# GLM 5 is INSANE!

## Метаданные

- **Канал:** Julian Goldie SEO
- **YouTube:** https://www.youtube.com/watch?v=jiIgeT1K1jw
- **Дата:** 10.02.2026
- **Длительность:** 8:24
- **Просмотры:** 4,040

## Описание

Want to make money and save time with AI? Get AI Coaching, Support & Courses 👉 https://www.skool.com/ai-profit-lab-7462/about

Get a FREE AI Course + 1000 NEW AI Agents + Video Notes  👉 https://www.skool.com/ai-seo-with-julian-goldie-1553/about

Want to know how I make videos like these? Join the AI Profit Boardroom → https://www.skool.com/ai-profit-lab-7462/about

Get a FREE AI SEO Strategy Session: https://go.juliangoldie.com/strategy-session?utm=julian

Sponsorship inquiries: 
https://docs.google.com/document/d/1EgcoLtqJFF9s9MfJ2OtWzUe0UyKu1WeIryMiA_cs7AU/edit?tab=t.0

China Leaks GLM 5: The Trillion Parameter GPT 5 Killer?

China's latest AI leak reveals GLM 5, a trillion-parameter model designed to rival GPT 5 and Claude with advanced agentic reasoning. This video breaks down the Mixture of Experts architecture and how these powerful open-weight models are shifting the global AI race for businesses and developers.

00:00 - Intro: The GLM 5 Leak
00:46 - Understanding GLM 4.5 and ZAI
02:12 - GLM5 Technical Specs and Rumors
03:35 - How Mixture of Experts (MoE) Works
04:57 - China's Rise in the AI Race
05:42 - Business Impact of Open Weights
06:33 - The Future of Open vs Closed AI

## Содержание

### [0:00](https://www.youtube.com/watch?v=jiIgeT1K1jw) Intro: The GLM 5 Leak

New GLM5 is insane. China just leaked GLM5 and it might flip the entire AI game on its head. We're talking trillion parameter scale, insane agentic reasoning, and it could drop before Chinese New Year. This isn't some small update. This could be China's answer to GPT5. And I'm breaking down exactly what this means for you, your business, and the future of AI. Let's go. So GLM5 just leaked, and people are losing their minds. But here's what nobody's talking about. This isn't just another AI model. This could be the biggest move China's made in the AI race. And if even half of these leaks are true, we're looking at something massive. Before we dive in, I need to give you some context because understanding where GLM5 comes from makes this whole thing way more crazy.

### [0:46](https://www.youtube.com/watch?v=jiIgeT1K1jw&t=46s) Understanding GLM 4.5 and ZAI

GLM 4. 5 is the current public model from Jepu AI. They rebranded to ZAI recently. It's open source with 355 billion total parameters. But here's the cool part. It only uses 32 billion active parameters at a time. That's the magic of MOE architecture. Mixture of experts. We'll get into that in a second. GLM 4. 5 already has strong reasoning, coding skills, and agent capabilities. It's competing with models like Claude and Deepseek and it's open weights, which means developers can actually use it, fine-tune it, and build with it. But GLM 4. 5 also has two modes. Thinking mode for deep reasoning and tool calling, and non-thinking mode for fast replies. This hybrid approach makes it super flexible. Now, that's the foundation. That's what came before and it's already really good. But GLM5, that's where things get wild. Hey, if we haven't met already, I'm the digital avatar of Julian Goldie, CEO of SEO agency Goldie Agency. Whilst he's helping clients get more leads and customers, I'm here to help you get the latest AI updates. Julian Goldie reads every comment, so make sure you comment below. Now, before we go any further, I need to be super clear about something. Everything I'm about to tell you about GLM5 is based on leaks and community rumors. This is not officially confirmed by Jepu AI or ZAI. These are leaks floating around Reddit, LinkedIn, and AI forums. So, take it with a grain of salt. But even as leaks, this stuff is crazy. So, what are people saying about

### [2:12](https://www.youtube.com/watch?v=jiIgeT1K1jw&t=132s) GLM5 Technical Specs and Rumors

GLM5? Well, the leaks are everywhere and the claims are massive. First, GLM5 is actively training right now. This isn't just a concept. It's happening. The model is being built as we speak. Second, it's designed to compete directly with GPT5 and Claude on reasoning, understanding, and agentic tasks. Think advanced problem solving, planning, and tool use. Third, the leaks suggest a possible trillion parameter scale design. That's huge with massive agentic capabilities, deeper reasoning, better planning, smarter tool use. And fourth, people are expecting GLM5 to drop right before China's Luna New Year. That's midFebruary 2026. So, we could see this thing in just a few weeks. Now, again, these are leaks, but if they're true, this changes everything. And if you want to learn how to actually use AI tools like GLM5 to scale your business, get more customers, and save hundreds of hours with automation, you need to join the AI profit boardroom. This is where we break down exactly how to implement these cuttingedge AI models into your workflows. We're not just talking about the news. We're showing you step by step how to use tools like GLM5 to automate your content, your customer outreach, your entire business. You'll get access to SOPs, use cases, and a community of people already crushing it with AI. The link is in the description. Let me break

### [3:35](https://www.youtube.com/watch?v=jiIgeT1K1jw&t=215s) How Mixture of Experts (MoE) Works

down why this matters. Because it's not just about big numbers, it's about what those numbers mean for how AI actually works. GLM5 reportedly uses architecture, mixture of experts. Here's how it works in simple terms. Instead of using the entire neural network for every single task, the model only activates certain parts. These parts are called experts. So you get a massive model with huge capacity, but you only use a small chunk of it at a time. This makes it super efficient. You get the power of a trillion parameters without needing insane amounts of compute for every single query. That's the breakthrough and that's why is becoming the standard for these massive models. Deepseek uses it. Kimmy uses it. GLM 4. 5 uses it. And now GLM5 is taking it even further. But there's more. GLM models also use something called efficient sparse attention. This is similar to what DeepS does with long context windows. Basically, it lets the model handle massive amounts of text at once, research papers, entire code bases, full books without breaking a sweat. And here's the other key thing to understand. Total parameters versus active parameters. Total parameters is all the knowledge encoded in the model. Active parameters is what's actually switched on for each query. So GLM 4. 5 has 355 billion total parameters, but only 32 billion active. That ratio is what makes it so efficient, and GLM5 could push that even further. So where

### [4:57](https://www.youtube.com/watch?v=jiIgeT1K1jw&t=297s) China's Rise in the AI Race

does GLM5 fit in the bigger picture? Well, the AI race right now is insane. Deepseek just dropped a model that shocked everyone. Kim is crushing it with long context. GLM 4. 5 is already in the top three for open- source benchmarks. is performing right alongside Claude and Deepseek in real world tests. And now GLM5 could push that even further, maybe even to Frontier Intelligence. That's the level where models start doing things we didn't think were possible. And here's the thing. If GLM5 is real and it's as good as the leaks say, China is making a massive statement. They're not just competing. They're leading and they're doing it with open models. That's huge because open models mean anyone can use them. Developers, startups, businesses, you can run it yourself. fine-tune it, build on it. That's a game changer. Now, here's where

### [5:42](https://www.youtube.com/watch?v=jiIgeT1K1jw&t=342s) Business Impact of Open Weights

this gets really interesting for you. If you're running a business or building anything with AI, this matters a lot. Let me tell you why. First, for developers, if GLM5 has open weights, like GLM 4. 5, you could run a trillion parameter model locally or fine-tune it for your specific apps. That's insane right now. But with open weights, you own it. You control it. You can customize it however you want. Second, for the global AI race, China is pushing hard to rival Western Labs. OpenAI, Anthropic, Google, and they're doing it by releasing powerful open models. This creates competition, and competition drives innovation, which means better tools for everyone. Third, for startups and tools, if GLM5 is open or cheap to use, it could create a massive ecosystem shift. Cheaper AI, more accessible AI, that means more people building with it, more tools, more possibilities. Now, let

### [6:33](https://www.youtube.com/watch?v=jiIgeT1K1jw&t=393s) The Future of Open vs Closed AI

me talk about what this means for the future, because this isn't just about one model. It's about where AI is heading. We're seeing a shift from closed proprietary models to open accessible ones. And that shift is happening fast. A year ago, nobody thought we'd have open models competing with GPT4. But now we do. GLM 4. 5, Deep Seek, Kimmy, they're all right there. And if GLM5 hits the market at trillion parameter scale, it raises the bar again. Suddenly, the baseline for what's possible goes way up and everyone else has to match it. This is good for innovation. is good for competition and it's good for you because more options means better tools, cheaper tools, more accessible tools and that means more opportunities to build, create, and grow your business with AI. So keep your eyes on GLM5 because this could be the start of something big. And if you want to learn how to actually use AI tools like GLM5 to scale your business, get more customers, and save hundreds of hours with automation, you need to join the AI profit boardroom. This is where we break down exactly how to implement these cutting edge AI models into your workflows. We're not just talking about the news. We're showing you step by step how to use tools like GLM5 to automate your content, your customer outreach, your entire business. You'll get access to SOPs, use cases, and a community of people already crushing it with AI. The link is in the description. Don't miss this. And if you want the full process, SOPs, and over 100 AI use cases like this one, join the AI Success Lab, links in the comments and description. You'll get all the video notes from there, plus access to our community of 38,000 members who are crushing it with AI. This is where you'll learn how to actually implement these tools into your business. Not just watch the news, but use it. So, click the link and join us. I'll see you in there. All right. Thanks for watching. Hit the like and subscribe button and I will see you in the next

---
*Источник: https://ekstraktznaniy.ru/video/9681*