overall. zh:有。Koyal.ai 测评了六个动画类别，Wan3.0 总分排名第一。 [Sarah]: Number one across six categories - that's not luck anymore. zh:六个类别综合第一——这就不是靠运气了。 [Mike]: They loved its high-energy motion, like fast action scenes, and its realistic style scored high too. zh:他们特别喜欢它的高动态表现，比如快节奏的动作场面，写实风格的得分也很高。 [Sarah]: So who is actually using this every day? Give me the industries. zh:那现在谁在天天用它？跟我说说有哪些行业。 [Mike]: Short drama studios first. One tech team said close-ups of actors' faces come out clean and steady, so they can produce shows at scale. zh:首先是短剧团队。一家技术公司说，演员面部的近景特写出来又干净又稳，所以他们可以规模化生产短剧。 [Sarah]: At scale meaning many episodes, on schedule. zh:"规模化"就是说很多集、按进度出片。 [Mike]: Right. A platform called LibTV made a car chase - cars, drivers, and city streets stayed consistent for the whole 30 seconds. zh:对。一个叫 LibTV 的平台做了一场追车戏——车辆、司机和城市街道在整个 30 秒里都保持一致。 [Sarah]: Without restarting fifty times to roll a lucky clip? zh:不用重开五十次去抽一段运气好的片段？ [Mike]: Almost no rerolling. Before, AI video felt like a slot machine - you pulled the lever again and again and hoped. zh:几乎不用重抽。以前 AI 视频像老虎机——你得一遍一遍拉杆，然后祈祷。 [Sarah]: And now the lever finally pays out every time. What about advertising? zh:现在这台老虎机终于每次都出货了。广告行业呢？ [Mike]: An ad company in Guangzhou said products and logos stay accurate, and talking presenters look natural - close to a real commercial shoot. zh:广州一家广告公司说，商品和 Logo 能保持准确，口播的人也很自然——接近真实广告拍摄的水平。 [Sarah]: Logos are a big deal. Brands get angry when their logo grows an extra corner. zh:Logo 这点很关键。品牌方看到自家 Logo 多出一个角是会生气的。 [Mike]: Tourism too. Creators rebuilt famous places with cinematic camera moves - an ancient city at sunset, red palace walls, cherry blossoms in Kyoto. zh:文旅也在用。创作者用电影级的运镜重现著名景点——晚霞中的古城、红色的宫墙、京都的樱花。 [Sarah]: So someone who may never visit Kyoto can still feel it first. zh:这样可能永远没机会去京都的人，也能先感受一下。 [Mike]: And musicians. One music company uploaded a single photo of a singer, marked lip-sync singing, and got several camera angles automatically. zh:还有音乐人。一家音乐公司上传了一张歌手的照片，标注对口型唱歌，就自动得到了好几个机位镜头。 [Sarah]: From one photo to a finished music video? That used to need a whole film crew. zh:从一张照片到一个成品 MV？这以前需要一整个摄制组。 [Mike]: During those ten-plus days of public testing, it also improved daily - better instruction following, steadier cross-shot consistency, cleaner audio. zh:公测的十多天里，它还在每天进化——指令遵循更好，跨镜头一致性更稳，音频也更干净。 [Sarah]: Okay, the important question - how much does it cost? zh:好，关键问题来了——多少钱？ [Mike]: Through the API, 480P costs 0.3 yuan per second, 720P is 0.6 yuan, and 1080P is 1.2 yuan. zh:API 价格是 480P 每秒 0.3 元，720P 是 0.6 元，1080P 是 1.2 元。 [Sarah]: Any launch discount? zh:上线有优惠吗？ [Mike]: Until September 23, two platforms offer a 30% discount. So yes, cheap enough to experiment with. zh:9 月 23 日之前，两个平台有 7 折优惠。所以说，拿来试错足够便宜。 [Mike]: And it's easy to find. It's live on Alibaba Cloud's Bailian platform, the Wan website, and several apps including Qwen. zh:而且很好找。阿里云百炼平台、万相官网，还有千问等多个 App 都能用上它。 [Sarah]: You know what just hit me? All those industries - drama, ads, tourism, music - they all sell stories. And this tool makes stories cheaper. zh:你知道吗，我突然想到一件事。刚才说的那些行业——短剧、广告、文旅、音乐——卖的都是故事。而这个工具让故事变便宜了。 [Mike]: Which means the bottleneck moves. It's no longer "can you afford to shoot it?" It's "do you have a story worth shooting?" zh:所以瓶颈就转移了。问题不再是"你拍不拍得起"，而是"你的故事值不值得拍"。 [Sarah]: Ooh, save that line for the show notes. Meanwhile, I'm going to upload my expense spreadsheet and ask for an emotional music video about taxes. zh:哇，这句要写进节目简介。而我嘛，打算把我的报销表格传上去，让它做一支关于报税的抒情 MV。 [Mike]: Please send it over when the violin comes in. Try Wan3.0, and tell us - what would you feed it: a photo, a document, or your wildest idea? zh:等小提琴响起来的时候一定要发给我们。试试 Wan3.0 吧，然后告诉我们——你会喂给它什么：一张照片、一份文档，还是你最疯狂的想法？ [Sarah]: We'd love to hear it. See you next time on Learn English with Podcasts! zh:我们很期待听到你的分享。下期 Learn English with Podcasts 再见！ [Mike]: See you next time! zh:下期见！

Episode: DeepSeek V4 Flash Vision Exp: AI Learns to See Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to Learn English with Podcasts! Sarah, quick question - can your AI read your vacation photos and actually do something useful with them? zh:欢迎回到 Learn English with Podcasts！Sarah，问你个小问题——你的 AI 能看懂你的度假照片，然后真的做点有用的事吗？ [Sarah]: Honestly, most AIs just say nice photo! But you're smiling like you found one that does more. zh:老实说，大多数 AI 只会说"照片不错！"但你笑得这么开心，肯定是发现了更厉害的。 [Mike]: I did. DeepSeek just released something called DeepSeek V4 Flash Vision Exp. It's experimental - Exp means it's a test version. zh:我确实发现了。DeepSeek 刚刚发布了一个叫 DeepSeek V4 Flash Vision Exp 的东西。它是实验性的——Exp 就是测试版的意思。 [Sarah]: DeepSeek V4 Flash Vision Exp - that's a long name! What makes it special? zh:DeepSeek V4 Flash Vision Exp——名字好长啊！它特别在哪里？ [Mike]: Two words: it can see. The normal V4 Flash is great with text. This new one keeps all that text power, but adds eyes. zh:两个字：它能看见。普通的 V4 Flash 就已经很擅长处理文字了。这个新版本保留了全部文字能力，还加了一双眼睛。 [Sarah]: Wait, so it didn't lose any text skills? Usually when you add vision, the text part gets weaker. zh:等等，所以它的文字能力一点没丢？通常加了视觉能力后，文字部分就会变弱。 [Mike]: Exactly, that's the balance. In pure text tasks - like solving problems or answering knowledge questions - it scores the same as V4 Flash. zh:没错，这就是它的平衡。在纯文字任务上——比如解决问题或回答知识类问题——它的得分和 V4 Flash 一样。 [Sarah]: Okay, that's good. But what about the vision part? How much better is it? zh:好，这很不错。那视觉部分呢？提升了多少？ [Mike]: Big jump. On agent benchmarks that need vision, the normal Flash was not strong. This Vision Exp version jumps up to almost the level of Opus-4.8. zh:提升很大。在那些需要视觉的智能体测评里，普通的 Flash 并不强。这个 Vision Exp 版本一跃提升到几乎和 Opus-4.8 同等水平。 [Sarah]: Opus-4.8? That's one of the top models. So this experimental model is already playing in the big league? zh:Opus-4.8？那可是顶级模型之一。所以这个实验性模型已经能和顶尖选手同场竞技了？ [Mike]: Yes. And the cool part is how it works inside agent tools. You give it a big goal, and it uses vision to get there. zh:是的。更酷的是它在智能体工具里的工作方式。你给它一个大目标，它就用视觉能力去实现。 [Sarah]: Give me a real example. I learn better with stories. zh:给我一个真实的例子。我听故事学得更快。 [Mike]: Okay, first one - a luxury Tibet travel PPT. The prompt was crazy detailed: one month, self-driving, south plus north Tibet, wild and raw style, not touristy photos. zh:好，第一个——一份高端西藏旅行 PPT。提示词非常详细：一个月，自驾，藏南加藏北，要野性原始的风格，不要游客照。 [Sarah]: A month in Tibet? And the AI has to make slides for rich customers? zh:在西藏待一个月？而且 AI 要给高净值客户做幻灯片？ [Mike]: Yes, for high-end private tours. And get this - the final PPT needed three real pricing plans, with real photo style images, rough and powerful look. No cute filters. zh:对，是给高端私人定制游的。而且关键是——最后的 PPT 需要三种真实的报价方案，配上真实摄影风格的图片，粗粝有力量感。不要小清新的滤镜。 [Sarah]: And the AI did it? Just from that long text description? zh:然后 AI 做出来了？就凭那段长长的文字描述？ [Mike]: With the vision ability, yes. It found the right images, matched the wild tone, and built a real-looking deck that you could actually send to customers. zh:有了视觉能力，是的。它找到了合适的图片，匹配了野性的基调，做出了一份看起来真的能发给客户的演示文稿。 [Sarah]: Wow, so it understands both the words and the pictures fit the feeling. What's the second example? zh:哇，所以它既理解文字，也能让图片贴合感觉。第二个例子是什么？ [Mike]: This one is fun. They asked it to rebuild a developer website - the DeepSeek Harness site - with a future style. Dark blue sea, glass UI, ASCII art. zh:这个很有趣。他们让它重做一个开发者网站——DeepSeek Harness 的官网——要未来感风格。深蓝色海洋、玻璃质感界面、ASCII 艺术字。 [Sarah]: ASCII art? Like old computer letters making pictures? zh:ASCII 艺术？就是用老式电脑字母拼成图片那种？ [Mike]: Exactly. And after many rounds of talking, the model turned that idea into a whole new site. It kept listening, kept seeing what it made, and fixed it step by step. zh:没错。而且经过多轮对话后，模型把那个想法变成了整个新网站。它一边听，一边看自己做出来的东西，然后一步步修改。 [Sarah]: So it's not one-shot. It's like working with a designer who shows you drafts? zh:所以不是一次成型。就像和一个会给你看草稿的设计师合作？ [Mike]: Perfect example. And third one - a mini website with cute 3D clay monsters dancing at a retro party. zh:完美的比喻。第三个——一个迷你网站，一群可爱的 3D 黏土小怪物在复古舞池里跳舞。 [Sarah]: Clay monsters dancing? That's so random. I love it. zh:黏土怪物跳舞？这也太随意了吧。我喜欢。 [Mike]: The prompt was random too: a group of cute 3D clay monsters joining a jumping retro dance party. And the agent built a moving demo in the browser. zh:提示词本身就很随意：一群可爱的 3D 黏土小怪物加入一个跃动的复古舞池派对。然后智能体直接在浏览器里做出了会动的演示。 [Sarah]: Okay, so it can see, understand tone, and build things. How do developers actually use it? Is it hard? zh:好，所以它能看，能理解基调，还能搭建东西。开发者实际怎么用它？难吗？ [Mike]: Super easy. You just set model='deepseek-v4-flash-vision-exp' in the API. That's it. zh:超级简单。你只需要在 API 里设置 model='deepseek-v4-flash-vision-exp' 就行。就这么简单。 [Sarah]: Just change one line? And the price? zh:就改一行？那价格呢？ [Mike]: Same as V4 Flash. Each image costs at most 384 tokens. So you pay by tokens, not by extra vision fee. zh:和 V4 Flash 一样。每张图片最多算 384 tokens。所以你按 token 付费，没有额外的视觉费用。 [Sarah]: 384 tokens per image - that's pretty cheap. What about sending images? zh:每张图片 384 tokens——挺便宜的。那怎么发送图片呢？ [Mike]: Three ways: you can paste base64 code, send a URL link, or use their new Files API. zh:三种方式：你可以粘贴 base64 编码，发一个网址链接，或者用他们新的 Files API。 [Sarah]: Files API? Is that new? zh:Files API？是新出的吗？ [Mike]: Yes, just launched. You upload the image once, get a file_id, and use that file_id in many requests. No need to upload again. And it's free to upload. zh:是的，刚刚上线。你上传一次图片，拿到一个 file_id，然后在很多请求里重复使用这个 file_id。不用重复上传。而且上传是免费的。 [Sarah]: Oh, that saves time and bandwidth. Does the API work with different formats? zh:哦，那省了时间和带宽。API 支持不同的格式吗？ [Mike]: Yes, three formats: Chat Completions, Messages, and Responses. So it fits into most agent tools easily. zh:支持，三种格式：Chat Completions、Messages 和 Responses。所以它能很轻松地接入大多数智能体工具。 [Sarah]: You know what I find interesting? This is an experimental model. Exp means they are still testing. zh:你知道我觉得有意思的是什么吗？这还是个实验性模型。Exp 意味着他们还在测试。 [Mike]: Right, but it's already strong enough to make real PPTs and websites. Imagine what the final version will do. zh:对，但它已经强到能做真正的 PPT 和网站了。想想最终版会做到什么程度。 [Sarah]: And it mixes text and images together. That feels more human. We don't just read - we look. zh:而且它把文字和图片混在一起处理。这感觉更像人类。我们不只是阅读——我们也在看。 [Mike]: Right. Before, AI was like a smart friend who could only hear your words. Now it can see your photos, your screenshots, your designs. zh:对。以前，AI 就像一个只能听你说话的聪明朋友。现在它能看到你的照片、截图、设计稿。 [Sarah]: So it's not about who talks the most. It's about who can see what you're pointing at. zh:所以重要的不是谁话最多，而是谁能看懂你在指什么。 [Mike]: Exactly! And here's a funny thought - we spent years teaching AI to read our messy handwriting. Now we're teaching it to judge our PowerPoint taste. zh:没错！还有个好笑的想法——我们花了好多年教 AI 认我们潦草的字迹。现在我们在教它评判我们的 PPT 审美。 [Sarah]: Ha! My PPT taste definitely needs judging. So if listeners want to try it, just try that model name? zh:哈！我的 PPT 审美确实需要被评判一下。所以如果听众想试试，只要试试那个模型名就行？ [Mike]: Yes, try model DeepSeek V4 Flash Vision Exp and see what it sees. And tell us - what would you ask an AI that can finally see? zh:对，试试模型 DeepSeek V4 Flash Vision Exp，看看它能看见什么。也告诉我们——如果 AI 终于能看见了，你会让它帮你做什么？ [Sarah]: We'd love to hear your ideas. See you next time on Learn English with Podcasts! zh:我们很期待听到你的想法。下期 Learn English with Podcasts 再见！ [Mike]: See you next time! zh:下期见！

Episode: Jeff Dean's Honest Talk: Gemini, Discovery Loop Duration: approximately 12 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to "Learn English with Podcasts"! Sarah, imagine your old boss speaks in public for the first time after quitting - and says exactly what they really think. Awkward, right? zh:欢迎回到"Learn English with Podcasts"！Sarah，想象一下，你的老上司离职后第一次公开讲话，而且把心里话全说了出来。尴尬吧？ [Sarah]: Very awkward! But also kind of refreshing. Did that actually happen? zh:非常尴尬！但也挺让人耳目一新的。真有这种事吗？ [Mike]: It just did. Jeff Dean - the legend who helped build Google - gave his first public interview after leaving. And he did not hold back. zh:刚刚就发生了。Jeff Dean——帮忙打造谷歌的那位传奇人物——在离职后做了第一次公开访谈，而且一点都没藏着掖着。 [Sarah]: Wait, the Jeff Dean? The one behind Gemini, Google's biggest AI model? zh:等等，是那个 Jeff Dean 吗？就是谷歌最大 AI 模型 Gemini 背后的那个人？ [Mike]: The same. He co-started Gemini and was its first technical lead. In this talk, he admitted something surprising about it. zh:就是他。他共同发起 Gemini，也是它的首位技术负责人。在这次访谈里，他承认了一件挺让人意外的事。 [Sarah]: Oh, now I'm curious. What did he admit? zh:哦，我现在好奇了。他承认了什么？ [Mike]: He said Gemini was not great at coding in the early days. They cared a bit too late about making it really good at writing code. zh:他说 Gemini 早期在写代码方面表现并不出色。他们把"让它在编程上真正惊艳"这件事，重视得稍微晚了一点。 [Sarah]: Really? But Gemini is huge now. So what did they get right? zh:真的吗？可 Gemini 现在很厉害啊。那他们做对了什么？ [Mike]: One big thing: they built it multimodal from day one. Text, images, video, audio - all in one model. He said that was the right call. zh:一件大事：他们从第一天起就把多模态做进去了。文字、图像、视频、音频——全在一个模型里。他说这个决定非常正确。 [Sarah]: Makes sense. If you want one brain for everything, it should understand everything. Did he talk about older projects too? zh:有道理。如果你想有一个能应对一切的大脑，它就该理解一切。那他也聊了更老的项目吗？ [Mike]: Yes. He looked back at TensorFlow, the tool that helped millions learn machine learning. He admitted two clear mistakes. zh:聊了。他回顾了 TensorFlow——那个帮无数人入门机器学习的工具。他承认了两个明显的失误。 [Sarah]: Two mistakes from the genius? I'm listening. zh:天才也会犯两个错？我洗耳恭听。 [Mike]: First, they didn't have eager execution at the start. Later, PyTorch and JAX made it popular, and TensorFlow added it too late. Second, they opened a contrib folder where outsiders added ten different ways to do the same thing. zh:第一，他们一开始没有即时执行模式。后来 PyTorch 和 JAX 让这种模式流行起来，TensorFlow 加得太晚了。第二，他们开了一个 contrib 文件夹，外部开发者往里塞了十种做同一件事的不同方法。 [Sarah]: Oh no, ten ways to do one thing? That must have confused everyone. zh:哦不，一件事十种做法？那肯定把大家都搞晕了。 [Mike]: Exactly. He said they should have kept the core simple. But he also praised TensorFlow for helping so many people start in AI. Fair and honest. zh:没错。他说当初应该让核心保持简洁。但他也称赞 TensorFlow 帮了那么多人踏入 AI 领域。很公道，也很诚实。 [Sarah]: Okay, but here's my big question - he spent 27 years at Google. Why leave? zh:好，但我最大的问题是——他在谷歌待了 27 年。为什么要走？ [Mike]: He loved his time there. But he said a tiny, focused team has a special power: everyone works on one mission, with almost no distractions. zh:他很喜欢在谷歌的岁月。但他说，一支极小而专注的团队有一种特别的力量：所有人都围绕同一个使命工作，几乎没有干扰。 [Sarah]: Like a band of friends building something in a garage? zh:就像一群朋友在车库里搞发明？ [Mike]: Perfect picture. And today, cloud computing lets a small team rent huge machines. You don't need to build everything yourself. A team of about 10 people can move fast. zh:就是这个画面。而且今天，云计算让小团队也能租到巨型机器。你不需要自己从头搭建一切。大约 10 个人的团队也能跑得飞快。 [Sarah]: Still, leaving Google after 27 years must feel scary. zh:不过，在谷歌待了 27 年再离开，肯定还是会有点害怕吧。 [Mike]: He said it's a little nervous, but very exciting. The big dream is what he calls recursive self-improvement. zh:他说确实稍微有点紧张，但也非常令人兴奋。那个大梦想，就是他所谓的"递归自我改进"。 [Sarah]: Recursive self-improvement? That sounds like a tongue twister. What does it mean? zh:递归自我改进？听起来像绕口令。这是什么意思？ [Mike]: Simply: use AI to make AI better. His co-founder Quoc Le once built a system that designs model architectures by itself. One result, the Evolved Transformer, was about 30% more efficient. zh:很简单：用 AI 来让 AI 变得更好。他的联合创始人 Quoc Le 曾经做过一个系统，能自己设计模型结构。其中一个成果，Evolved Transformer，效率大约提升了 30%。 [Sarah]: Thirty percent better, just by letting AI redesign itself? That's wild. zh:提升了 30%，只是因为让 AI 重新设计自己？太疯狂了。 [Mike]: And that's only the start. His new company, Discovery Loop, wants to automate the whole scientific loop - build, test, learn, repeat. zh:而这只是开始。他的新公司 Discovery Loop，想把整个科学循环自动化——构建、测试、学习、再来一遍。 [Sarah]: Discovery Loop... so they want AI to do science? What's the end goal? zh:Discovery Loop……所以他们是想让 AI 来做科学？最终目标是什么？ [Mike]: He described a model with the power of 20 PhDs. No single human has 20 doctorates. But a model could understand many fields at once, and send agents to solve sub-problems together. zh:他描述了一个拥有 20 个博士能力的模型。没有任何一个人能拿 20 个博士学位。但一个模型可以同时理解很多领域，并派出智能体一起解决子问题。 [Sarah]: Twenty PhDs in one machine. My brain can barely handle one topic at a time! zh:一台机器里装 20 个博士。我的大脑一次都很难应付一个话题！ [Mike]: Right? And here's the crazy part: a round of real experiments that takes a day, or even a week, could shrink to a minute, or just an hour. zh:对吧？还有更疯狂的：一轮真实的实验，原本要花一天、甚至一周，未来可能压缩到一分钟，或者仅仅一小时。 [Sarah]: A week becomes an hour? I need that for my laundry. zh:一周变成一小时？我的衣服也想这么处理。 [Mike]: Ha! But beyond jokes, he shared a real study tip. Instead of reading one paper very carefully, skim 10 papers - even 100 abstracts. zh:哈！不过说正经的，他分享了一个真正的学习建议。与其非常仔细地读一篇论文，不如快速浏览 10 篇——甚至 100 篇摘要。 [Sarah]: Wait, that feels backwards. Shouldn't we go deep, not wide? zh:等等，这感觉反了。我们不该求深，而不是求广吗？ [Mike]: His point: you want to connect ideas nobody joined before. If you know many things are becoming possible, you see a hard problem in a new light. Then you find five parts are already solvable. zh:他的意思是：你要去连接别人从没连起来的想法。如果你知道很多事正变得可行，你就能用新眼光看难题。然后你会发现，其中五个部分其实已经有解法了。 [Sarah]: Oh, so reading wide gives you puzzle pieces. Clever. zh:哦，所以广泛阅读是给你拼图的碎片。聪明。 [Mike]: He also uses quick order-of-magnitude estimates. Like: will moving this data take 10 seconds, or 100 years? Those are completely different worlds. zh:他还习惯做快速的数量级估算。比如：搬运这批数据要 10 秒，还是 100 年？那可是完全不同的两个世界。 [Sarah]: So he does math in his head like a human calculator. I like that. zh:所以他在脑子里算得像人形计算器。我喜欢这点。 [Mike]: Now the serious turn. He warned that AI in cybersecurity can already match top human hackers - maybe even go further. zh:现在说点严肃的。他警告，网络安全领域的 AI 已经能媲美顶级人类黑客——甚至可能更进一步。 [Sarah]: That sounds scary. Can AI really find holes in our systems? zh:这听起来挺吓人。AI 真能在我们的系统里找出漏洞吗？ [Mike]: Yes, on both sides. The same tool can attack, but it can also find and fix real-world holes before the bad guys do. It's a double-edged sword. zh:会，而且两边都行。同一个工具能攻击，也能在坏人动手前，找出并修补现实世界里的漏洞。这是一把双刃剑。 [Sarah]: So the tool that scares us is the same tool that protects us. Maybe the real power isn't the AI - it's the questions we choose to ask it. zh:所以吓我们的工具，也正是保护我们的工具。也许真正的力量不在 AI 身上——而在我们选择向它提出什么问题。 [Mike]: Beautifully said, Sarah. And that's a good place to pause. If you enjoyed Jeff Dean's honest talk, tell us what you think - and we'll see you next time on "Learn English with Podcasts"! zh:说得太好了，Sarah。我们就停在这里吧。如果你喜欢 Jeff Dean 这次坦诚的分享，告诉我们你的想法——下期"Learn English with Podcasts"再见！ [Sarah]: Thanks, Mike. Until next time! zh:谢谢你，Mike。下期见！

Episode: Xiaomi Q2: Another 100 Billion Yuan Quarter Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to "Learn English with Podcasts"! Sarah, guess what. I just counted the Xiaomi devices in my apartment. Do you know how many there are? Eight. My phone, my watch, my earphones, and even my rice cooker. zh:欢迎回到"Learn English with Podcasts"！Sarah，你猜怎么着。我刚数了数我公寓里的小米设备。你知道有多少个吗？八个。手机、手表、耳机，连我的电饭煲都是。 [Sarah]: Eight Xiaomi things in one apartment? That is an ecosystem, not a home. But wait, why are you telling me this today? zh:一个公寓里八件小米产品？那已经是个生态系统了，不是家。不过等等，你今天为什么跟我说这个？ [Mike]: Because this week, Xiaomi told the world how much money it makes. On August 18, it shared its numbers for the second quarter of 2026. And the first number is a big one: 108.9 billion yuan in a single quarter. zh:因为这周，小米向世界公布了它赚了多少钱。8月18日，它公布了2026年第二季度的业绩。而第一个数字就很大：一个季度营收1089亿元。 [Sarah]: 108.9 billion yuan in just three months? That is... a lot of rice cookers. zh:就三个月就营收1089亿元？那是……好多电饭煲啊。 [Mike]: Ha, it is! And here is the part that surprised me the most. The fastest-growing business is not the phone. It is the car. zh:哈，可不是嘛！而最让我意外的是这个。增长最快的业务不是手机，而是汽车。 [Sarah]: The car? Xiaomi makes cars now? I thought they only made phones. zh:汽车？小米现在做汽车了？我还以为他们只做手机。 [Mike]: They do both, and the car business is flying. In the second quarter, Xiaomi delivered 104,199 electric cars. That is more than 1,000 cars every day. zh:两个都做，而汽车业务正在起飞。二季度，小米交付了104,199辆电动汽车。相当于每天交付超过一千辆。 [Sarah]: More than 1,000 cars a day? I cannot even cook rice that fast. So does the car business make money? zh:每天一千多辆？我煮米饭都没这么快。那汽车业务赚钱吗？ [Mike]: It brought in 24.9 billion yuan this quarter. And its new SU7 is the best-selling pure electric car above 200,000 yuan in China. Its other car, the YU7, is number one for keeping its value after one year. zh:这个季度它带来了249亿元收入。而新款SU7是中国20万以上纯电轿车里的销量冠军。另一款车YU7，一年后保值率排名第一。 [Sarah]: Wait, the YU7 keeps its value better than other electric cars? A phone company's car holds its price better than real car companies' cars. I have to think about that one. zh:等等，YU7比其他电动汽车更保值？一家手机公司造的车，比真正的汽车公司的车还保值。这个我得琢磨琢磨。 [Mike]: I know, right? Now, are you ready for the second surprise? Xiaomi's phones are getting more expensive, and people are happy to pay. zh:对吧！现在，准备好听第二个意外了吗？小米的手机在变贵，而人们很乐意买单。 [Sarah]: More expensive phones, and people like it? That sounds backwards. Usually we want things to get cheaper. zh:手机变贵，大家还喜欢？这听起来反过来了。通常我们都希望东西越来越便宜。 [Mike]: Xiaomi phones now cost on average 25.9% more than last year. That is because people are buying the pricier models. Phones above 3,000 yuan now make up 32% of Xiaomi's domestic sales, a record high. zh:小米手机现在的平均售价比去年上涨了25.9%。这是因为人们在买更贵的机型。3000元以上的手机现在占小米国内销量的32%，创下历史新高。 [Sarah]: So "cheap Xiaomi" is becoming "premium Xiaomi." But does the rest of the world notice? I mean, Xiaomi is a Chinese company, right? zh:所以"平价小米"正在变成"高端小米"。那世界其他地方注意到了吗？我是说，小米是中国公司，对吧？ [Mike]: It is, but it is everywhere. Xiaomi has been in the global top three for phone shipments for 24 straight quarters. And in 53 countries, it is the number one, two, or three seller. zh:是，但它无处不在。小米已经连续24个季度稳居全球手机出货量前三。而在53个国家里，它都是销量第一、第二或第三。 [Sarah]: 53 countries! So when I travel, I will see Xiaomi shops everywhere. What about the other things, the rice cooker, the watch, the earphones? zh:53个国家！那我出国旅行时，会到处看到小米门店。那其他东西呢——电饭煲、手表、耳机这些？ [Mike]: That is the smart home side, and it is huge. This quarter, Xiaomi made 31.3 billion yuan from those products. And it says 1.16 billion devices are now connected to its smart home platform. zh:那就是智能家居这块，非常庞大。这个季度，小米从这些产品上赚了313亿元。而它说，目前有11.6亿台设备连接着它的智能家居平台。 [Sarah]: 1.16 billion devices... There are more Xiaomi gadgets on earth than there are people in Europe. My toaster feels very lonely now. zh:11.6亿台设备……地球上的小米小玩意比欧洲的人口还多。我的面包机现在觉得自己好孤单。 [Mike]: Ha! And here is where the money really comes from. Xiaomi's apps, the software inside all those phones, made 9 billion yuan. And that business has a margin of almost 77%. zh:哈！而真正赚钱的地方在这里。小米的应用——就是那些手机里的软件——赚了90亿元。而这个业务的毛利率接近77%。 [Sarah]: 77%! So selling a phone makes a little, but the software inside it makes a lot. The phone is the door, and the apps are the shop. zh:77%！所以卖手机赚得少，但手机里的软件赚得多。手机是门，应用是店铺。 [Mike]: Exactly. And Xiaomi is putting a lot of that money back into new ideas. This quarter alone, it spent 9.2 billion yuan on research. And its new AI model, MiMo V2.5, is now the most used model on OpenRouter, a popular AI platform. zh:没错。而小米把很多钱又投回新想法上。仅这个季度，它就在研发上花了92亿元。而它的新AI模型MiMo V2.5，现在成了热门AI平台OpenRouter上使用量最大的模型。 [Sarah]: Their AI model is number one? A phone-and-car company is beating the AI companies at their own game? zh:它们的AI模型排名第一？一家做手机和汽车的公司，在AI公司自己的赛场上赢了它们？ [Mike]: For this week, yes. So think about it. Xiaomi started as a cheap phone. Now it sells cars, watches, earphones... and its brain is inside everything you own. zh:至少这周是这样。所以想想看。小米从一部便宜的手机起步。现在它卖汽车、手表、耳机……而它的大脑住进你拥有的每一样东西里。 [Sarah]: You know, after hearing you, I counted my devices. My phone is Xiaomi. My lamp? Xiaomi. My toothbrush? Also Xiaomi. Maybe I should buy the car too and finish the collection. zh:你知道吗，听完你的话我也数了数我的设备。手机是小米。台灯？小米。牙刷？也是小米。也许我该把车也买了，集齐全套。 [Mike]: Ha, a full Xiaomi family! Just remember: their cars hold their value. Your toothbrush might not. zh:哈，一整套小米家族！只要记住：他们的车保值。你的牙刷可不一定。 [Sarah]: True. The car holds value, the toothbrush just holds my teeth. Thanks for listening to "Learn English with Podcasts", see you next time! zh:说得对。汽车保值，牙刷只管保住我的牙。感谢收听"Learn English with Podcasts"，下次见！ [Mike]: And hey, if you ever wonder where Xiaomi is going next... just look around your home. It is probably already there. Bye! zh:还有，如果你想知道小米下一步要去哪……就看看你家里吧。它可能已经在那了。再见！

Episode: Unitree: The First Humanoid Robot Stock Duration: approximately 7 minutes Level: B1 (Intermediate) --- [Mike]: Welcome back to "Learn English with Podcasts"! Sarah, quick question. Do you remember the little robot dogs that danced on TV during the Spring Festival? zh:欢迎回到"Learn English with Podcasts"！Sarah，快问快答。你还记得春晚在电视上跳舞的小机器狗吗？ [Sarah]: Of course! Little white dogs moving in perfect rhythm. My nephew watched them again and again. He said they were cooler than real dogs. zh:当然记得！动作整齐划一的小白狗。我侄子反复看了好多遍。他说它们比真狗还酷。 [Mike]: Those dancing dogs have a company name, and it is Unitree. This week, Unitree is doing something huge. It is selling pieces of itself to the public. In one word: it is going public. zh:那些跳舞的小狗有一个公司名字，叫宇树科技。本周，宇树要干一件大事。它要把自己的一部分卖给公众。用一个词说：它要上市了。 [Sarah]: Going public? That means the company is joining the stock market, right? So people can buy and sell it? zh:上市？就是说这家公司要进股票市场了，对吧？人们可以买卖它？ [Mike]: Exactly. On August 19, Unitree will start trading on Shanghai's STAR Market. And it is the first humanoid robot company to list on the Chinese stock market. zh:没错。8月19日，宇树将在上海的科创板开始交易。它也是中国股市上第一家上市的人形机器人公司。 [Sarah]: "Humanoid" — a robot that looks like a person. But wait, those dogs do not look human at all! zh:"人形"——就是长得像人的机器人。可是等等，那些小狗一点也不像人啊！ [Mike]: Ha, good catch. Unitree makes both kinds. Four-legged robots like the Go2, and two-legged human-shaped robots called the H1. The H1 can walk, run, and even climb stairs. zh:哈，问得好。宇树两种都做。有像Go2这样四条腿的机器人，也有叫H1的双腿人形机器人。H1能走路、跑步，甚至还会爬楼梯。 [Sarah]: A robot family: some on four legs, some on two. So what exactly happens when a company goes public? Can I just buy one share? zh:一个机器人大家庭：有的四条腿，有的两条腿。那公司上市到底会发生什么？我能直接买一股吗？ [Mike]: Yes, you can. When a company goes public, it sells shares — small pieces of itself — to anyone. One Unitree share costs 150.80 yuan. Buy one, and you own a tiny part of the company. zh:可以。公司上市时，会把股份——也就是它自己的一小部分——卖给所有人。宇树一股卖150.80元。买一股，你就拥有这家公司的一小部分。 [Sarah]: One hundred fifty yuan for a piece of a robot company? That sounds much cheaper than an actual robot dog. zh:150元就能买机器人公司的一小部分？听起来比真正的机器狗便宜多了。 [Mike]: An actual robot dog costs tens of thousands of yuan. But the whole company is worth about 61 billion yuan. That is a lot of robot dogs! zh:真正的机器狗要好几万元。而整个宇树市值大约610亿元。那是好多机器狗啊！ [Sarah]: Wait, but does Unitree really earn 61 billion yuan a year? That cannot be right. zh:等等，可宇树一年真的能赚610亿元吗？那不可能吧。 [Mike]: You are right, it does not. And this is the interesting part. People are not just buying what Unitree makes today. They are buying what robots will do tomorrow. zh:你说得对，它赚不了那么多。而有趣的地方就在这里。人们买的不仅是宇树今天做的东西，更是机器人明天能做的事。 [Sarah]: So the share price is like a bet on the future? zh:所以股价就像是对未来的一场赌注？ [Mike]: A very expensive bet. Unitree's price is 219 times its yearly profit. The industry average is only about 38 times. That means the market believes this robot company is special. zh:一场很贵的赌注。宇树的市盈率是219倍。行业平均只有大约38倍。这说明市场相信这家机器人公司很特别。 [Sarah]: Ah, the P/E ratio. Price divided by yearly profit. A high P/E means people expect fast growth. But is Unitree even making money now? zh:啊，市盈率。价格除以年利润。市盈率高，说明人们预期它会快速增长。但宇树现在到底赚钱吗？ [Mike]: Good news: yes. In 2023, it lost money — small sales and a loss. Then in 2024, it turned a profit. And in 2025, sales jumped to 1.7 billion yuan, with a profit of 278 million. zh:好消息：赚钱。2023年它还在亏钱——销售额小，还有亏损。然后2024年它扭亏为盈。到了2025年，销售额跳到17亿元，利润2.78亿元。 [Sarah]: From a loss to 1.7 billion in two years. That is one fast-growing robot family. So what will Unitree do with all the money it raises? zh:两年时间从亏损到17亿销售额。这个机器人家庭成长得真快。那宇树要拿募来的这些钱做什么呢？ [Mike]: It plans to raise about 6.1 billion yuan. And here is the surprise: almost half of that money, about 2 billion yuan, is going to one thing — smarter AI models for its robots. zh:它计划募资约61亿元。而让人意外的是：这笔钱里几乎一半，大约20亿元，会投到一件事上——为它的机器人打造更聪明的AI模型。 [Sarah]: Not new robot bodies? The money is for the brain, not the legs? zh:不是造新的机器人身体？钱是给大脑的，不是给腿的？ [Mike]: Right. Unitree wants its robots to do more than dance. It wants them to open doors, pick up boxes, and work in factories. That needs a smart brain, not just strong legs. zh:对。宇树想让它的机器人做的比跳舞更多。它想让它们开门、搬箱子、在工厂里干活。这需要聪明的大脑，而不只是强壮的腿。 [Sarah]: So the dog that dances today is learning to work tomorrow. And who else is buying in? Only small investors like us? zh:所以今天跳舞的狗，明天正在学着干活。那还有谁在买？只有像我们这样的小投资者吗？ [Mike]: No, some big names joined too. The social security fund bought shares. And so did an AI company called DeepSeek — the same DeepSeek that made the famous chatbot. zh:不，还有一些大机构也加入了。社保基金买了股份。一家叫DeepSeek的AI公司也买了——就是做出那个著名聊天机器人的DeepSeek。 [Sarah]: Wait, the chatbot company is buying robot shares? The "brain" company wants a piece of the "body" company. This is starting to sound like a family reunion. zh:等等，做聊天机器人的公司来买机器人的股票？"大脑"公司想要"身体"公司的一份。这越来越像一场大团圆了。 [Mike]: A family reunion of tech companies, exactly. And in two days, the dancing dogs officially become a stock market star. zh:一场科技公司的大团圆，没错。而两天后，这些跳舞的小狗就正式成为股市明星了。 [Sarah]: You know what my nephew would say? Last year he told me, "Those dogs are going places!" And look at them now. They really are going places — all the way to the stock market. zh:你知道我侄子会说什么吗？去年他告诉我："那些狗要去好地方了！"再看看它们现在。它们真的去好地方了——一路走到了股票市场。 [Mike]: Ha! Literally going public. So if you believe robots will change the world, you can buy a tiny piece of that belief for 150.80 yuan. Cheaper than a robot dog... and almost as loyal. zh:哈！字面意义上的"上市"。所以如果你相信机器人会改变世界，你可以花150.80元买下这份信念的一小片。比机器狗便宜……而且几乎一样忠诚。 [Sarah]: Almost. A share will not bring you your slippers. Not yet. Thanks for listening to "Learn English with Podcasts" — see you next time! zh:差一点。股票现在还不会帮你叼拖鞋。感谢收听"Learn English with Podcasts"——下次见！ [Mike]: And remember: today you buy the shares, tomorrow the robots do the dishes. Bye! zh:还有记住：今天你买股票，明天机器人来洗碗。再见！