Экстракт Знаний

China Launches FREE AI Research Tool + 10 Major Updates This Week

14:18

China Launches FREE AI Research Tool + 10 Major Updates This Week

Vaibhav Sisinty 24.09.2025 43 548 просмотров 1 099 лайков обн. 18.02.2026

Смотреть на YouTube

Поделиться Telegram VK Бот

Транскрипт Скачать .md

Анализ с AI

Описание видео

🔗 Join Our AI Updates WhatsApp Channel: https://whatsapp.com/channel/0029VbBALDP84OmAI59i4b0R The AI world just went nuclear. This week alone: OpenAI achieved a PERFECT score at the world's hardest coding competition - something NO human team could do. Meta shipped $799 glasses that literally read your mind. Google accidentally leaked Gemini 3.0 (and quickly deleted it). And China? They just released AI tools for FREE that Silicon Valley charges thousands for. Every week I track 100+ AI updates. This week broke all records. These 11 updates aren't just iterations - they're revolutions happening simultaneously. In this episode: Meta's neural wristband that controls AR with your thoughts. OpenAI's 12/12 perfect ICPC score that proved AI codes better than humans. Google's leaked Gemini 3.0 that could end GPT's dominance. China's Hunyuan 3D giving away Hollywood-grade tools. Notion's AI agents that work while you sleep. Chrome becoming an AI operating system for 3 billion users. Plus World Labs turning single photos into 3D worlds, Luma's reasoning video model, and the AI that can now use your credit card. ⏰ TIMESTAMPS: 00:00 - Introduction: 11 Game-Changing AI Updates 00:10 - Meta Connect 2025: The End of Smartphones? 00:32 - Ray-Ban Meta 2.0: Double battery, 3K video 00:59 - Conversation Focus: Volume control for humans 01:20 - Live AI Assistant: 2-hour always-on Jarvis 01:44 - Oakley Meta Vanguard: Sports glasses with Strava/Garmin 02:07 - Ray-Ban Display: 5000 nit micro display AR glasses 02:31 - Neural Band: Mind-reading wristband control 02:54 - Real-time translation subtitles in your glasses 03:16 - OpenAI Perfect Score at ICPC World's hardest coding competition 12/12 problems solved (no human team achieved this) GPT-5 + experimental reasoning model 04:30 - Google Gemini 3.0 Ultra Leak GitHub repository leak and deletion Benchmarks and December 2025 predictions 05:36 - Google Agent Payment Protocol (APP2) AI that can buy things with your credit card Agent-to-agent transactions Enterprise launch Q1 2026 07:33 - World Labs: One Photo to 3D World Fei-Fei Li's $230M startup Single image to explorable environment Real-time browser-based navigation 08:40 - GPT-5 Codex: Autonomous Engineering 74.9% on SWE-bench verified Codes while you sleep VS Code and GitHub integration 09:34 - China's Hunyuan 3D 3.0 Free, open-source 3D generation 3.6 billion voxels Tencent's ecosystem play 10:21 - Luma Ray3: First Reasoning Video Model Studio-grade HDR generation Draft mode 5x faster Hollywood-quality AI video 11:07 - Notion 3.0 Agents AI teammates that work 20 minutes autonomously $500M annual revenue milestone Custom agents coming soon 11:58 - Google Chrome AI Upgrade Gemini integration for 3 billion users Multi-step task automation Real-time scam detection 13:52 - Outro & Next Week Preview -------- To Know More, Follow Vaibhav Sisinty On ⤵︎ Instagram @VaibhavSisinty https://www.instagram.com/vaibhavsisinty Twitter @VaibhavSisinty https://twitter.com/VaibhavSisinty Facebook @VaibhavSisinty https://www.facebook.com/vaibhavsisinty/ LinkedIn - Vaibhav Sisinty https://www.linkedin.com/in/vaibhavsisinty --------

Оглавление (19 сегментов)

0:00 Introduction: 11 Game-Changing AI Updates 22 сл.
0:10 Meta Connect 2025: The End of Smartphones? 69 сл.
0:32 Ray-Ban Meta 2.0: Double battery, 3K video 75 сл.
0:59 Conversation Focus: Volume control for humans 64 сл.
1:20 Live AI Assistant: 2-hour always-on Jarvis 69 сл.
1:44 Oakley Meta Vanguard: Sports glasses with Strava/Garmin 64 сл.
2:07 Ray-Ban Display: 5000 nit micro display AR glasses 65 сл.
2:31 Neural Band: Mind-reading wristband control 67 сл.
2:54 Real-time translation subtitles in your glasses 65 сл.
3:16 OpenAI Perfect Score at ICPC 194 сл.
4:30 Google Gemini 3.0 Ultra Leak 149 сл.
5:36 Google Agent Payment Protocol (APP2) 240 сл.
7:33 World Labs: One Photo to 3D World 186 сл.
8:40 GPT-5 Codex: Autonomous Engineering 139 сл.
9:34 China's Hunyuan 3D 3.0 125 сл.
10:21 Luma Ray3: First Reasoning Video Model 122 сл.
11:07 Notion 3.0 Agents 146 сл.
11:58 Google Chrome AI Upgrade 343 сл.
13:52 Outro & Next Week Preview 69 сл.

Introduction: 11 Game-Changing AI Updates

This week in AI has been wild. I tracked over 100 updates, but only 11 are absolute game changers. Let's dive in.

Meta Connect 2025: The End of Smartphones?

Okay, I think Mark Zuckerberg just low-key ended smartphones. Like, not in 10 years, not someday in the metaverse. Right now, this week at Meta Connect 2025. And the craziest part, most people think it was just about some upgraded AR glasses, but no, this was so much bigger. Zuck basically came out and said, "We're building 10 different ways to make your phone useless. " Let me show

Ray-Ban Meta 2.0: Double battery, 3K video

you what just dropped. It's insane. First up, Rayban Meta 2. 0. Looks almost identical to last year's version, right? But inside, double the battery. 3K video recording literally puts your iPhone to shame. But here's where it gets wild. They added something called conversation focus. Imagine this. You're at a loud cafe, music playing, people chatting. You tap your glasses, and suddenly your friend's voice gets louder, and everything else fades out. It's like

Conversation Focus: Volume control for humans

giving someone a volume knob. Basically, AirPods Pro's transparency mode, but on beast mode. Then they casually drop this live AI assistant. And no, it's not, "Hey Siri, what's the weather? " This thing stays active for 2 hours. It's always listening, learning your behavior, and predicting what you need before you even ask. Like I don't know whether to be amazed or mildly

Live AI Assistant: 2-hour always-on Jarvis

terrified, but yeah, it's basically having your own Jarvis whispering in your ear all day. Next, Oakley Meta Vanguard. This one's for the runners, cyclists, people who hate phones bouncing around during workouts. 122° wide-angle camera, waterproof, syncs with Straa and Gin automatically. So now you go for a run and come back with a full documentary of it. Okay, but this was the moment my jaw dropped. Ray-B

Oakley Meta Vanguard: Sports glasses with Strava/Garmin

band display. They literally put a 5,000nit micro display inside normallook glasses. You could wear them to brunch and people wouldn't even know. 42 pixels per degree. That's sharper than the Quest 3. And yeah, videos floating in your vision as you walk. No bulky headset. No weird goggles. Just straight up sci-fi in sunglasses form. Then they went full black mirror. This wristband

Ray-Ban Display: 5000 nit micro display AR glasses

they call the neural band, it reads your muscle signals. So if you pinch your fingers, it clicks. Rotate your wrist, it scrolls. You're not touching anything, but you're controlling everything. Nobody around you even sees it happening. You're just silently navigating the world like a psychic ninja. Travelers, you're going to love this. Someone speaks Spanish, you see English subtitles floating in your lens

Neural Band: Mind-reading wristband control

instantly. No need to pull out Google Translate, no awkward pauses, just seamless conversation anywhere in the world. And then it got straight up cinematic. They dropped Horizon Studio. You scan your room and the space around you literally transforms. This isn't editing, it's realtime reality remixing. All of this starts at $799. Sales of the Ray-B band Meta glasses already tripled this year. This isn't a

Real-time translation subtitles in your glasses

Vision Pro competitor. This is Meta trying to replace your phone altogether. And honestly, after watching that keynote, I kind of think they might pull it off. If this is where AR is in 2025, we're about to see the wildest platform shift since the iPhone. So, OpenAI just did something that no human team could. They got a perfect score at ICPC, 12 out

OpenAI Perfect Score at ICPC

of 12 problems. The best human team, they got 11 out of 12. OpenAI beat them all. ICPC, by the way, is the Olympics of programming. This happened on September 4th in Baku, Azarbaijan. 139 university teams, 5 hours, 12 problems. You either get it perfect or you get nothing. So, here's how OpenAI did it. They used two models. GPT5 solved the first 11 problems. No problem. First try every time. But for the 12th problem, GPT5 hit a wall. So they brought in this experimental reasoning model. And after nine tries, they cracked it. One human team solved it in three tries. Open AAI needed nine, but they got it. Now, here's the crazy part. Problem C, it stumped every single human team. 139 teams, none of them solved it. But Google's Gemini solved it in 30 minutes. And guess what? OpenAI did too. Oh, and get this. Jakob Pachi, OpenAI's chief scientist, tweeted his congratulations. Guess what? He won ICPC gold in 2012. The guy who built the AI that just crushed ICPC. He used to compete in it himself. Kind of wild, right? If AI had its own ranking, it

Google Gemini 3.0 Ultra Leak

would have been first Open AAI 12 of 12, second Google Gemini 10 of 10, 123rd St. Petersburg University 11 of 12. This week, someone spotted Gemini 3. 0 Ultra in Google's GitHub and the AI world went crazy. Then Google daily it saying it was misleading test data. Too late. Screenshots were everywhere and speculation exploded. Then came the benchmarks. Some anonymous post claimed Gemini 3. 0 scored 32. 4% on something called humanity's last exam. The problem, that benchmark doesn't even exist. just made up numbers spreading like wildfire. Here's what we know. Google's release pattern is predictable. Gemini 1. 0 December 2023 Gemini 2. 0 December 2024 Gemini 2. 5 mid 2025. So Gemini 3. 0 probably December 2025. But why leak this now? Open AAI just dominated ICPC with GPT5 and Google needs to catch up. Classic move. Leak something. Test the reactions, then go

Google Agent Payment Protocol (APP2)

silent. Was the leak real? Probably not. Is Gemini 3. 0 coming? Definitely. Expect it in December 2025. And if those fake benchmarks are true, OpenAI should be worried. Google just gave AI the keys to your wallet. Introducing the agent payment protocol, AP2. Now your AI can actually make purchases for you. From booking flights to ordering groceries, it uses your credit card to handle everything seamlessly. Adobe, Salesforce, Intuitit, and Accenture are already testing the system. The real kicker, agentto agent transactions. Your AI can negotiate with Amazon's AI for better prices. Restaurant AIs can bid for your reservation, and insurance AIS can adjust your premium in real time based on your driving data. AP2 will be available for enterprise partners in the first quarter of 2026. And by 2027, Google expects 30% of all online transactions to be AI initiated. By 2030, we might never see a checkout page again. The real question isn't whether AI can handle your money. It's whether you trust an algorithm that never forgets a deal, never impulse buys, and never sleeps. Google just gave AI the keys to your wallet, introducing the agent payment protocol, AP2. Now your AI already testing the system. The real kicker, agent to agent transactions. Restaurant AIs can bid for your reservation and insurance AIS can adjust your premium in real time based on your driving data. By 2030, we might never see a checkout page

World Labs: One Photo to 3D World

again. The real question isn't whether AI can handle your money. It's whether you trust an algorithm that never forgets a deal, never impulse buys, and never sleeps. Wordlabs just turned one photo into a fully explorable 3D world. No, it's not just a depth map. It's an actual environment you can walk around in. Pretty wild, right? This is the brainchild of Feay Lee, the genius behind ImageNet. They've raised $230 million and are already valued at over $1 billion. Here's how it works. You upload a single photo and in real time it creates a 3D scene you can navigate right in your browser. It's not just a flat image. It's like stepping into a new world and it's got real physics. Everything reacts like it should. Unlike traditional photoggramometry which needs hundreds of photos or nearfs that need multiple angles, World Labs only needs one. Now, it's not perfect yet. You can't walk too far and there are some minor rendering glitches, but it's an early preview. The first full product is coming. Co-founder Justin Johnson put it perfectly. Virtual worlds cost hundreds

GPT-5 Codex: Autonomous Engineering

of millions. This changes that game developers, film studios, architects, they're all waiting for to get their hands on this. If you want to try the demo, head over to worldlabs. ai. OpenAI just dropped GPT5 CEX, not a coding assistant, an autonomous engineer. This isn't just another upgrade. GPT5 Codeex is built for realworld software engineering. It scored 74. 9% on the SWE bench verified benchmark, outperforming previous models. In a complex refactoring task, it achieved 51. 3% accuracy, a significant improvement over GPT5's 33. 9%. The standout feature, the handoff capability. Start coding in VS Code, then shut down. Codex picks up where you left off, runs tests, commits to GitHub, and continues working while you sleep. You can access it through the Codeex CLI extension, GitHub integration, and the Chat GPT mobile app. GPT5. Codex isn't

China's Hunyuan 3D 3.0

here to replace developers, but it changes the workflow. Code while you work. Let a I finish while you sleep. The race is on. China just released Hanuan 3D 3. 0 free open-source 3D generation with 3. 6 billion voxels and 1,536 cubed resolution. It's faster and more detailed than Western tools, and it's already being downloaded millions of times. Unlike Western 3D tools that charge per model, Hanuan 3D is free for personal use. Tencent launched Hanuan 3D Studio slashing workflows from days to minutes and it runs on consumer hardware. There are some limitations. Commercial use needs permission and models can be too heavy for AAA games. But Tencent's building a massive ecosystem with 90% of Chinese gaming companies using Tencent Cloud. Tencent's

Luma Ray3: First Reasoning Video Model

giving away 3D tech for free while restricting commercial rights creating long-term dependency. Lumemer 3 just launched something huge. The world's first reasoning video model that generates studiograde HDR footage. That means AI generated videos so realistic they're cinema quality. No more generic low res clips. This is the future of video creation. With Ray 3, the AI doesn't just create footage. It thinks it critiques, refineses, and perfects itself, delivering realistic and consistent results every time. And with draft mode, creators can generate rough cuts five times faster, explore dozens of concepts, and then upgrade the best shots to 4K HDR. It's available now for free and Dream Machine, bringing Hollywood level video creation to everyone from indie creators to major

Notion 3.0 Agents

studios. Just look at some of the output people are already creating. The detail and consistency are next level. Notion just launched agents AI teammates that do everything a human can in Notion. Your agent works autonomously for 20 minutes at a time, creating documents, building databases, searching Slack and Google Drive, and executing multi-step workflows. It can even handle an entire project, creating plans, assigning tasks, and drafting documents. Complete automation. The real game changer is memory. Give your agent your preferences, and it remembers over time, becoming your assistant, not just a bot. Ben Levik from RAM says, "We can now instantly spin up systems that used to take hours. " Nine out of 10 ramp employees now use notion AI monthly. The killer feature custom agents. Each aitmate will handle specific tasks. Customer feedback, weekly updates, IT requests. They work on autopilot even

Google Chrome AI Upgrade

while you sleep. Notion just crossed $500 million in annual revenue with over 50% of customers paying for AI features. Enterprise adoption doubled and 99% of organizations are planning to deploy agents. As Akshai Kotari, co-founder of Notion said, with 3. 0 zero. You get agents that do the work for you. This isn't just an update, it's workforce replacement, available now for all Notion users. The age of AI co-workers has begun. Google just dropped the biggest Chrome upgrade ever, and most people aren't talking about it. Sundar Pichai called it our next Chrome era. AI features powered by Gemini that completely change what a browser can do. Chrome is now your personal AI agent. Why does this matter? The browser wars just became the AI wars. Google rolled out Gemini to Chrome's 3 billion users for free. No one else has this kind of reach. Starting today, Gemini is live for Mac and Windows users in the US. It can see your entire page, clarify information, and work across tabs. No extensions needed. But it gets crazier. Soon, Gemini will handle multi-step tasks autonomously. Book a haircut, it'll do it for you. Buy groceries on Instacart? Gemini does that, too, while you work in another tab. Gemini also remembers past pages, integrates with calendar, YouTube, and maps. Ask it for a YouTube timestamp. Type complex questions and get detailed answers. Plus, Gemini detects scams in real time and blocks fake virus alerts, saving you from online fraud. It's even smarter with permissions, learning what you like and blocking intrusive requests. And when it comes to passwords, Gemini can autogenerate and update them on supported sites with one click. This is rolling out to Google Workspace soon, and it could change how businesses reach consumers. On mobile, Gemini will help you across apps, summarize pages, and even act as a personal assistant. As Sundar Pichai said, Chrome isn't just displaying websites anymore. It's understanding, acting on, and protecting you from them. The browser is now the operating system for AI. Every website

Outro & Next Week Preview

now has an AI layer, automating tasks, blocking scams, and making browsing smarter. This isn't just Chrome's biggest update. It's the moment browsers became AI agents. Every week the impossible becomes inevitable and it's only speeding up. Next week's going to be even crazier. Hit subscribe, ring the bell, and join our exclusive WhatsApp community for updates you won't find on YouTube. Link in description. See you next week.

Ещё от Vaibhav Sisinty

Ankur@warikoo Untold Podcast: From Employee to Founder to Creator

Vaibhav Sisinty | 07.10.2023 | 34 сегм. | 875 628

Google's SECRET 7 AI Tools Just DESTROYED ChatGPT (100% FREE Stack)

Vaibhav Sisinty | 12.11.2025 | 20 сегм. | 361 978

Google's FREE Tool Just DESTROYED 10 AI Subscriptions (NotebookLM Mastery)

Vaibhav Sisinty | 23.12.2025 | 13 сегм. | 261 010

⁠Podcast with @Sahil_Bloom on Content Creation, Side Hustles & Fitness | GSTH 04

Vaibhav Sisinty | 20.04.2024 | 20 сегм. | 218 777

This Viral AI Bot Can do your Work For FREE | Here's How to Set It Up

Vaibhav Sisinty | 29.01.2026 | 15 сегм. | 185 947

Master ChatGPT Agent Builder Before It's Too Late: Dev Day Breakdown + Full Tutorial

Vaibhav Sisinty | 08.10.2025 | 21 сегм. | 163 843

Ctrl+V

Экстракт Знаний в Telegram

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

Подписаться