Transcription
In terms of AI model releases, this has been one of the biggest weeks we have ever seen in AI. We got extended access to Fable 5. Meta and SpaceX AAI both released new models, and OpenAI released their much anticipated 5.6 series. Kind of nuts.
But the big question is, what does this actually mean for you? Do these new models actually make a difference? We're going to talk about that and a lot of other AI news in this video, like the incredible new chat GBT Live and the confusing Clawude redesign. My name is Paul J. Lipsky. Every week, I release an AI news recap video just like this one where I talk about the news that regular people actually care about. I'm not here to hype up these new models or talk about their coding capabilities. I want to share the news that will actually make a difference in your life. So, if that's what you're looking for, make sure to subscribe. And now let's get to the news.
And I want to start by talking about this update, GBT 5.6. With this update, OpenAI is giving us access to three new models: Soul, Terra, and Luna. Soul is their new flagship model, their most powerful model. That's what you're going to use for your most ambitious work. Terra is the balanced everyday model. That's great for everyday work. And then Luna is their fast, more affordable model.
Now, according to the benchmarks that OpenAI released, these new models should be some of the most powerful models that we can access. In fact, if we look at the new Terra and Soul, those are supposed to be more powerful than Claude Fable 5. And Soul is supposed to be more powerful than Mythos, which we don't even have access to. But if you guys know me, you know I don't care so much about benchmarks. I want to know in the real world, how do these perform and how are we going to be using them and will they actually make a difference?
Now, I did not get early access to these models. So, it's a little bit too soon for me to say whether I think they're better than Fable and what I'm going to be using these for long term, but I wanted to share some thoughts that other people had who have had access to 5.6 for a few weeks now. So Dan Shipper on X says that he believes that 5.6 is a much better writer than Fable. Now, that kind of blows me away because I think Fable is incredible at writing. I think it's one of its best strengths and one of the things that I was using it for.
To give you a little bit of context, I write out the scripts or the outlines for all of my YouTube videos. And I've always done that on my own because every model I've used has been pretty bad at kind of capturing my tone of voice and language and putting together a script or even just parts of a script that really sound like they're in my my tone of voice. Fable has been the only one who has nailed that at least somewhat. I think it gets about like 80% of the way there, which is pretty incredible. So if GPT 5.6 6 is that much better, that is truly mind-blowing. And for everyday tasks and knowledge work, that is a huge game-changer because GPT 5.6 is accessible. It's cheaper. We don't really have access to Fable in a few days. It's going to cost us extra. So, if we have this powerful of a model available to us for knowledge work, that is pretty incredible.
Now, just to offer a slightly different opinion, Theo OnX, who's a coder, he's a developer, so take that into consideration, he said that while 5.6 is a really good model, it's not quite as smart as Fable, but for everyday use, it is a really good go-to model. And he also said that it is incredible, probably the best in the world at computer use. And that seems to line up with the limited experience I have with GBT 5.6. It doesn't feel as smart as Fable, but it feels like an incredible everyday model that is great for knowledge work.
So, what does this mean for you? Well, as we know, Fable 5 is going to start costing us extra money on top of our subscription in just a few days, which means that really, it's not going to be easily available to most people. But GPT 5.6 6 is available. So, I think for knowledge work, GBD 5.6 makes a lot of sense. It's also excellent because you don't have to spend as much time telling it what you want it to do. It just executes and does it. And the first drafts that we're getting out of it are much closer to the finished version that we're looking for, which means that you don't have to constantly keep prompting it over and over again to get the results that you want. But of course, just because it's the newest, most powerful model doesn't mean you should blindly trust it. Always make sure that you keep yourself in the loop and you're always checking the work that the model does for you.
Before we get to the next story, I want to take a second to talk about the sponsor of today's video, the Plaude Note Pro. If you're anything like me, your best ideas don't happen when you're sitting at your computer. They happen on a phone call, in a meeting, or driving in a car. And by the time you sit down to write them out, half of it's already gone. So, for the past few months, I've been using the Plaude Note Pro to capture all of my ideas before I forget them. It's a credit card thin device that captures your conversations and turns them into a searchable knowledge base you can come back to later.
One long press and it starts recording. It has four microphones with AI beam forming, so you can leave it in the middle of the table and it'll pick up everyone clearly. Just be sure to get everyone's consent before you record. Then if something important gets said, one quick tap highlights that exact moment. When you're done, long press the button again and it stops recording. From there, the app will take over and you'll be able to see your recording at the top along with a transcript that identifies each speaker. You'll also get an AI summary that will summarize the entire conversation. It will pull out key quotes if you want and action items. And one of my favorite features is ask Plaude. So, if you have any follow-up questions about this conversation or you're trying to remember something down the line about what was said, you don't have to go through the entire transcript or even the entire summary, you can just ask Plaude here. It will even give you a citation to the exact part of the transcript that it used to answer that question.
They also have Plaude desktop now, which captures your Zoom and Google Meet calls. So, if you want every conversation, idea, and meeting captured and searchable in one place, grab the Plaude Note Pro at the link below, use my code at checkout to save. And if it's not for you, there's a 30-day return policy, so there's really no risk in trying it. All the details are in the description. And now, back to the news.
Okay, we have some other news from OpenAI, which I'll cover in a moment, but I first want to cover all the news about the new models. And we got two other new models that dropped this week. The first is Gro 4.5. This is Space XAI's newest model, and it's supposed to be built for coding, agentic tasks, and knowledge work. The benchmarks look pretty solid. It's better than Opus 4.8, but it's not quite as good as GPT 5.5 or 5.6 for that matter or Fable, but still a pretty solid model. Now, unless you're using this though for coding or agentic tasks, I don't think you're going to really notice a huge difference with it.
The other model that dropped is Meta's Muse Spark 1.1. And this one is also designed for agentic tasks and coding. Now, I personally don't use Grock all that much. I really just use it to help me find trending topics from X, and I don't use Meta's models at all. But overall, I think this is a really good thing. This puts these two companies, BAXI and Meta, really on the map. They were severely lacking before, and I think they are starting to catch up, but they're not quite there yet. I still feel like Chat GBT and Claude and Gemini still have better tools and models than what we're seeing from Grock and from Meta. But competition is always a good thing, and I'm certainly going to be keeping an eye on what they're doing.
So, let's talk now about Fable 5. I feel like this is a never-ending story. It's just so convoluted everything that has happened with Fable 5, but we have another update to it as well. So, we were supposed to effectively lose access to it on July 7th. At that point, it was supposed to switch over to additional usage, meaning that it would cost you extra on top of your subscription. Well, Anthropic extended that deadline to July 12th. And right after chat after OpenAI released chat GBT 5.6, they also reset the fable limits. So now we have an entire full use of it until July 12th. That is thousands of dollars worth of access to the best model that is publicly available in the world for free. And if you're not maxing that out, you are literally throwing away money.
So, what I want to do right now is kind of go over how to get the best or the the most out of Fable 5 while you still have so much use of it until July 12th. So, all that starts with understanding what Fable is so good at. First, it's really good at understanding you. So, when you put in a prompt, you want to put in what your goals are. And it's really just great at knowing what it is that you're looking for, even if you don't explain it very well. It also has really good taste in design. So, if you're building websites or you need new new a new website or new landing pages or slide decks, it's really good at design. And it's also really good at spotting gaps in knowledge that you didn't even know to ask about.
So, here's how I suggest that you use it. Here are some sample prompts that you can use. So, first, you can use it as a consultant, maybe like a business consultant. So, you can ask it something like, "You're my business strategist. Help me find a business I could realistically start that fits who I actually am." And then, of course, the more context you give it, like access to your knowledge base, the more it's going to know about you to answer those questions. And because it is so good at spotting the gaps and is so good at connecting the dots between what might seem like unrelated information, it's really good at being a consultant. And I've had some big breakthroughs in the way I structure my business and run it thanks to Fable and just using it as a brain uh brainstorming partner.
Another thing you can do with Fable is use it for a full life review. So basically, you could use it as a life coach. It can look at everything it knows about you and it will ask you questions about your life, your health, your work, your money to understand where you're struggling. And again, it can kind of connect those dots and see the gaps and really gives you some solid advice on ways to improve your life.
Next, you can use it to audit any existing system that you have. So on my computer, I have one folder that's like my agent folder. Every time I use Claude, I point it at that folder, give it access to it, so it's able to see the full context of everything we've ever worked on. And so what I did was I pointed Fable to that folder and said, just do an audit here. Look through everything. What is missing? What is redundant? What can be refined? And tell it not to execute right away, but instead to put together a plan that will incorporate that will tell you all the changes that it suggests that you make to your system. You can go through that and look at it and then if you're happy with that say go ahead and execute it or use a weaker model to actually execute that and then it'll make all those changes and make your entire system just run a lot smoother and be better.
Along those same lines, what I suggest you do is for one day log every single thing that you do with or without AI and then feed it into Fable and ask it, hey, this is everything I did today. How can you help me? Because a lot of times there are things that you don't even know that AI can do for you. And once you give Fable a list of everything that you do every day, it will tell you, I can help you with this. I think I can help you with this. We can try setting up a system for this. And that's how you go from just having a chatbot to a full assistant.
And the final prompt I have for you is any sort of website design. So, if you have a site that you already have and need it to be improved, you can just ask it to go through it and to suggest changes where things can be clarified, new copy, or if you're starting from scratch, you can ask it to create a new website for you completely from scratch. I'll have all these prompts down below if you want to grab these and use them yourself. Just be sure you use them before July 12th so that you can get the most out of Fable.
All right, we're done talking about the models. Let's talk about some other updates from this week. The most exciting one I want to talk about is from Open AI. They have a new super app called Chat GBT work. This is an environment where everyday knowledge workers can use OpenAI's model to help them with knowledge work. So, I've made videos in the past about Claude or sorry, OpenAI's Codeex, which is an app you download onto your computer that's very similar to Claude Code and Claude Co-work that has now been rebranded into the Chat GBT app. So, this app right here used to be Codeex. It is now called the Chat GBT app. And up here you can switch between work and codeex, similar to how you can switch between Claude Co-work and Claude Code in the Claude desktop app. If that analogy makes sense to you, if it doesn't, don't worry. All you need to know is that if you're a coder, you would want to use Codeex. That's for coding. And for more knowledge everyday work, that is what you use Chat GBT work.
So using this, you can add these projects on the side. And projects are just folders on your computer. So I have a folder here called invoices. That's where I obviously have a lot of invoices in there and like manage and keep them. I have one with some health information in here. I have one where I keep like a business journal. And then this one right here called Frankbbot. This is one folder that contains all of my other projects and folders. So these are all folders locally on my computer and they are linked up with these projects on Chat GBT work. So if I were to start a new chat inside this folder and I tell it to look for a certain spreadsheet I have on my computer, it's going to be able to find that. It has access to this entire folder on my computer. It can add new folders. It can change uh it can add new files. It can change existing files. So, it really allows you to work where you already do your work on your computer.
So, that actually isn't all that new. You could already do that with Codeex, but this is a rebranding. And I think this is a really smart idea because even though I've made videos about Codeex in the past, it was sort of confusing because people searching for Codeex, most of them are developers and coders. And the people I'm trying to reach are I'm trying to show people that they can use this tool for everyday tasks. Think like emails, creating documents, creating spreadsheets, uh creating slideshows. And but the problem is when people search, they're getting a lot of videos about coding and developing. And so now we have this distinction. I can make videos about Chat JBD work and everyone knows exactly what I'm talking about. So, I think this was a really smart move for them to do this and to make this distinction so that more knowledge workers know that this is a very powerful tool that they can use.
There is a whole lot you're going to be able to do with this new Chat GBT work super app. If you want to see the prior video I did about Codeex, you can find that right up here in the card. Everything you could do in that video with Codeex, you're going to be able to do with Chat GBT work and a lot more. So, I will be doing a full video on how to get the most out of Chat GBD work very soon. So, make sure you subscribe so you don't miss out on that one.
The other big update from Open AI is that we now have a better Chat GBT Live. It is an update to their voice model that sounds a lot more natural and humanlike. This is now available for all users. So, let me start a new voice chat and show you what makes this so unique.
"Hey, I want to practice a speech or a presentation that I'm going to give. Feel free to jump in if you hear me say anything that's inaccurate."
"Totally. I'm ready. Just go ahead and I'll only step in if something's off."
"Our solar system is incredible. We have the Earth and the Sun, which is 10 million miles away. And we also have the Moon."
"Quick correction there. The sun's actually about 93 million miles away on average, not 10 million. Everything else sounding good?"
"Okay, so the sun is 93 million miles away and it's a star and it's actually the biggest star in the Milky Way galaxy, which is pretty cool when you think about it because..."
"Tiny fix. Again, the sun's a star, yes, but it's not the biggest in the Milky Way. Not even close. It's more of a medium-sized star."
"Oh, okay. Um, how can I improve this presentation?"
"Start by leaning into that sense of wonder you already have. Then in creating a clean accurate fact or two, maybe say something like..."
"Can I talk about how we have like we're one of the few that actually has well probably the only one we know of that actually has life on one of the planets and how that could be pretty unique?"
"Yeah. So what I'm trying to highlight for you there is that it's able to both talk and listen at the same time while you're talking. So, it's not going to uh in the past, it used to just interrupt you sort of randomly before you were done talking. Now, it interrupts you with purpose. It heard me say something that was incorrect and it jumped in and told me, 'Hey, that fact was incorrect.' And you'll also notice that as I was speaking, it would say things like, 'Yeah,' or 'Uh-huh,' just to acknowledge that it was listening. So, it really feels a lot more conversational than the prior version. You kind of have to use it to really feel that and understand it.
There's also a lot more happening behind the scenes. So, it's able to delegate to more powerful models. So, as you ask questions or interact with it, it's it's it's delegating to those models to be able to help you better in real time. And the speed of it is quite remarkable. It's definitely something you should try out for yourself for yourself, especially since it is free and now available to all users."
Next, let's talk about Anthropic and what they've been up to. We got some good news for anyone who likes Claude Co-work. Anthropic has doubled the Co-work usage limits through August 5th, so that's good news. They've also made some updates to the Claude desktop app. In the past, up here on the top left, we used to have three tabs: one for chat, one for Co-work, and one for code. Now, they've collapsed together chat and Co-work into home. And when you come over to the chat here, you'll have to switch here between chat and Co-work.
Now, at first, this didn't make a lot of sense to me because if we have to choose here anyway, why not just leave it up here? But what they've also done is basically combine chat and Co-work entirely. So all your chat and Co-work folders are now one. You can both chat in fold in inside of projects or do Co-work tasks inside of them. You'll see all of them here. And you have one running list of recent both chats and tasks. The chats are identified by this chat icon and the tasks are identified by this check mark icon. So personally, I like this. I think it looks a lot cleaner and now I'll actually start using chat again. I wasn't using chat for months because I use Co-work for pretty much everything. And then if I do if I did have a chat like question, I would just use Co-work anyway because I wanted it all in one space. So it is nice for me that all of this stuff on the left has been collapsed together. But I think some people are not going to be happy with this change.
If you come over to the web version of Claude, you'll also notice that you have access to Co-work inside of here as well. So this is what Anthropic has said about it. So they've said that Claude Co-work is coming to mobile and web. Hand Claude a task on your desk and pick it up finished. Pick up your finished work from the phone. Close the laptop and Claude keeps going. But here's the thing, it still can only do that for cloud-based tasks.
If we return back to the desktop app, one of the reasons I love Co-work and why most people love Co-work is because it gives Claude access to local files and folders on your computer. So with Co-work selected on this dropdown, I can select this YouTube folder. That's a folder on my computer. I don't have it like in Google Drive, for instance. It's a folder on my computer. And now I can say, "Help me with the outline of one of my videos inside this YouTube folder," and it will be able to do that.
If you try to do the same thing using the web version of Claude Co-work, you can't do it, or at least I have not been able to do it. It keeps telling me that it is unable to access local files and folders on my computer. And if I want to run this task, I have to do it from the desktop Claude app. Same thing if I try to start the task using Co-work inside the iOS Claude. So, I'm just a little bit confused as to why we now have access to Claude Co-work on web and on our phones. What does that really add over just using regular chat? I'm not quite sure yet, to be honest, but I just got access to this. I haven't had a lot of time to really play around with it. But yeah, that just kind of leaves me wondering if we can access local files and folders, which is one of the most, you know, standout features and the reason we use Claude Co-work, then what is the purpose of it?
Once I figure that out, I'll be making a whole video about it. In fact, I'm going to be making an entire video about this whole new Claude ecosystem because it looks so different now. The fact that we get Co-work on web, the fact that the Claude desktop app looks so much different. So, I want to do a whole video just dedicated to this. If you want to make sure you don't miss that, make sure you definitely subscribe.
So, that's all the latest AI news from the past week. If you enjoyed this video, do me a quick favor, subscribe to the channel. I release a news recap like this every week. So, if you want to stay informed and on top of the latest AI news that actually matters, you do want to make sure you are subscribed. Otherwise, thanks so much for watching and I'll see you in the next video. Bye for now.