Экстракт Знаний
The Most INSANE AI News This Week! 🤯
12:40

The Most INSANE AI News This Week! 🤯

Julian Goldie SEO 12.04.2025 5 245 просмотров 155 лайков обн. 18.02.2026
Смотреть на YouTube
Поделиться Telegram VK Бот
Транскрипт Скачать .md
Анализ с AI
Описание видео
🚀 Get a FREE SEO strategy Session + Discount Now: https://go.juliangoldie.com/strategy-session Want to get more customers, make more profit & save 100s of hours with AI? Join me in the AI Profit Boardroom: https://go.juliangoldie.com/ai-profit-boardroom 🤯 Want more money, traffic and sales from SEO? Join the SEO Elite Circle👇 https://go.juliangoldie.com/register 🤖 Need AI Automation Services? Book an AI Discovery Session Here: https://juliangoldieaiautomation.com/ Click below for FREE access to ✅ 50 FREE AI SEO TOOLS 🔥 200+ AI SEO Prompts! 📈 FREE AI SEO COMMUNITY with 2,000 SEOs ! 🚀 Free AI SEO Course 🏆 Plus TODAY's Video NOTES... https://go.juliangoldie.com/chat-gpt-prompts - Want a Custom GPT built? Order here: https://kwnyzkju.manus.space/ - Join our FREE AI SEO Accelerator here: https://www.facebook.com/groups/aiseomastermind - Need consulting? Book a call with us here: https://link.juliangoldie.com/widget/bookings/seo-gameplanesov12 In this video, I'll cover the most groundbreaking AI developments of the week, including Meta's Llama 4 with its massive 10 million token context window, Microsoft's new memory features for Copilot, Google's Firebase Studio for app creation, the mysterious Quasar Alpha model, OpenAI's strategy shift, Claude Max pricing, and an incredible new way to connect Claude to any tool using MCP agents. 0:00 - Introduction 0:32 - Llama 4 Explosion: Meta's new AI models 0:44 - Llama 4 Scout with 10M token context window 0:58 - Llama 4 Maverick and upcoming Behemoth 1:19 - The Llama drama: Open source controversy 1:42 - Leaked claims about Meta's benchmark manipulation 2:06 - LM Arena rankings controversy 2:33 - Microsoft's Memory Revolution 2:47 - Copilot's new memory features 3:08 - AI-generated Quake game demo 3:24 - John Carmack's defense of AI in gaming 3:37 - Google's AI updates this week 3:53 - Agent-to-agent protocol (A2A) 4:06 - Google Workspace AI enhancements 4:24 - Firebase Studio: Google's no-code app builder 4:43 - How Firebase Studio works 5:03 - Firebase Studio's key features 5:27 - Firebase Studio example: Building an expense calculator 5:49 - OpenAI's changed roadmap 6:02 - GPT-4o and GPT-4o Mini announcements 6:16 - ChatGPT's memory upgrade 6:37 - Quasar Alpha: The mysterious AI model 6:56 - How to access Quasar Alpha through Open Router 7:20 - Building apps with Quasar Alpha 7:37 - Creating AI agents with Quasar Alpha 7:57 - Claude Max: Anthropic's premium plans 8:16 - Claude 4 expected within 6 months 8:25 - YouTube's new AI music tool for creators 8:37 - DaVinci Resolve 20's AI features 8:55 - Video generation updates (Runway & Amazon) 9:11 - Together AI's DeepCoder 14B open-source model 9:23 - New APIs (Gemini, V2, Grok 3) 9:48 - Real-world AI applications (Amazon Zoox, Samsung Ballie) 10:02 - Claude's new MCP agents integration with n8n 10:15 - How the n8n MCP integration works 10:38 - Step-by-step setup process 11:08 - Why this integration is revolutionary 11:36 - The big picture: AI's accelerating progress 12:05 - Call to action and final thoughts

Оглавление (39 сегментов)

  1. 0:00 Introduction 106 сл.
  2. 0:32 Llama 4 Explosion: Meta's new AI models 27 сл.
  3. 0:44 Llama 4 Scout with 10M token context window 42 сл.
  4. 0:58 Llama 4 Maverick and upcoming Behemoth 63 сл.
  5. 1:19 The Llama drama: Open source controversy 68 сл.
  6. 1:42 Leaked claims about Meta's benchmark manipulation 65 сл.
  7. 2:06 LM Arena rankings controversy 86 сл.
  8. 2:33 Microsoft's Memory Revolution 36 сл.
  9. 2:47 Copilot's new memory features 73 сл.
  10. 3:08 AI-generated Quake game demo 54 сл.
  11. 3:24 John Carmack's defense of AI in gaming 35 сл.
  12. 3:37 Google's AI updates this week 47 сл.
  13. 3:53 Agent-to-agent protocol (A2A) 39 сл.
  14. 4:06 Google Workspace AI enhancements 53 сл.
  15. 4:24 Firebase Studio: Google's no-code app builder 63 сл.
  16. 4:43 How Firebase Studio works 58 сл.
  17. 5:03 Firebase Studio's key features 74 сл.
  18. 5:27 Firebase Studio example: Building an expense calculator 71 сл.
  19. 5:49 OpenAI's changed roadmap 38 сл.
  20. 6:02 GPT-4o and GPT-4o Mini announcements 40 сл.
  21. 6:16 ChatGPT's memory upgrade 61 сл.
  22. 6:37 Quasar Alpha: The mysterious AI model 60 сл.
  23. 6:56 How to access Quasar Alpha through Open Router 70 сл.
  24. 7:20 Building apps with Quasar Alpha 59 сл.
  25. 7:37 Creating AI agents with Quasar Alpha 59 сл.
  26. 7:57 Claude Max: Anthropic's premium plans 56 сл.
  27. 8:16 Claude 4 expected within 6 months 33 сл.
  28. 8:25 YouTube's new AI music tool for creators 24 сл.
  29. 8:37 DaVinci Resolve 20's AI features 54 сл.
  30. 8:55 Video generation updates (Runway & Amazon) 53 сл.
  31. 9:11 Together AI's DeepCoder 14B open-source model 27 сл.
  32. 9:23 New APIs (Gemini, V2, Grok 3) 80 сл.
  33. 9:48 Real-world AI applications (Amazon Zoox, Samsung Ballie) 39 сл.
  34. 10:02 Claude's new MCP agents integration with n8n 41 сл.
  35. 10:15 How the n8n MCP integration works 67 сл.
  36. 10:38 Step-by-step setup process 85 сл.
  37. 11:08 Why this integration is revolutionary 85 сл.
  38. 11:36 The big picture: AI's accelerating progress 74 сл.
  39. 12:05 Call to action and final thoughts 106 сл.
0:00

Introduction

This week in AI is mindblowing. I'm about to show you a 10 million token model that's free, a new way to make AI remember everything about you, and video tech that finally gets human bodies right. Plus, the drama behind Llama 4 that nobody's talking about. It's the craziest week in AI yet, and you need to know all of it. Hey, if we haven't met already, I'm the digital avatar of Julian Goldie, CEO of Goldie Agency. while he's helping clients get more leads and customers. I'm here to help you get the latest AI updates. Let's start with the Llama 4 explosion. The
0:32

Llama 4 Explosion: Meta's new AI models

biggest story this week, huge news. Meta just dropped Llama 4, and it's a gamecher. Not just one model, but three different versions. First, Llama 4 Scout
0:44

Llama 4 Scout with 10M token context window

with a crazy 10 million token context window. That's insane. We're talking about 7. 5 million words or about 9 novels in one prompt. Google's Gemini 2. 5 only has 2 million tokens. So, this is five times bigger. Then there's Llama
0:58

Llama 4 Maverick and upcoming Behemoth

for Maverick. Bigger parameters, but smaller 1 million token context. Still huge though. You could fit the entire Harry Potter series in there. And the beast that's coming soon, Llama for Behemoth with two trillion parameters. That would make it the biggest model we know about. And get this, they're calling all of these open source. But here's where things get juicy. There's
1:19

The Llama drama: Open source controversy

some serious llama drama happening with Llama 4. First, Meta's definition of open-source isn't exactly what most people think. If you have over 700 million active users on your Llama based tool, you need Meta's permission. And if you make a new model based on Llama, you have to call it Llama something, not truly open source. But it gets wilder. Someone claiming to be from Meta's AI
1:42

Leaked claims about Meta's benchmark manipulation

team leaked that they weren't being honest about everything. They said that after training, Llama 4 couldn't reach state-of-the-art levels. So, MetaMix benchmark test data into the training to make it look better in benchmarks. Meta's team denied this completely, saying they would never do that. They blame the performance issues people are seeing on implementation problems. The plot thickens. When Llama 4 first showed
2:06

LM Arena rankings controversy

up on Elmarina, the leaderboard where users blind test models, it ranked number two behind Gemini 2. 5 Pro. But check this out. Now it's dropped all the way to number 32. What happened? LM A Marina called Meta out. They said Meta used a different version of Llama 4 on the leaderboard than what they released to everyone else. They optimized a special version just for the leaderboard. That's like bringing a souped-up car to the race, then selling people the regular version. Next up
2:33

Microsoft's Memory Revolution

Microsoft's Memory Revolution. Microsoft just celebrated their 50th anniversary with Bill Gates, Steve Balmer, and Satia Nadella all on stage together. Talk about a power trio. They announced a huge update to Microsoft Copilot. It now
2:47

Copilot's new memory features

has memory. This means Copilot will remember your past conversations to learn what you like, what you don't like, and details about your life, your dog's name, your work projects, even your workout motivation. This is a big deal because it makes AI feel more personal and less like talking to a blank slate every time. Think about not having to repeat the same information over and over. They also showed off an
3:08

AI-generated Quake game demo

AI generated version of Quake built with their Muse AI model. Every single frame of the game is created by AI as you play. It's not perfect. Characters disappear and there are bugs. But the fact that it's generating a playable game on the fly is mind-blowing. But not everyone's happy. Game developers called
3:24

John Carmack's defense of AI in gaming

it disgusting. But interestingly, John Carmarmac, the original creator of Quake, defended it. He said AI tools will help the best developers reach greater heights while giving smaller teams more power. Google also made some
3:37

Google's AI updates this week

big moves this week. They improved their AI search mode to handle longer questions, comparisons, and how-tos. Plus, it can now see and search with images. Take a picture of books on your shelf and Google will find similar ones. At Google Cloud Next, they announced a
3:53

Agent-to-agent protocol (A2A)

new TPU tensor processing unit coming later this year and something called A2A, an agent to agent protocol. This lets AI agents talk to each other and work together without human input. Google Workspace got some AI love, too.
4:06

Google Workspace AI enhancements

There's a new audio feature in Google Docs. Help me refine for better writing, AI enhancements in Sheets, and Gemini features in Google Meet for meeting summaries. For developers, they launched Chirp 3 for audio generation, Imagine 3 for text to image, and Leah for text to music. Google's not playing around. And
4:24

Firebase Studio: Google's no-code app builder

Google just released Firebase Studio, and it's a gamecher for anyone who wants to build apps. Think of it like Google's answer to Bolt and Lovable, a completely free full stack AI workspace. This is crazy because you can build entire apps with just one prompt. I've already created several tools with it and the process is super simple. All you need is
4:43

How Firebase Studio works

a free Gmail account to get started. Just go to studio. firebase google. com, type in what you want to build and Firebase Studio does all the coding for you. I built an SEO tool that analyzes keyword density in seconds and a flashcard app for learning SEO in minutes. Here's what makes Firebase Studio so powerful. It
5:03

Firebase Studio's key features

uses Gemini 2. 5 Flash to generate your apps. You can choose custom colors, layouts, and animations. It publishes your app to a free subdomain instantly. You can share your creations with a simple link or QR code. Everything gets saved automatically in your workspace. The best part, when you run into any issues, just click fix error and it auto troubleshoots for you. No coding skills needed at all. I created an expense
5:27

Firebase Studio example: Building an expense calculator

calculator app while recording this video. just typed the prompt and it built everything, including a nice UI with my brand colors. It would have cost thousands to hire a developer for this before. Firebase Studio even lets you switch to code view if you want to make custom changes, but you don't need to touch a single line of code if you don't want to. And OpenAI has changed course
5:49

OpenAI's changed roadmap

completely. Remember when Sam Alman said they'd skip GPT4 Red and jump straight to GPT5? Well, now he's backtracking. Open AAI will release GPT40 and GPT40 Mini after all. Reading between the lines, this probably means GPT5 is
6:02

GPT-4o and GPT-4o Mini announcements

taking longer than planned, and they don't want to look like they're falling behind while waiting for it. But Altman promises GPT5 will be much better than what everybody's expecting. Chat GPT also got a major memory upgrade. It can
6:16

ChatGPT's memory upgrade

now reference all your past chats. When I logged in, I saw a box saying, "Introducing new improved memory. " The AI wrote me a personal description based on everything we've ever talked about. This is next level personalization. But be careful. Your random chats from months ago could now influence how the AI responds to you. And have you heard
6:37

Quasar Alpha: The mysterious AI model

about Quaza Alpha? It's blowing up right now. And there's a high chance it's actually Open AI's new model. Even Chat GPT says it might be a prototype for GPT5 or a refined GPT4. 5. This is huge because OpenAI has never released a free model like this before. Quaazar Alpha gives you a million token context window API for
6:56

How to access Quasar Alpha through Open Router

free. You can build and automate almost anything with it. The craziest part, it's trending on Open Router right now. Just go to Open Router, select Quazar Alpha as your model, and you're good to go. It's completely free to use and insanely fast. Want to build apps with it? Grab an API key from Open Router settings, plug it into Visual Studio Code, and start coding. You can connect
7:20

Building apps with Quasar Alpha

it to MCP servers like Perplexity to get the latest knowledge beyond its October 2023 cutoff date. You can also use it with Bolt Local to build apps, tools, games, anything you want. I built a full to-do list app in seconds and a quiz game with custom brand colors in under a minute. What about AI agents? Connect
7:37

Creating AI agents with Quasar Alpha

Quaza Alpha to tools like N and create your own AI agents in literally 30 seconds. I made a social media content creator that automatically generates tweets based on the latest news. The fact that this might be OpenAI's model but completely free is mind-blowing. This could be your chance to test GPT5 level performance without paying a dime.
7:57

Claude Max: Anthropic's premium plans

This week, Claude Max arrived. Anthropic rolled out a max plan for Claude, their answer to chat GPT plus's higherend versions. For $100 per month, you get five times more usage than the pro plan. Want even more? For $200 per month, the maximum flexibility plan gives you 20 times more. We should also expect Claude
8:16

Claude 4 expected within 6 months

for within the next 6 months. Jared Kaplan, chief scientist at Anthropic, confirmed it's coming this year. The AI race keeps heating up. YouTube launched a free AI music tool for creators. Need
8:25

YouTube's new AI music tool for creators

royalty-free background music for your videos? No more paying for subscription services? YouTube's got you covered with AI generated tunes? Da Vinci Resolve 20
8:37

DaVinci Resolve 20's AI features

just dropped with tons of AI features. One of the coolest lets you upload a script and your footage and it automatically lines up shots based on your script. It's got better magic mask features and you can train it on your voice for AI voiceovers. Video editing is getting crazy fast. Runway introduced
8:55

Video generation updates (Runway & Amazon)

Gen 4 Turbo, a super fast video generator that creates 10-second videos in just 30 seconds. Amazon's got their own video generator called Nova Real that can make AI videos up to 2 minutes long. The quality is impressive, on par with the best models out there. For coders, Together AI announced Deepcoder
9:11

Together AI's DeepCoder 14B open-source model

14B, an open- source coding model at the 01 or 03 mini level. They're releasing everything. Data set, code, training recipe, the whole package. New APIs are
9:23

New APIs (Gemini, V2, Grok 3)

everywhere. Gemini 2. 5 Flash and Pro, live API, and V2 are all in the Gemini API. Now, want to build a platform where people can generate AI videos using V2? Now, you can. Grocery finally has an API. This is huge because now it can be integrated into tools like Windsurf and Cursor. GitHub Copilot got an agent mode and MCP support for easier API connections. In the real world, Amazon's Zox is rolling out robo taxis in Los
9:48

Real-world AI applications (Amazon Zoox, Samsung Ballie)

Angeles. Samsung is finally releasing Bali, that little ball robot that projects stuff on the floor. And Kawasaki showed off Coro, a robot dog you can ride like a motorcycle. And you can now connect Claude to anything with
10:02

Claude's new MCP agents integration with n8n

new MCP agents. N just dropped a massive update for MCP agents that changes everything. You can now connect Claude directly to Gmail tools and your custom workflows without writing a single line of code. Before this update, you had to
10:15

How the n8n MCP integration works

self-host everything and deal with a ton of tech headaches. Now, it's super simple. Just add the new MCP server trigger in N8N and you're halfway there. Here's what makes this so mindblowing. You can make Claude send emails directly from your Gmail account. I literally sent an email using Clawude Desktop by connecting it to my N8N MCP server. It took seconds. The setup is surprisingly
10:38

Step-by-step setup process

easy. Update your N8N workspace to the latest beta 1. 88. Add the new MCP server trigger node to your workflow. Connect whatever tools you want, Gmail, AI agents, etc. Install Claude desktop and enable developer mode. Connect Claude to your NAN MCP server using a simple config. Once that's done, you can just tell Claude, "Send an email to johnacample. com about our meeting tomorrow. " And it happens instantly. No jumping between apps, no copy pasting, just one seamless experience. This is a
11:08

Why this integration is revolutionary

huge deal because you're basically giving Claude superpowers. Instead of Claude being isolated, it can now trigger any workflow you build in N8 end. Think about connecting Claude to the internet, databases, your calendar, or custom AI agents you've built. There are some limitations. It only works with SSE endpoints, not MPX. But this is just the beginning. If you've been wanting to build a real personal AI assistant that can actually do things for you, this is your chance. So, what does this all
11:36

The big picture: AI's accelerating progress

mean? This week shows us AI is advancing at lightning speed in every direction. We're seeing massive context windows with Llama 4's 10 million tokens, personal AI with memory features in Chat GPT and Microsoft Copilot creative tools transforming how we make videos, music, and games. AI agents starting to work together without human input. The gap between what was science fiction and what's reality keeps shrinking every week. Julian reads every comment, so
12:05

Call to action and final thoughts

make sure you leave your thoughts below. What excites you most? Which of these tools are you going to try first? If you want to scale your business, get more customers, and save hundreds with AI automation. Check out my AI profit boardroom link in the comments and description. Need help with SEO? Book a free SEO strategy session. link in the comments and description. And don't miss my free AI SEO course and community with 3,600 AI SEOs and 50 plus free AI SEO tools. Link in the comments and description, too. Thanks for watching and I'll see you in the next mind-blowing AI update.

Ещё от Julian Goldie SEO

NEW ChatGPT Images Update is WILD! 🤯 6:31

NEW ChatGPT Images Update is WILD! 🤯

Julian Goldie SEO | 28.03.2025 | 2 сегм. | 277 262
This NEW AI Operating System is INSANE! 🤯 10:06

This NEW AI Operating System is INSANE! 🤯

Julian Goldie SEO | 09.07.2025 | 3 сегм. | 261 385
DeepSeek-R1 FULL 1 Hour 40 min Course 1:43:28

DeepSeek-R1 FULL 1 Hour 40 min Course

Julian Goldie SEO | 27.01.2025 | 29 сегм. | 206 428
Build anything with DeepSeek-R1 🤯 18:31

Build anything with DeepSeek-R1 🤯

Julian Goldie SEO | 26.01.2025 | 11 сегм. | 200 170
Google Veo 3 is WILD! 16:06

Google Veo 3 is WILD!

Julian Goldie SEO | 24.05.2025 | 8 сегм. | 127 210
NEW 1-Click Google AI Agents are INSANE! 🤯 16:16

NEW 1-Click Google AI Agents are INSANE! 🤯

Julian Goldie SEO | 08.02.2025 | 9 сегм. | 119 586
Ctrl+V

Экстракт Знаний в Telegram

Транскрипты, идеи, методички — всё самое полезное из лучших YouTube-каналов.

Подписаться