OpenAI Just Gave India ₹4,788 worth of ChatGPT Pro For FREE | Top 10 AI Updates of the week

This week, 150 plus AI updates dropped. I sorted through them all to find the top 10 that actually matter. Every week, I do this vade through the noise so you don't have to. And here's what I've learned. Most AI news is just hype. But this week, four things genuinely change the game. Open AAI made their 399 rupees per month plan completely free in India for an entire year. Google dropped coordinated updates across quantum computing and robotics all at once. Claude launched desktop apps that work while you sleep. And a 16-year-old got arrested because AI thought his Doritos bag was a gun. Let's break down all 10, starting with the companies making the biggest moves. Quick thing though, I've got a weekly newsletter where I break down all the latest AI updates, the tools worth trying, and the ones you should skip. It's completely free, and honestly, it's the best way to stay ahead with all this AI stuff. Links in the description below. OpenAI just did

something pretty wild. Charge GBT Go. You know their 399 rupees a month plan. It's completely free in India for a full year starting November 4th. And I'm not talking about some sketchy trial that charges you later. And you're getting the good stuff. GPT5. 10 times more messages than the free version. 10 times more image generation. 10 times more file uploads. The free tier barely lets you do anything. This is way better. But here's what's actually happening. India is their second biggest market after the US. — India is an incredibly important market for AI in general for OpenAI in particular. It's our second biggest market. Tripled users here in the last year and everyone's fighting for Indian users right now. Perplexity teamed up with Airel giving 360 million people free perplexity pro. Google made Gemini Pro free for students. So OpenAI looked at all this and just said you know what free for literally everyone. While OpenAI is doing this

Elon Musk just declared war on Wikipedia. He launched a brand new AI powered encyclopedia called Groipedia and he's already bragging it's better than Wikipedia and also that the upcoming version 1. 0 will be 10 times smarter. According to him, the goal of Groipedia is to deliver the whole truth and nothing but the truth. Groedia launched with 885,000 articles. That's tiny compared to Wikipedia's 7 million. Though every article is written and fact checked by Ex's AI, can a robot encyclopedia outsmart millions of humans? Grokipedia is already getting attention for challenging biased information online. Supporters say it finally goes deeper on sensitive topics without filtering opinions, but critics challenge this view. They say some articles are clearly biased. The search sends you to weird pages and they're sneaking in controversial statements. There are also concerns that many pages look heavily inspired by Wikipedia just without the proper citations. Experts warn that even if the AI is trying to remove bias, it might just be swapping one bias for another. So the real question is, can an AI encyclopedia truly stay neutral? Time will tell, but for now, Google dropped

some major news. Google quantum chip Willow just solved a quantum computing benchmark 13,000 times faster than regular computers. Here's why that matters. Some problems are so hard that normal computers would take forever to solve them. Think drug discovery. Finding new medicines requires running millions of calculations. Stuff that would take regular computers years. Willow does it in minutes. They're calling the algorithm quantum echoes. And the results are verifiable. We're watching the moment quantum computing goes from theory to reality.

Next up, while Elon and Open AAI are literally fighting on Twitter like it's high school drama, Google just quietly dropped four massive announcements. And like nobody's even talking about them. Let me tell you what they did. First, they helped cure cancer. Google's Gemma 2-based AI model just discovered a new drug combination that makes tumors 50% more visible to your immune system in lab tests. Scientists at Yale and Google DeepMind tested it on actual human cells and it worked. Next, robots. Google DeepMind just launched Gemini Robotics 1. 5, and they're basically bringing AI into the real world, like robots that can actually help you with stuff in your house. But here's where it gets wild. Gemini Robotics 1. 5 has something called as thinking VA. Before taking any action, the robot generates an internal monologue in natural language. It's literally talking itself through the task. This makes it way better at multi-step tasks like folding origami, packing lunch boxes, or preparing a salad. Visual instruction, action, and task generalization. Gemini Robotics 1. 5 dominates everything. Next up, well, Google is building its second

biggest data center outside the United States in India. And before you go, another office just for the sake of it. No, Google dropped $15 billion on India and it's about to completely flip India's economy upside down. So what's Google building? They're partnering up with Adani group and Bharti Airtel to build a 1 gawatt AI facility in Visag. I used to take this train through the south of India and there's this town called Visag. It's a beautiful coastal town growing up and we announced our AI investment outside the US. A $15 billion 1 gawatt plus data center 80% powered by clean energy with subc cables going in. It's an entire campus powered by renewable energy. The best part, it's set to create thousands of jobs, opening up massive opportunities for us in India. Google has decided to make

learning fun. Okay, imagine this. Google just turned your boring study notes into an anime. No, I'm not joking. Notebook LM just dropped six brand new visual styles powered by Nano Banana. And trust me, students are flipping out. Notebook LM already had the video overviews that turned documents into narrated videos, but let's be honest, they were pretty basic. But now you can create watercolor, anime, whiteboard, retroprint, heritage, and papercraft. All you do is upload your lecture notes, hit video overview, and pick your style. The AI isn't just slapping random images on the slides either. It reads your content and creates contextual illustrations that actually match what you're learning. I mean, wow. But wait, there's more. Google's also introduced a brief format. Need a quick rundown? Choose brief and get bite-sized insights. But if you want the deep dive, you can go with explainer and it connects all the dots across your sources. You decide how much you want to absorb. Genius, right? And did I mention this is going to be available in all supported languages? Pretty global if you ask me. But Google also has its

eye on the real issues. Next up, Google just dropped Vibe coding in AI Studio powered by Gemini models. It's basically drag and drop app creation. Think of it as if Figma and a developer had a kid. There is something called annotation mode. You click on any part of your app and tell Google what to change. And if you get stuck while brainstorming, simply hit the I'm feeling lucky button and Google gives you random app ideas to get you started. There's also an app gallery now full of project ideas you can preview, learn from, and remix into your own creations. Just click build and start talking. You can literally build a full app by pointing at things and

talking. That's insane. While Google was dropping all that, Anthropic quietly made that might be even smarter. Claude is now on a desktop available for both Windows and Mac. This means with one keyboard shortcut, Claude pops up over whatever you're working on. No context switching, no losing your flow. Drag and drop any doc, spreadsheet, or slide into Claude for instant summaries, edits, and analysis. And here's the game changer, desktop extensions. These let you connect Claude to your local tools and files through something called the model context protocol. Claude can now read your project files, run your code, and access your databases with your permission. Of course, the setup is dead simple. Download from claude. ai, AI. Install it, sign in with your account, and you're done. If you think this will make your work easier, let me tell you about Claude

Skills. Here's why Claude skills could change how you work. They're folders containing instructions, scripts, and resources that Claude automatically loads when you need them. You can now build custom skills to teach Claude specialized tasks. Claude scans available skills to find relevant matches. When one matches, it loads only the minimal information needed, keeping Claude fast. Multiple skills stacked together automatically. This is an AI that finally remembers how you work. Next up, they moved Claude

code to your browser. Claude can now write code for you while you're not even at your computer. Yeah, this is wild. Now, you just tell Claude what you need fixed in plain English, and it does the work in the cloud. Multiple projects at once while you're doing literally anything else. The insane part, you can do this from your phone now. Imagine you're at a coffee shop. Just tell Claude, "Fix that bug and it's done by the time you're back at your desk. " But here's the catch. It's only available for iOS for now. If you have Claude Pro or Max, just go to claude. ai/code, connect your code, and start telling it what to do. They have also

dropped Claude Haiku 4. 5 and it's insanely cheap. Remember their old best model? This new one works just as good, but costs three times less and runs twice as fast. But here's the crazy part. Haiku 4. 5 is the first Haiku model to get extended thinking. That means it can now think through hard problems like the expensive bigger models, but for much cheaper and faster. And lastly, Anthropic is

officially coming to India. They're establishing a major presence in Bengaluru in early 2026, following in the footsteps of companies like OpenAI. The timing is absolutely calculated. Just a month ago in September, Anthropic released its 2025 economic index report and found out that India is the second largest user base for Claude globally right behind the US. CEO Dario Amade visited India this month and met with government officials and partners. Aropic will also be working towards education, healthcare and agriculture, real world AI applications that could actually change lives. We're actually breaking down that entire economic index report. Here is a short teaser of the video. Over half of our usage is concentrated in coding. We're automating instead of augmenting. We're optimizing one narrow domain instead of transforming our entire economy. Meanwhile, 77% of global businesses are already deploying AI to automate complete tasks. And they're not starting with the cheap, easy stuff. They're going after the high value complex work. The gap isn't just wide. It's widening every single day. So, here's the question I want you to ask yourself. Are you replaceable or indispensable? Are you using AI to get answers faster or to get fundamentally smarter? Are you copying outputs or learning the logic behind them? Are you automating yourself out of a job or augmenting yourself into an irreplaceable position? Because that 0. 27 score, it doesn't have to define us if we change how we think about AI. That video is dropping soon. And trust me, the data is wild. Subscribe now so you catch it. OpenAI launched Atlas

an AI that can actually browse the web and take actions for you, book flights, research topics, click buttons, fill forms, and a lot more. We actually did a full breakdown on both of these. You can check the full video after this link is in description. They bought a company called Software Applications Incorporated. This company made an app called Sky for Mac computers. Here's what Sky does. It's like having an AI assistant that sits on top of your screen and actually does stuff for you. But here's the wild part. Sky can actually see what's on your screen and click buttons for you. And the people who made Sky, they're the same team that built workflow, which Apple bought years ago that became shortcuts on your iPhone. So, OpenAI just hired the people who know how to make AI actually do things. They're putting Sky into Chat GPT. So, soon ChatGpt won't just answer your questions, it'll actually help you get work done. But OpenAI is playing a even

bigger game. All right, for my next AI update, I need you to do something for me. Just stop and actually imagine this. You're 16. You just finished a long football practice. You're tired. You're hungry. You grab a bag of Doritos while you wait for your ride. You finish the bag, crumple it up, and you're just waiting. Suddenly, eight police cars swarm you. Officers are jumping out, guns drawn, shouting at you to get on the ground. They cough you. You're terrified. You have no idea what's happening. And when you ask them why, they tell you an AI security camera flagged you as a threat. It thought your crumpled bag of Doritos was a gun. This isn't a Black Mirror episode. This is a true story that just happened to a 16-year-old kid in Baltimore. And it perfectly shows the single biggest problem with AI right now. putting it in charge of the real world without a human sanity check. An AI gun detection system from a company called Omniert flagged a crumpled Doritos bag he was holding as a potential firearm. Could you believe that? The crazy part isn't that the AI made a mistake. The problem is that the human system failed to catch it before an algorithm's mistake became a kid's trauma. Figure AI

just made Time magazine's best inventions of 2025 cover with figure 03. Their third generation humanoid robot built for one audacious goal to work in your home. Not in 5 years, not as a concept. They've already built Bot Q, a factory designed to manufacture 12,000 units this year, scaling to 100,000 robots over the next 4 years. This is the first humanoid robot engineered from the ground up for mass manufacturing, not lab experiments. Here's what makes Figure 03 different from every other household robot prototype. It's designed around Helix AI. It's a figures proprietary vision language action system that lets the robot learn directly from humans and reason through real world tasks. The hardware is purpose-built to enable this AI. Figure 03 features a completely redesigned sensory suite with cameras delivering twice the frame rate, one quarter the latency, and 60% wider field of view than figure 02. Each hand now has embedded palm cameras that maintain visual awareness even when reaching into cabinets or working in confined spaces where the main cameras can't see. The hands themselves represent a major leap. Each fingertip sensor detects forces as small as 3 g, that is the weight of a paperclip. This precision lets figure 03 distinguish between a secure grip and an impending slip before it happens, enabling fine grained control over fragile or irregular objects. The robot is 9% lighter than Figure02 with significantly less bulk, making it easier to navigate household spaces. It's covered in soft washable textiles instead of hard machine parts with strategically placed foam to protect against pinch points. And it charges wirelessly. Charging coils in the robot's feet let it step onto a pad and charge at 2 kW, meaning it can automatically dock and recharge throughout the day without human intervention. They're hiring human pilots who wear VR headsets and perform household tasks, generating massive training data sets for Helix AI. Every failed grasp, every drop towel, every awkward movement becomes training data. Think Tesla's approach to self-driving. Collect billions of miles from real drivers. Let the AI learn from failure at scale. Figure is doing the same thing

for hands. Alibaba dropped Quen 3 wheel and it's wild how small this thing is. There are two versions, a 2B model and a 32B model. For context, GPT and Claude are way bigger than that. Here's the crazy part. Quen 3VL 32B is beating GPT 5 mini and Claude 4 sonnet in science problems, video understanding, and agentic tasks. Think about that. It's like a 9-year-old beating the world chess champion.

Researchers from the MIT Media Lab have introduced a new frontier in personalized technology called Neuroadaptive AI. Available in the form of a headband, their system, dubbed Neurohat, is the first system to directly integrate an LLM, like GPT4, with real-time brain data. Here's where it gets absolutely insane. Neurohat continuously monitors your brain activity to calculate an engagement index. It knows if you're focused, overloaded, or just plain bored. Based on that index, the AI automatically changes its conversational style, adjusting the complexity, the tone, and the pacing of the content. If you're locked in, it offers more challenging material. If your focus dips, it instantly simplifies the explanation. But here's the tiny catch. A pilot study showed Neurohat massively increased user engagement, both measured and self-reported. — But they didn't see an immediate difference in short-term learning outcomes like quiz scores. So this technology does not make you smarter overnight. It only simplifies learning for you. This study proves that the future of learning is accessible. The researchers now envision this technology moving into everything from personalized coding assistants to creativity amplification tools that adapt to your internal cognitive state. This work signals the beginning of systems designed to collaborate with the human mind rather than just respond to it. The days of simply typing to your AI are officially over. Now your AI is listening to your mind. So think

big. To stay updated in all this chaos, subscribe so you don't miss what's dropping next. All right, I'll see you in the next