The Best Free AI Just Got WAY Better & More AI Use Cases

Hello. Quick voice warm up. Welcome to another week in AI where all these consumer products are not just maturing, but they're really competing for your attention. Claude now has some of his best features and models available on the free tier. While OpenAI is trying to win the PR battle with several stories. And then we have China doing its own thing with all of these new models and a few useful features in between. This has really been a busy week in AI. And on this show, music and news, we pull together all the different stories. We filter for the ones that matter to you and I get to present them back to you. So, let's get into it cuz there's a lot to talk about as I

mentioned, starting with the Claude stories. There was a phase about a month or two ago where Google was just ripping story after story, feature after feature, new releases, and then it seems like Claude took over from them, and now they're on this streak of just releasing useful stuff to consumers. And this week is no different. There's really two big components to this. First of all, is the Sonnet 4. 6 six release. We'll talk about that in a second. And secondly is them bringing some of the more premium features to the free plan. I think most interestingly, they added the ability to create Microsoft Office documents within Claude on the free tier now. And Claude was already bestin-class at this in my opinion. Creating Excel sheets, creating Word documents. If you ask for it, it will just do it. And you can do that on the free plan now, too. There's a few more things like connectors to certain applications. I personally don't use those much, but they're available on the free tier now. and the ability to further refine your custom instructions making the free product hm probably the most capable one. I mean especially in the light of another story that we'll talk about later that chat GPT now started rolling out ads on the free tier. You have a adree super potent product on this side with claude. It's just hard to compete if claud is putting all these features into the free tier on a completely adree product. Plus there's also the new model release sonnet 4. 6. So let's talk about that briefly. This is not the biggest story, but it's essentially a massive upgrade to their most costefficient model. If you're not familiar with the Claude ecosystem, they have Opus, which is their state-of-the-art model, the best they have, but on the web app, it's not accessible to the free tier, understandably so. And when you use it for the API, it's really pricey. Sonnet, on the other hand, might not be quite as smart on complex reasoning task and some of these agentic use cases, but for most users, it usually hit the sweet spot between performance and cost. And now they had a massive upgrade to their sonnet models. They've actually been upgraded so much that they get super close to Opus level performance at a fraction of the cost. Now that's for people using it for the API for people just using claude and being on the free account. Well, the good news is you have sonnet now too. And to link it back to use cases that you can take advantage of today, they especially highlighted how good it is at knowledge work, especially creating word documents and Excel sheets. So, not just that is available to you in the free plan now, but you also have a way upgraded model that does this way better than anything before on the free plan. Worth considering if you're into that. I would just say if you haven't given Claude a try it, this is the perfect time to do it. And if you're creating Word documents or Excel sheets, then it's kind of a no-brainer. Okay, next up, I

want to highlight this feature again relating to Word documents or in this case, Google Docs. I suppose I like talking about those because I know that most viewers of this channel will be working with documents and there's a new feature that I really like that Google brought into here. And if you're familiar with the notebook LM audio summaries where it turned the document into podcast, well, they basically brought that feature to Google Docs. So if you have a document and don't feel like reading it, you should be able to do this right now. Although in typical Google fashion, this hasn't rolled out to everybody yet. They just announced it and neither me nor the team have this yet. It's super fresh though. So if you go to tools, audio, you should be able to summarize your documents. So as you can see for me, I have the listen to this tab now, but I don't have the summary. So this just turns it into audio whereas the other one compresses it. check your own account and start using this. I think this is really

useful. Okay, so now let's get to OpenAI updates. There has been a few, but nothing really that you will be able to put to work right away. I mean, first of all, they started rolling out this ads feature. I don't personally actually know anybody who's seen the ads there yet, but to be fair, most people that I know are on the pro accounts, so I would love to hear from you in the comment section if you've been hit over the head with these chatb ads yet. Furthermore, they released a bunch of news and that's not what the show is about. I just want to quickly touch on it because it is really interesting. They're trying to get the public's attention through worthy stories like GPT 5. 2 derives a new result in theoretical physics. Apparently, this is a verified discovery. If you want to read about it, there's a blog article about it now. And then also, they hired the founder of OpenClaw, which we'll talk about in a second, and I cover it in separate videos. It's basically that crazy eye agent that just does things these days. There's a lot of security risks with it, but it's really potent and people have been making it work. I strongly recommend you check out my other video that we uploaded yesterday, I think, where I give my take on it because it really is revolutionary, but I don't recommend most people use it. It needs a few more weeks or months for it to be turned into a consumer product, but now it's just crazy but also powerful. And OpenAI hired the guy who made it, Peter Steinberger. Fun fact, when I first moved to Austria, we moved to a little village called Steinbeck, which is his name. It's Steinbag and then anyway, I don't think he's from there, but just wanted to throw that in. Oh, and as we're talking about that, another fun fact is that one village over was the birthplace of Arnold Schwarzenegger. Um, I don't know what you're going to do with these facts, but I the next one cuz this is actually quite useful. We're talking about OpenAI models and model releases, right? And I'm going to clump a few stories here together because there have been several model releases this week. We have GPT 5. 3 Codex Spark. Okay, that's a coding model that is just more efficient, nothing revolutionary for most people. But then inside of chat, this is something that I don't know, I suppose was overdue. They cleaned up their legacy models. If you look into the interface and go to legacy models, you see that a lot of them disappeared. Many people used to prefer GPT40 and that thing is actually gone now. They really like the tone of it. They got kind of built the emotional connection with it. I guess GBT 4. 1 and a few more are gone. So, they're really cleaning up this experience. I know that some people were still relying up on some of these models and they just removed them from the web interface, which honestly I think is a good move. this product should be kept simple, but I understand why some people are upset about this.

And then beyond the new model from OpenAI and Claw that we featured earlier, there's also a lot of Chinese models. Now, as you know, usually on this show, we really don't go much into examples cuz there's no point. Like, they produce very similarish results. They're close to the Frontier models like GPT 5. 2 or Opus 4. 6, but they're 10 or 50 times cheaper or something like that. It's no difference with these stories. So the new models that came out this week are Quen 3. 5, Mini Max M2. 5, and Seed 2. 0 from Bite Dance. So CE 2. 0, for example, almost matches GPT 5. 2 in performance, but it's nearly 110th the price. Chinese don't mess around. And then the interesting thing about Miniax M2. 5 is that concretely on the Agentic tasks, it's close to OPOS 4. 6 or GPT 5. 2. And then Quen 3. 5 actually claims to beat GPT 5. 2 2 and Gemini 3 Pro at one/ird the price. Okay, so that's a lot of information, but basically what this means is China is not asleep at the wheel. Matter of fact, they're moving full throttle forward. And I think in the context of the next story that I want to talk about, which is Open Claw and all of this agentic AI assistant madness that has been going on over the past few weeks, these model releases really do matter because if your computer is capable enough, you don't have to pay 100, 200, $300 a day anymore to experiment with some of these agents. Now, if at this point in the video you're just confused and it's like, "What's all these models? What's open claw? What's Oh my god, this is a lot. " Yeah. Yeah. It is. I agree. Even when I was going through this and preparing for this video, I'm like, "How do I even fit all of this in here and make it not be confusing? " I think the answer is just it's not possible to a certain sense if I want to cover all of these significant stories. But I think also partial answer is just, hey, some of these topics are really big and the philosophy I follow now is I create standalone videos on them. So, I think this entire open claw story, the agent assistant that has been going viral recently, deserves its own attention in a separate video, and that's why I made that. So, check out that upload if you want to get caught up on my latest thoughts on that. Long story short, first I thought it's no big deal, but now I'm actually blown away and it's kind of reorganizing my worldview. But

while Open Claw might be a separate conversation, there are some competitors that have already been popping up. So, if you watch that video or you're just aware of Open Claw and it's too technical, too expensive, too risky for you. I think that's a fair assessment. That would be my recommendation. There are some alternatives that came out now. The first one would be Manos AI coming out with agents. And while honestly, I haven't tested this much. It just seems like an inferior version of Open Claw where I spend all my time right now, it does look super userfriendly. And regularly, I have someone talking to me about Manos and really praising it for how much they like it in their everyday workflow. So my recommendation would be if you're interested in open claw but don't want to put up with all the negatives then check out either Manus or the second release Kim Claw or there's also Chinese alternative. I don't know if I can recommend this one though. But look, Kim Claw is basically their deployment of it where they made it really easy to use. If you want to experiment, this might be worth it. Other than that, right now in the consumer landscape, Manos is really trying to be the personal assistant agent for everybody. And I guess if we're talking about that category already, I should also point out that Claude's version of this that is arguably a bit less capable, but also more secure and even easier to use. Claude Co-work is now finally available on Windows. I know a lot of viewers when I made videos about Cloud Co-work and what you can do with it. We're like, ah, why is this Mac only? Well, now your time has come. It's finally available on Windows also. So many options to get your hands dirty in this agentic space. I created videos on cloth co-work use cases. Those would also work in Manus, Open Claw, Kimclaw, all of those. You'll find all these links below. And ultimately, I would just say just take a few hours and follow your curiosity. Whatever in this video catches your attention, that's probably the thing where you should be spending a little bit of time to poke around yourself. I

have a few more quick hits for you, things that are relevant for you to stay on top of this and might just be interesting conversation starters or if that is your specific domain, you might want to go deeper. But I think with this many stories, it's just important to compress some of the stuff because it's just so much. So, let me go through the things that happened this week that I thought were interesting enough for you to know about. And this story I thought was so interesting because it contextualizes all this madness that we're talking about here. It's that Spotify developers admitted to not having written any code themselves since December. They're improving what is probably the world's most popular music streaming app. Is that right? With AI written code only. We talked about this reality for a long time, but now it's happening in front of us. If you're not a developer and not in that world, then there you go. This is your update on how people are actually using this

stuff. The next one is also interesting. The Vatican launched an AI translation service where it translates its masses into 60 different languages for people. This is the bright side. I love that. I think this should be available for most cultural experiences. And the way it works is kind of interesting. It starts with a QR code scan on your phone and then you get a real-time translation in your language. Pretty neat. And that's pretty much it for this week. I know this was a lot and it was maybe more like keeping up with where the space is moving right now rather than a ton of new buttons to click for you this week, but nevertheless, I hope this was helpful. I expect a wave of more technical releases now to be followed by a wave of more consumer focused ones and I'll be here to cover those for you too every Friday. I hope you have a wonderful weekend and I'll see you soon.