跟讀練習: Did Anthropic just save its $1 trillion IPO? - 通過影片學習英語口語
選擇練習模式
每種模式都會開啟全屏練習室。
74 句
如果句子過短或過長,請點選 Edit 進行調整。
1
While Honeypot Wong was busy falling head over leather jacket for a succulent Chinese spy, These Chinese models are excellent.
2
Dario was putting the final touches on his prep work for the Anthropic IPO roadshow that started yesterday.
3
Unfortunately, it's coming at a pretty rough time for Dr. Doomsday.
4
The U .S midterms turned into a referendum on data centers where voters made it clear they hate them,
5
and Wall Street responded by panic selling every AI stock it could find.
6
That dragged Anthropik's expected valuation down from $2 trillion in September to a mere $1 trillion today.
7
So yesterday, Dollar Store Dario gave investors something else to look at by releasing Fable and Mythos 5 .2,
8
which you'll never believe are faster and cheaper than the versions that just came out two months ago.
9
During testing, it built an entire operating system starting from an empty folder, it finished a math project in 30 hours that Opus failed to finish,
10
and it even designed a working Tamagotchi.
11
In today's video, we'll look at everything these new models can do and answer the question I know you've all been wondering.
12
Can it create an interactive thoroughbred with an appropriate amount of realness?
13
It is November 10th, 2026, and you're watching the Code Report.
14
If you don't follow the Claude Cinematic Universe closely, Fable and Mythos are the same model,
15
but Mythos is the unrestricted one reserved for large companies with budgets for cybersecurity insurance,
16
and Fable is the one with guardrails that won't let you research how to cook up smallpox.
17
Now, we'll get to the fun stuff in a second, but let's get the boring stuff out of the way first.
18
The price is now $8 in and $40 out, which is down from $10 and $50 from $5 .1,
19
but Anthropic suggests that you should stick with Opus 5 .5 for most work
20
and only swap to Fable if you really need it.
21
For the operating system demo, an engineer pointed Fable at an empty folder and let 1 ,000 subagents have a Jensen -inspired freakfest for 8 days.
22
What comes out is a fairly legit OS that boots on a real laptop with its own kernel and drivers.
23
And you may remember back in August
24
when an unreleased Claude spent 11 days turning Fermat's last theorem into lean code that a computer could verify.
25
This time, a swarm of 1 ,000 Fable agents did it in 30 hours, and the proof was 7 times shorter,
26
while Opus 5 .5 crashed out at 26 % with the same setup.
27
And apparently it's decent at hardware too.
28
Working with a circuit board company called Diode, Fable took a one -paragraph prompt and used it to design a voice assistant hub,
29
a USB speakerphone, and a handheld virtual pet.
30
On the TMBBs, it beats GPT -6 Astra everywhere Anthropic listed an OpenAI score, it loses to Gemini 4
31
Carbon on business automation, and it more than doubles Astra on a robotics benchmark that Anthropic ran itself.
32
But enough glazing.
33
As you know, there's only one benchmark I trust.
34
A benchmark that's so good, it's become the industry standard by which humanity measures its new god -in -a -box toys, Thoroughbench.
35
So I open Claudecode and type those five famous words create a realistic interactive 3D horse with 3JS.
36
Make no mistakes.
37
Fable cooked for 43 minutes, and instead of downloading a horse model like a normal developer,
38
it sculpted the entire animal out of math using signed distance fields
39
that get meshed up and rigged to a skeleton every time the page loads.
40
It even screenshotted its own first attempt and roasted it.
41
The mane is bluish, the tail is stick -like, and the neck and hind legs are thin.
42
Then it went back and fixed every one of them.
43
The end result is the most realistic horse Thoroughbench has ever seen.
44
The skin texture and lighting look great, its head follows your mouse cursor,
45
and you can steer it with the arrow keys or orbit around it in 3D.
46
It's very cute and friendly, but gets extremely shy if you get too close.
47
The transitions between gates are buttery smooth, from a standstill all the way to a gallop,
48
although it's a little more awkward when coming to a full stop.
49
Under the hood, Opus 5 .5's horse had 95 ,000 vertices and 35 bones, while Fables has over 120 ,000 vertices and 38 bones,
50
and it plants each hoof with inverse kinematics, so unlike Kimmy, it's not just warping the mesh.
51
Fable 5 .2 didn't hallucinate a giant low -poly phallus like Gemini, but it equipped our horse with those rear leg shock absorbers,
52
which probably come in very handy when the horse decides to rear, as you can see here.
53
And the mane is a little lopsided, unless this was a custom request for his barber.
54
And even though I told it to make no mistakes, it admitted the horse is not photoreal, and the eyes are half -lidded.
55
So the final score for Fable, 5 .2 on Thoroughbench, is a 7 .77, making it the new state -of -the -art in horse.
56
And this really got me thinking, the models are starting to get so good
57
that we should really be naming the horses they create to keep track of them.
58
So we've decided to call this one TypeScript in honor of Secretariat, the greatest racehorse of all time,
59
who I can only assume would have loved to type if it had fingers.
60
And if you want to take him for a ride yourself, he now lives at typescript .dev, where you can walk, trot,
61
and gallop them right in your browser.
62
And if you ever need to run a video in your browser, you can do so with MUX, the sponsor of today's video.
63
Their highly customizable API is by far the easiest way to add video features to your application.
64
We've used it for years to host our courses, but it does a lot more than just infrastructure.
65
When you upload a video to MUX, you automatically get transcripts, storyboards, thumbnails, and clips.
66
From there, MUX Robots gives you AI workflows that can find key moments or pick the best thumbnail.
67
without you needing to host a model or maintain a pipeline.
68
You can automate all of this with directives where you define a workflow once and it runs on every new upload.
69
And robots keep up with new AI models automatically
70
so you can safely ignore the next video about whichever model just changed everything.
71
Perplexity, Patreon, and many other prestigious companies all trust MUX, and you can try robots on your own video in the dashboard.
72
The free plan includes 100 ,000 robots units and delivery minutes every month,
73
and you get an extra $50 credit at mux .com slash fireship or at the link below.
74
This has been The Code Report, thanks for watching, and I will see you in the next one.
同頻道
FOR YOU
推薦影片
本課的詞彙與口語要點
這節口語課以影片“Did Anthropic just save its $1 trillion IPO?”為素材。 影片中反覆出現的詞有:Fable, horse, model, mux。 這段影片共有 74 個句子、1127 個單詞可供跟讀。 講話部分時長為 5:31。 說話人語速較快,每分鐘約 204 個詞,會出現較多連讀和弱讀。 只有 74% 的單詞屬於英語最常用的 3,000 詞,詞彙難度較高。
影片中的重點詞彙
影片中 15 個值得學習的單詞,附發音和釋義:
| 單詞 | 發音 | 釋義 |
|---|---|---|
| robot 名詞 | /ˈɹoʊ.bɑt/ | 機器人 /機器人, 機械人 /機械人 |
| bone 名詞 | /boʊn/ | гўдў, гўдўзы |
| realistic 形容詞 | /ˌɹɪəˈlɪs.tɪk/ | 現實 /現實, 栩栩如生 |
| upload 動詞 | /ˈʌpˌloʊd/ | 上載 /上載, 上傳 /上傳 |
| mesh 名詞 | /mɛʃ/ | 網孔 /網孔 |
| folder 名詞 | /ˈfoʊldɚ/ | 卷宗, 資料夾 /資料夾 |
| finger 名詞 | /ˈfɪŋɡɚ/ | 手指, 指頭 /指頭 |
| ignore 動詞 | /ɪɡˈnɔː/ | 無視 /無視, 忽略 |
| gate 名詞 | /ˈɡeɪ̯t/ | 門口 /門口, 門 /門 |
| thin 形容詞 | /ˈθɪn/ | бә, 薄 |
| infrastructure 名詞 | /ˈɪnfɹəˌstɹʌkt͡ʃə/ | 基礎設施 /基礎設施 |
| shock 名詞 | /ʃɔk/ | 衝擊 /衝擊 |
| transition 名詞 | /tɹænˈzɪʃ.ən/ | 過渡 /過渡, 轉變 /轉變 |
| custom 名詞 | /ˈkʌstəm/ | 習慣 /習慣, 習俗 /習俗 |
| decent 形容詞 | /ˈdiː.sənt/ | 體面 /體面, 正派 |
影片中出現的短語動詞
| 單詞 | 發音 | 釋義 |
|---|---|---|
| come out 動詞 | /ˌkʌm ˈaʊt/ | 出, 出來 /出來 |
| go back 動詞 | 回去, 返回 |
需要注意的發音
說話人用了 10 次縮略和弱讀形式,例如 we'll, we've, didn't。請按聽到的簡短形式來說。
- “th” 音: math /mæθ/, thin /ˈθɪn/, smooth /smuːð/, leather /ˈlɛðɚ/, theorem /ˈθiərəm/
- “sh” 和 “zh” 音: benchmark /ˈbɛn(t)ʃmɑːk/, mesh /mɛʃ/, shock /ʃɔk/, transition /tɹænˈzɪʃ.ən/, automation /ˌɔ.təˈmeɪ.ʃən/
- 長單詞——注意重音位置: interactive /ˌɪn.tɚˈæk.tɪv/, infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, humanity /hjuˈmænɪti/, developer /dɪˈvɛləpɚ/, referendum /ˌɹɛfəˈɹɛndəm/
中文母語者容易讀錯的音:
- /v/ — 不要讀成 /w/,上齒輕觸下唇: interactive /ˌɪn.tɚˈæk.tɪv/, virtual /ˈvɜɹt͡ʃuəl/, developer /dɪˈvɛləpɚ/, voter /ˈvoʊtɚ/, investor /ɪnˈvɛs.tə(ɹ)/
- 詞尾子音 — 要讀清楚,後面不要加母音: robot /ˈɹoʊ.bɑt/, benchmark /ˈbɛn(t)ʃmɑːk/, math /mæθ/, realistic /ˌɹɪəˈlɪs.tɪk/, interactive /ˌɪn.tɚˈæk.tɪv/
- /r/ 和 /l/ 的區別: realistic /ˌɹɪəˈlɪs.tɪk/, virtual /ˈvɜɹt͡ʃuəl/, guardrail /ˈɡɑɹdˌɹeɪl/, perplexity /pəɹˈplɛksəti/
如何用這段影片練習
- 先完整聽一遍影片,不要開口,記下不認識的單詞。
- 先用 0.75 倍速逐句跟讀,熟練之後再回到正常速度。
- 錄下自己的聲音並與原聲對比,特別注意 robot, bone, realistic 這類單詞。
什麼是跟讀法?
跟讀法 (Shadowing) 是一種有科學依據的語言學習技巧,最初開發用於專業口譯員的培訓,並由多語言者Alexander Arguelles博士普及。這個方法簡單而強大:您在聽英語母語原聲的同時立即大聲重複——就像是一個延遲1-2秒緊跟說話者的影子。與被動聽力或語法練習不同,跟讀法強迫您的大腦和口腔肌肉同時處理並模模擬實的講話模式。研究表明它能顯著提高發音準確性,語調,節奏,連讀,聽力理解和口語流利度——使其成為雅思口語備考和真實英語交流最有效的方法之一。































