Shadowing Practice: A $6.3 billion open-weight model just got embarrassed by the French... - Learn English Speaking with Video
Choose a practice mode
Each mode opens a full-screen practice room.
63 sentences
If sentences are too short or too long, click Edit to adjust them.
1
Yesterday, while you were being force -fed every iteration of that Rumpelstiltskin video, the Paris -based AI Lab Mistral released Mistral Large 4,
2
a trillion -parameter open -weight model that they're claiming is the best of its kind, assuming you adopt the American ideology of just ignoring any technology coming out of China.
3
It's natively multimodal, has a million -token context window, and claims to blah blah blah.
4
The more interesting part of this release was the timing, because it landed in the middle of what many are calling the dumbest week in the history of the open weight race.
5
It kicked off early in the week
6
when Trump announced a super intelligence force whose job it is to keep America at the frontier of the AI race,
7
with their first order of business claiming that AI is actually dumb and the real future is SI.
8
Then Reflection AI, the $25 billion Nvidia -backed startup that calls itself America's Open Frontier Lab, will finally ship their first model after two years of fundraising.
9
It's called Beam, it has 501 billion parameters, it was trained on 6 .3 billion dollars worth of GPUs, it benchmarked itself against an already outdated Chinese model,
10
and 24 hours later it got out -benchmarked itself by Mistral.
11
Then, just yesterday, Moonshot AI, the Chinese lab behind Kimi, closed its final funding round at a 50 billion dollar valuation and started prepping a Hong Kong IPO,
12
which I'm told the superintelligence force has no interest investing in.
13
In today's video, we'll look at what the French, the Americans, and the Chinese each ship this week.
14
Find out why every government on earth suddenly wants an open model to call its own, and figure out which one you should actually use.
15
It is October 7th, 2026, and you're watching the Code Report.
16
If you're still not convinced about the benefits of open -weight models, last month a man in San Francisco learned them the hard way.
17
When he had the authorities looking for him, after a joking conversation he had with Claude triggered a safety review, and just last week, a woman in Florida got arrested just days after telling Claude
18
she was going to do something that's typically frowned upon in modern society.
19
And even if you're not planning on committing a felony, with proprietary models, you're still sending every prompt to a data center, which is scanned by a classifier, but potentially read by a safety team,
20
and then used in whatever way the lab wants, like training future models or kicking off the process to throw you in jail.
21
And whole -ass countries are having the same problem too.
22
Take a government that wants to use AI to find issues in their own power grid.
23
Using a model like Mythos, which is designed for this exact purpose, would require handing a foreign company your private schematics and then hoping
24
that they never get leaked or that someone on their payroll isn't a foreign spy, which Dario himself has said may already be the case, where if you were using an open -weight model,
25
most of these problems go away, which is why nearly every country in the world is trying to build their own if they can, or find ones they can trust if they can't.
26
Which brings us back to this week and the great model mid -off.
27
The first announcement came from Reflection AI, which was started by two DeepMind researchers who raised $4 .6 billion, a billion with a B,
28
on the premise that the best open model in the world should come from the West, and Llama 4 wasn't cutting it.
29
Well, this week, we finally got to witness Manifest Destiny realized when they released Beam.
30
It's got 501 billion parameters, and its base model was trained in just under four weeks on about 24 trillion tokens.
31
After that, they spent four more weeks post -training it to get it better at writing code, using tools, and finishing long agent tasks.
32
The result is a model that Reflection says keeps up with Zhipu AI's GLM 5 .2, while needing only a quarter of the hardware.
33
But sadly, as of today, you still need an invite to use it, and the waits aren't actually out yet.
34
Which means the best open American model is somehow still Gemma 4, which Google released under Apache 2 .0 back in April.
35
Then the next day, Mistral showed up with something that sounds like you'd order it from McDonald's.
36
It's a mixture of experts model, which means it has about a trillion parameters, but only 49 billion of them actually turn on for a given token.
37
So the idea is
38
that you get a trillion parameters worth of knowledge for roughly the cost of running a model 20 times smaller, which if that sounds familiar,
39
it's probably because it's the same trick DeepSeq used last year to tank NVIDIA.
40
On the TMBBs, it barely beats GLM 5 .3 at long coding jobs, it beats Beam by 18 points and lands in the top five for finding security bugs,
41
which is the one Mistral actually cares about
42
because the power grid scenario from earlier is the thing they're actually trying to sell.
43
The catch is the numbers really are, trust me bros, because they were provided by Mistral
44
and they actually haven't finished training yet and the weights don't come out until later this month under their own custom license.
45
Which brings us to China, which already has the thing both of them are trying to build with Kimi K3.
46
It's a 2 .8 trillion parameter open model that Moonshot released back in July,
47
which became so popular they had to stop taking new subscribers two days later because they ran out of GPUs.
48
Since then, Microsoft, Amazon, and Google have all been negotiating to host it.
49
Anthropic accused them of running 300 ,000 requests through fake accounts to copy Claude, and the company went from a $4 billion valuation last November
50
to $50 billion yesterday with China's own state AI fund involved in the round.
51
So which one should you actually use though?
52
If all you've got is a gaming PC, go with Gemma 4 and get your money up.
53
If you have your own data center and no compliance department, go with Kimi K3.
54
Or if you have a data center and a compliance department, then go with Mr. Large 4.
55
And if you're a patriot, Beam's weights are coming later this month.
56
But no matter which model you choose, they'll need to get on the internet, which they can do with Kernel, the sponsor of today's video.
57
They provide open source browser infrastructure that lets your agents act on the internet blazingly fast.
58
But none of that actually matters if your agent keeps getting blocked by bot checks, which is why they just launched Config Registry.
59
Send it the URL you want to access from either their dashboard or the API, and it'll give you back the browser and proxy settings that have worked for that site.
60
Right now, my agent keeps getting stopped by a bot check, but when I spin up a browser with the config Kernel recommends, it's able to access the page without any issues.
61
Kernel builds these configs by continuously testing real browser and proxy setups against the live site, and each one comes with a success rate, sample size, and last verified date.
62
Use the code Fireship at the link below to get $30 in free credits, which covers your first month on the hobbyist plan this has been the code report thanks for watching
63
and i will see you in the next one
Download App
Everything you need to speak fluently
AI PronunciationScore every sentence
IPA PracticeMaster every sound
VocabularyBuild your word bank
Vocab GameLearn while playing
Same Channel
C11
PostgreSQL in 100 Seconds
Daily ConversationWorking Adults
C11
React Native in 100 Seconds
Daily ConversationUniversity Students
C12
Did Google just dethrone OpenAI and Anthropic?
Daily ConversationUniversity StudentsWorking Adults
C13
Every app on Earth is now open source... whether it likes it or not
Daily ConversationWorking Adults
C14
Brainf**k in 100 Seconds
Daily ConversationUniversity Students
C13
OpenAI’s lead engineer was secretly a spy...
Daily ConversationUniversity StudentsWorking Adults
C19
The most interesting "hack" in history...
American AccentBusiness EnglishUniversity StudentsWorking Adults
C12
Mind-bending new programming language for GPUs just dropped...
Daily ConversationUniversity Students
C12
China just dropped a new frontier model... and nobody can explain how
Daily ConversationUniversity StudentsWorking Adults
C12
The last real programmers are dying out...
Daily ConversationUniversity StudentsWorking Adults
C11
Did the AI bubble just pop?
Daily ConversationUniversity StudentsWorking Adults
B26
Intel chips can’t possibly be this bad… 100% crash rate?
American AccentBusiness EnglishUniversity StudentsWorking Adults
B23
OpenAI’s new browser feels familiar…
American AccentBusiness EnglishUniversity StudentsWorking Adults
C130
Docker in 100 Seconds
American AccentBusiness EnglishUniversity StudentsWorking Adults
B23
Firefox and Tor hit with 9.8 critical level exploit
American AccentBusiness EnglishUniversity StudentsWorking Adults
B23
5 open source tools that replaced my $320/mo AI stack...
American AccentBusiness EnglishUniversity StudentsWorking Adults
C12
C++ in 100 Seconds
American AccentBusiness EnglishUniversity StudentsWorking Adults
C12
Rust in 100 Seconds
American AccentBusiness EnglishUniversity StudentsWorking Adults
B24
Linux in 100 Seconds
American AccentBusiness EnglishUniversity StudentsWorking Adults
C15
The plan to break apart Google... RIP Chrome
American AccentBusiness EnglishDaily ConversationUniversity Students
FOR YOU
Recommend Video
C11
462 sentences
Expensive RTX 5090 for LLMs? NO. Use This Instead. (Tesla V100 SXM2 + Z8 G4, #RACERRRZ)
Daily ConversationUniversity StudentsWorking Adults
C141 sentences
No, Really — Is A.I. a Bubble? | The Ezra Klein Show
Daily ConversationUniversity StudentsWorking Adults
C12
171 sentences
Why the Arab World never united
Daily ConversationUniversity StudentsWorking Adults
C12
182 sentences
We Let $200 GPT-Pro Think for 7 Days on Unsolved Math
Daily ConversationUniversity StudentsWorking Adults
C11
198 sentences
Jack Dorsey’s New Dessa AI Voice Sounds Scary Human
Daily ConversationUniversity StudentsWorking Adults
C12
136 sentences
Is ML Engineering Dead? (AI vs ML Engineer)
Daily ConversationUniversity StudentsWorking Adults
C11
407 sentences
Why AI Tokens are so Expensive - Computerphile
Daily ConversationUniversity StudentsWorking Adults
C11
83 sentences
Leading AI CEOs Warn UN Security Council of Existential Risks | Sam Altman On AI Controls | N18G
Business EnglishDaily ConversationUniversity Students
C11
510 sentences
Inside the secret race to build artificial intelligence | The Gray Area
Daily ConversationUniversity StudentsWorking Adults
C14
168 sentences
The Rise Of Open-Source Software
Daily ConversationUniversity StudentsWorking Adults
Vocabulary and speaking notes for this lesson
This speaking lesson is built on the video “A $6.3 billion open-weight model just got embarrassed by the French..”. The speaker keeps coming back to these words: model, billion, Mistral, parameter, weight. This video has 63 sentences and 1255 words to shadow. The speech runs for 6:09. The speaker talks fast, about 204 words per minute, so expect linked and reduced sounds. 82% of the words are among the 3,000 most common in English; the rest is worth studying before you start.
Key vocabulary in this video
15 words from the video worth learning, with pronunciation and meaning:
| Word | Pronunciation | Meaning |
|---|---|---|
| parameter noun | /pəˈɹæm.ə.tɚ/ | A value kept constant during an experiment, equation, calculation, or similar, but varied over other versions of the experiment, equation, calculation, etc. |
| browser noun | /ˈbɹaʊ.zəɹ/ | One who or which browses. |
| token noun | /ˈtoʊkən/ | Something serving as an expression of something else. |
| grid noun | /ɡɹɪd/ | A rectangular array of squares or rectangles of equal size, such as in a crossword puzzle. |
| compliance noun | /kəmˈplaɪ.əns/ | An act of complying. |
| frontier noun | /ˈfɹʌntɪə/ | The part of a country which borders or faces another country or unsettled region. |
| valuation noun | /ˌvæ.ljuːˈeɪ.ʃən/ | An estimation of something's worth. |
| proxy noun | /ˈpɹɑk.si/ | An agent or substitute authorized to act for another person. |
| benchmark noun | /ˈbɛn(t)ʃmɑːk/ | A standard by which something is evaluated or measured. |
| ignore verb | /ɪɡˈnɔː/ | To deliberately not listen or pay attention to. |
| trick noun | /ˈtɹɪk/ | Something designed to fool, dupe, outsmart, mislead or swindle. |
| infrastructure noun | /ˈɪnfɹəˌstɹʌkt͡ʃə/ | An underlying base or foundation for a building, organization, or system. |
| custom noun | /ˈkʌstəm/ | Frequent repetition of the same behavior; way of behavior common to many; ordinary manner; habitual practice; method of doing, living or behaving. |
| potentially adverb | /pəˈtɛnʃ(ə)li/ | In a manner showing much potential; with the possibility of happening in a given way. |
| spin verb | /spɪn/ | To rotate, revolve, gyrate (usually quickly); to partially or completely rotate to face another direction. |
Phrasal verbs you will hear
| Word | Pronunciation | Meaning |
|---|---|---|
| kick off verb | To make the first kick in a game or part of a game. | |
| come out verb | /ˌkʌm ˈaʊt/ | To be discovered; to be revealed. |
| figure out verb | To come to understand; to discover or find a solution; to deduce. | |
| find out verb | To discover, as by asking or investigating. | |
| get out verb | To leave or escape. | |
| go away verb | To depart or leave a place. | |
| keep up with verb | To manage to remain beside or just behind someone or something that is moving away from one. |
Grammar in this video
The structures the speaker uses most, with the exact words from the video:
| Structure | In the video |
|---|---|
| Passive voice be + past participle — the focus is on what happens, not who does it | was trained · is scanned · is designed |
| Relative clauses who / which + clause — extra information about a person or thing | center, which is · Mythos, which is · away, which is |
| Present perfect have/has + past participle — a past action that still matters now | has said · haven't finished · have worked |
Pronunciation to watch
The speaker uses 19 contractions and reduced forms, such as you're, they're, aren't. Say them the short way, as you hear them.
- The “sh” and “zh” sounds: valuation /ˌvæ.ljuːˈeɪ.ʃən/, benchmark /ˈbɛn(t)ʃmɑːk/, potentially /pəˈtɛnʃ(ə)li/, negotiate /nəˈɡəʊ.ʃi.eɪt/, dashboard /ˈdæʃˌboɹd/
- Long words — get the stress right: parameter /pəˈɹæm.ə.tɚ/, valuation /ˌvæ.ljuːˈeɪ.ʃən/, infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, potentially /pəˈtɛnʃ(ə)li/, ideology /aɪ.diˈɑ.lə.d͡ʒi/
How to practise with this video
- Listen to the whole video once without speaking and note the words you do not know.
- Start at 0.75× speed, shadow it sentence by sentence, then go back to normal speed once it feels easy.
- Record yourself and compare with the original, paying attention to words like parameter, browser, token.
What is the Shadowing Technique?
Shadowing is a science-backed language learning technique originally developed for professional interpreter training and popularized by polyglot Dr. Alexander Arguelles. The method is simple but powerful: you listen to native English audio and immediately repeat it out loud — like a shadow following the speaker with just a 1–2 second delay. Unlike passive listening or grammar drills, shadowing forces your brain and mouth muscles to simultaneously process and reproduce real speech patterns. Research shows it significantly improves pronunciation accuracy, intonation, rhythm, connected speech, listening comprehension, and speaking fluency — making it one of the most effective methods for IELTS Speaking preparation and real-world English communication.