쉐도잉 연습: Meta's new model wants "deep access" to your personal life... - 영상으로 영어 말하기 배우기

연습 모드 선택

각 모드는 전체 화면 연습실로 열립니다.

57 문장

문장이 너무 짧거나 길면 Edit를 눌러 조정하세요.

On Monday, the company that spent the last year starving Llama, poaching researchers with nine-figure Zuck bucks, and locking its new models behind a Facebook auth page,

just released a free and open-source model under the Apache 2.0 license.

It's called Muse Glimmer, a name I believe was inspired by Zuck's fanfiction My Little Pony character,

and it's a 30 billion parameter agentic model that's small enough to run on your poverty spec PC.

That means you can now have an always-on mini Zuck agent living rent-free in your gaming PC, reading your email, managing your calendar,

and using your data to overthrow local elections in middle America.

And according to Meta, an agent like that needs deep access to personal context, which is a sentence they've been dreaming about writing since 2004,

except this time, the surveillance can run entirely on your own hardware.

In today's video, we'll find out what Muse Glitter Glimmer can actually do, learn how Meta shrunk it down with distillation,

quantization, and speculative decoding, and determine whether Zuck's open-source redemption arc is sincere, or if this is just what losing looks like right now.

It is August 12th, 2026, and you're watching The Code Report.

It wasn't long ago, when Meta was seen as the champion of the open weight class.

Their Llama models lopped up over a billion downloads and created an entire ecosystem of custom models built on top of them.

That is, until Llama 4 was released, and it became the biggest Facebook bust since Justin Timberlake partied with those underage girls.

In the release were two models, Scout and Maverick.

Initially, they benchmarked well, but that was only

because Meta submitted a secret juiced up version to LMArena that none of us normies could actually download.

And the models we could download had the intelligence of farm animals, which was fitting given their names.

And while this was happening, comrade Xi Jinping was dropping open models that were better, cheaper, and released in accordance with Mao Zedong's thought.

Then Zuck responded to this humiliation the only way a billionaire knows how.

He spent $14 billion on a 49% stake in scale AI just to acquire its CEO, Alexander Wang.

Then he went on a poaching spree for researchers at OpenAI and Google, and rebranded the whole operation as Meta Super Intelligence Labs.

The first thing his new dream team did was rebuild the stack from scratch and abandon open source entirely, shipping MuseSpark earlier this year as a closed API-only model.

That created a new problem, though, because no sane person would choose Muse Spark over something like Claude or Gemini.

So with Wall Street starting to ask questions about the $145 billion of CapEx being incinerated with nothing to show for it,

I assume Zuck flew to a secret compound in Kauai that the locals say he stole from them and formed a plan, which eventually led us here to the launch of Muse Glimmer.

It's a dense 30 billion parameter model distilled directly from Muse Spark, their big but closed model using logit distillation, which basically means they had the big model whisper its exact

probability distributions into the ear of Glimmer until it learned how to fake it.

And in Zuck's manifesto he made after the release, he described distillation as an important principle of how the open-source ecosystem works, which is ironic

because every American lab spent the last year accusing the Chinese of doing this exact same thing to their outputs.

So how good is it actually?

Well, according to the TMBBs, it clearly beats Gemma 4 and goes bar for bar with Quen 3.6.

But my favorite number is the prompt injection benchmark, where attacks against Glimmer succeeded 28% of the time, but they're listing that as a win because with Quen, it's 40%.

And the way they got it to run on consumer GPUs was interesting.

At full precision, the model would need over 55 gigs of memory, so their first move was quantization, where Meta compressed the weights down to about 4 bits,

which shrank it to just under 20 gigs.

And the second was speculative decoding, which is basically autocomplete for your autocomplete.

The way it works is a tiny model called D-Flash blurts out an entire block of tokens, then the big model reads the whole block in one pass, and throws out the bad guesses,

which they say led to a 3x speedup on a 5090.

But to be honest, the actual model is way less interesting than Zuck's manifesto I mentioned earlier.

In it, he argues that the real risk in AI isn't rogue superintelligence, but a small handful of companies owning it,

and that he doesn't understand why anyone who believes AI will end humanity would rush to build it, which is a reasonable point coming from a guy whose company

was just fined $567 million last week for being a public nuisance in New Mexico.

He also wants Frontier Labs to hand the US government mid-training checkpoints of unreleased models, and he announced a billion dollar fund for the towns willing to host his data centers,

probably so they'll stop causing all that ruckus.

So, is the redemption arc sincere?

Probably not, but an Apache license is an Apache license, and they can't put that genie back in the bottle.

Zuck and Wang both say open waits for MuseSpark 1.2 are coming soon, which would mean you could self-host the exact model behind Meta's own coding agent, which is pretty cool.

And if you want to test out Muse Glimmer for yourself, you can do it with the sponsor of today's video, OpenRouter.

It gives you a single API to access every LLM, so you only have one endpoint and one bill to worry about.

You can switch between models manually, or use their routers to pick one for you based on a combination of factors, like price and accuracy.

My startup still hasn't been able to raise a seed round, so I used Open Router to rewrite the entire codebase in Rust to attract more investors.

Because Rust is a notoriously easy language, I used Pareto Router to route every request to cheaper coding models that could still cargo build.

But I eventually ended up splurging for some fancier models for the UI,

the dedicated image and speech endpoints to generate a few especially attractive profiles to seed the community and get the matches flowing.

Open Router is easily the best tool I've seen for discovering new models without getting stuck in 25 different subscriptions, and you can try it out for free at the link below.

This has been The Code Report, thanks for watching, and I will see you in the next one.

앱 다운로드

유창하게 말하는 데 필요한 모든 것

AI 발음 평가문장마다 점수 확인

IPA 발음 연습모든 소리 마스터

단어 학습나만의 단어장 만들기

단어 게임놀면서 외우기

QR 스캔

Download on theApp Store Get it onGoogle Play

같은 채널

FOR YOU

이 비디오로 배울 실용적인 영어 말하기 기술

이 영상은 빠른 속도로 진행되는 대화에서 복잡한 정보를 정확히 전달하는 기술과 비판적인 내용을 유머와 함께 표현하는 말투, 전문 용어를 자연스럽게 섞어 사용하는 방법을 연습할 수 있습니다. 특히 "유튜브 영어 공부"를 하며 실생활에서 자주 쓰이는 구어체 표현을 익히기에 좋습니다.

듣고 주목해야 할 발음 현상

영상에서 자주 나타나는 연결 발음과 축약형을 주의깊게 듣습니다. 예를 들어 "it's called"가 /its kɔːld/가 아닌 /its kɔːld/로 연결되거나, "that means"가 /ðæt miːnz/에서 /ðæt miːnz/로 축약되는 것을 확인하세요. 또한 "poaching spree"에서 /ˈpoʊtʃɪŋ spriː/의 스트레스가 "poaching"에 가해지는 것을 느껴보세요. 이러한 현상은 "shadow speech" 연습 시 정확히 모방해야 합니다.

원어민처럼 말하는 비결: 리듬과 스트레스

원어민처럼 말하기 위해선 리듬과 스트레스를 중점으로 연습합니다. 이 영상의 화자는 중요한 정보를 강조하기 위해 단어의 스트레스를 강하게 하고, 부가적인 내용은 빠르게 넘깁니다. 예를 들어 "Meta's new model wants 'deep access' to your personal life"에서 "deep access"와 "personal life"에 스트레스를 주어 강조합니다. "shadowing site"를 이용해 반복적으로 따라하며, 화자의 말속도와 강세를 정확히 모방하세요. 또한 "speculative decoding"과 같은 전문 용어는 발음을 확실히 하면서도 자연스럽게 연결하는 연습이 필요합니다. 이렇게 연습하면 "shadow speak" 기술이 크게 향상될 것입니다.

이 영상의 문법

화자가 가장 많이 쓰는 문형을 영상 속 실제 표현과 함께 정리했습니다.

문형	영상 속 표현
관계절 who / which + 절 — 사람이나 사물에 대한 추가 정보	context, which is · animals, which was · works, which is
수동태 be + 과거분사 — 누가 하는지보다 무슨 일이 일어나는지에 초점	was inspired · was seen · was released
현재완료 have/has + 과거분사 — 과거의 일이 지금도 관련이 있을 때	hasn't been · I've seen · has been

쉐도잉이란? 영어 실력을 빠르게 키우는 과학적 방법

쉐도잉(Shadowing)은 원래 전문 통역사 훈련을 위해 개발된 언어 학습 기법으로, 다언어 학자인 Dr. Alexander Arguelles에 의해 대중화된 방법입니다. 핵심 원리는 간단하지만 매우 강력합니다: 원어민의 영어를 들으면서 1~2초의 짧은 지연으로 즉시 소리 내어 따라 말하는 것——마치 '그림자(shadow)'처럼 화자를 따라가는 것입니다. 문법 공부나 수동적인 청취와 달리, 쉐도잉은 뇌와 입 근육이 동시에 실시간으로 영어를 처리하고 재현하도록 훈련합니다. 연구에 따르면 이 방법은 발음 정확도, 억양, 리듬, 연음, 청취력, 말하기 유창성을 크게 향상시킵니다. IELTS 스피킹 준비와 자연스러운 영어 소통을 원하는 분들에게 특히 효과적입니다.

섀도잉 방법: 단계별 전체 가이드 읽기 →

쉐도잉 연습: Meta's new model wants "deep access" to your personal life... - 영상으로 영어 말하기 배우기

연습 모드 선택

앱 다운로드

Did Google just dethrone OpenAI and Anthropic?

Mind-bending new programming language for GPUs just dropped...

PostgreSQL in 100 Seconds

A $6.3 billion open-weight model just got embarrassed by the French...

Every app on Earth is now open source... whether it likes it or not

OpenAI’s lead engineer was secretly a spy...

Intel chips can’t possibly be this bad… 100% crash rate?

NestJS in 100 Seconds

Godot in 100 Seconds

Grok 4 pushes humanity closer to AGI… but there’s a problem

Array Map in 100 Seconds

Docker in 100 Seconds

The most interesting "hack" in history...

OpenAI’s new browser feels familiar…

10 weird OSS projects you need right now...

Firefox and Tor hit with 9.8 critical level exploit

5 open source tools that replaced my $320/mo AI stack...

C++ in 100 Seconds

DHH has gone completely off the rails...

Rust in 100 Seconds

Meta’s Llama 4 is mindblowing… but did it cheat?

Meta is pivoting again... everything you missed from Connect 2026

Transformers, the tech behind LLMs | Deep Learning Chapter 5

Why everyone is quitting social media

WWDC26: Run local agentic AI on the Mac using MLX | Apple

Mark Zuckerberg: “Alex, are you sure this is going to work?”

This 25-Year-Old Built a 2-Year Moat Nobody Can Bet Against | Corgi, Nico Laqua & Emily Yuan

World Models explained in 10min..

How I'd Learn AI From Scratch in 2026 (skip the useless 80%)

Is Amazon Too Big?

The perfect technology for creeps?

Codex Is Winning… So Why Am I Still Paying for Claude?

이 비디오로 배울 실용적인 영어 말하기 기술

듣고 주목해야 할 발음 현상

원어민처럼 말하는 비결: 리듬과 스트레스

이 영상의 문법

쉐도잉이란? 영어 실력을 빠르게 키우는 과학적 방법