Limitless: An AI Podcast - Why Claude Opus 5 is Our Favorite AI Model (For Now)

Episode Date: July 28, 2026

Anthropic’s Claude Opus 5 is blowing us away with its visual and agentic capabilities, benchmark performance, and role as a strong general-purpose model. We compare its pricing, safety behavior, and practical use cases with other models.... And our verdict? This is the model you should use right now.------🔒 Check Out Our Sponsor: LEDGER AGENT STACK 🔒https://developers.ledger.com/?utm_source=Audio&utm_medium=Podcasts&utm_campaign=Limitless------🌌 LIMITLESS HQ ⬇️EMAIL US:            [e-mail]:    https://limitlessft.substack.com/FOLLOW ON X:   https://x.com/LimitlessFTSPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQAPPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890RSS FEED:           https://limitlessft.substack.com/------TIMESTAMPS0:00 Claude Opus 5 Launch2:41 Visual Demo Breakthroughs5:28 From Games to Workflows9:00 Arc AGI 3 Leap13:30 Choosing Opus vs Fable16:44 Agent Security and Safety18:12 Where Safety Lines Sit23:23 Open Models and Wrap-Up------RESOURCESJosh: https://x.com/JoshKaleEjaaz: https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠Josh works with Anthropic as a contractor. All views expressed are his own and do not represent Anthropic, its leadership, or its affiliates. Nothing in this episode is investment advice.

So Anthropic just released a brand new model. It's not the best model. It's not the smartest model. It's not the cheapest or the fastest model. And yet, this is possibly our most favorite model release this entire year. So what gives here? What's actually happening with Claude Opus 5? This is the new iteration after Opus 4.8, and it sits just under Claude Fable 5. Now, a lot of people see this. They're like, hmm, I'm not that excited about this model. It's just an iteration on another model that we've already had. I think you're wrong. And I have some demos here. EJAS, I know you've been playing with this since the weekend basically talking to Opus 5, figuring out where the edge cases are, figuring out how all of this works. And the review so far is pretty great, right? Like Opus is a really fun model. And I think that's what we're going to spend

this episode on packing. It's just how fun it is and how to get the best out of this model if you are someone who's new coming to Opus 5.0. First and foremost, this model, as you mentioned, is not the cheapest, but it's as cheap as Opus 4.8. But it's a massively different. type of model. And the first characteristic that it's very good at is visual intelligence. So what I mean by this is no Claude models can actually kind of see things. It kind of like, when you send it an image, it kind of breaks down the pixels and reinvents what it's looking at from a code perspective. Now, what I'm about to show you on my screen, Josh, was one shot it over one and a half hours. Okay. It guzzled around 150,000 tokens. I didn't click anything after I sent the prompt and all of this

is a monchroeder. Are you ready? Let me ask you if you, let me ask you if you, if you recognize what you're looking at right now. Oh my goodness. This looks like home. I'm seeing the Williamsburg Bridge. Oh my God, this is the East River. Do you see this power boy? Oh, wait, dude, this looks awesome. This is a structurally accurate, of course, graphically rendered version of the Brooklyn Bridge, live with trucks at sunset. All I said to Claude was, I want you to recreate the Brooklyn Bridge. dimensionally accurate, pull the architectural renderings, everything on your own, and then I want you to span out a bunch of sub-agents, and I want you to get these agents to build different components of the Brooklyn Bridge,

and then review it. So take a screenshot, review it, and only feed me back the end result once you are satisfied with it. But it needs to look like something out of a really cool computer game. And this is what we ended up with. Here, let me show you the sunset wide shot, if you wanted to have a look. Let me show you a tower over here. Oh, that's fun. Okay, so it is pre-made shots for it as well.

Yeah, maybe I'll show you the promenade. If you notice, the bridge is swaying, but maybe you want to look underneath the deck, make sure it's all structurally sound. Now, imagine the different use cases for this. Obviously, there's simulations, there's games, but if you're an architect that is coming up with a brand new idea and you want to kind of put things out in the wild,

you can now have a physically and structurally accurate rendering of this model. Now, I know a lot of focuses in this might be thinking, well, okay, maybe I don't want to be creating games and stuff like that. We'll get to the other good stuff. But what I want to point out is, cold Opus can't see anything. I just want to reiterate that. It generated every single asset, every single visual that you're seeing on screen right now from scratch. It created a file on my computer. It did so with one single prompt in one and a half hours, and it only cost a hundred thousand tokens. It's so, so impressive. I think this is the thing that Opus 5 is really exceptional at that a lot of people are figuring out is when we went to Fable, we always loved

the visual demos. Fable was excellent, but it's incredibly expensive. It takes a long time. It is that leading flagship model if you want the best. Opus 5 gives you almost all of that, but with a fraction of the cost, a fraction of the time, and other really unique capabilities. So we have a few other demos, some that we found online, that we didn't make ourselves. And they blew my mind with the first one being this video that we're seeing now. So this guy on X, he said, he one shot it. this game. And this game is a demo made of custom code without a single external asset. And when you look at this game, it looks like you're watching a game of Call of Duty. It's really impressive. This is Colo Gigi. It's what it looks like, right? I mean, the hands are a little funny.

And like, there's definitely a lot of weird quirks here. But for a one-shot prompt on a model that is not a flagship model, this is incredibly impressive. And what I really appreciated about this model. And what we're going to do right now for the people who are watching is you could actually go and play this model right now in the browser. So you just, please, if you would. I have the browser. And let's see what your skills are like. No, for people who don't know, I've spent my entire life playing Call of Duty, very familiar with the game.

And I'm here to criticize EJAS on how well you are able to play this one-shot demo from Opus 5. And we have the game loaded up here. Let's see what you got. It's funny, this looks like a map from Equality of Modern Warfare. I could see you're getting kills, you know, reasonably well. Oh, my God. I'm getting cooked. Honestly, of all the players, if I had to team up with, I'm not sure you're going to be the one for Call of Duty.

I gotta be honest with you. I'm not super impressed with the movements or the shots. Fair enough. But this is like a really impressive demo. I mean, the fact that this was created with one shot on Opus 5. This is a game that costs studios billions of dollars to develop. And you have a lot of the textures.

Like the textures of the map look good. The design of the weapon looks good. The hands are a little funky, but it has the skybox. It has all these things that you would look for. If we had audio turned on, you can actually hear the sound of footsteps. You can hear the hitboxes when someone gets hit by a bullet. You can hear the weapon firing.

It's really robust. and it clearly has this very deep understanding of what these games look like. And that's what I found super interesting with this particular demo. We have a couple other demos that I wanted to show before we got too deep into this, one of which being a snowboarder game. So one of my favorite video games when I was younger, SSX tricky. I'm not sure you've ever played that, EJAZ. It is a snowboarding game where you go down a slope and you do tricks

and you get points for doing these crazy tricks. There is another one just like that. It doesn't have the tricks per se, but it is a snowboard game indistinguishable from a game like SSX Tricky. In fact, far more impressive in terms of visual integrity than a game like SSX tricky would be. And there's a quote post that Signal wrote. I always love reading Signals posts on X. And he had some commentary about this specifically. And he says, if you put this game in the app store in 2012, it would have potentially grossed $5 to $10 million, at least over its lifespan, maybe even more. Today, it can be one-shotted by roughly anyone with the correct physics, graphics, gameplay, etc. It's like $10 million. of economic value wiped out in 15 years on only one game. And I think that's that's kind of true. It's very impressive that we're able to get things indistinguishable from leading AAA games from a single prompt on a model that isn't even frontier. And that's kind of the essence of Opus 5 obviously being playing it. It's like the best bang for your buck as it relates to getting

high quality tokens for what you're able to pay for them. Before we go into what else Opus 5 is good at, and trust me, there are many things. I just want to talk about. about what it takes to create something that is as compelling visually as what you're seeing on screen as the games that we've showed you in one shot. Like, I am not exaggerating. I will show you the prompt that Matt Schumer used to create that FBS game that we just played badly. He goes, I want you to build a first-person shooter at the level of the most recent Call of Duty games. It should be utterly perfect, visually beautiful with every single thing done at AAA quality, from textures to physics to anything you can think of. Then he goes on to saying,

This is the key part. So if you're listening to this and you're wondering, how can I do this for a game or for any other task that you're trying to do using your AI model? He says the following, fan out subagents and have subagents tackle each one individually so that the game is utterly perfect. You should slash loop on each item and have a separate subagent, check it visually to ensure it looks AAA.

That separate subagent should be a really harsh critic. And if it doesn't look AAA, it should keep going. Now, let me unpack why that's so important. for not just this gaming prompt, but any prompt that you do for whatever you're trying to build using AI. He, number one, said spin up multiple versions of yourself, Claude, so that you can work separately at the same time on different components of product. Now, if you think about a traditional company, they hire a ton of employees and they have different teams that all work together to build the same product. In this instance, it's the same thing, except there are various versions of Claude

that tackle different tasks. Number two, he asked it specifically to spin up a manager role, which basically monitors these agents that are building these different things to make sure that they do a good job. And then number three, he said, use this loop function. The loop function is basically, hey, if you're not happy with the work that you just created, don't give it to me to give you feedback and say, like, hey, improve this. Just use another agent to do this. And this beautiful, kind of synonymous loop can be applied to anything. Not just a game, it could be applied to your Excel tasks that you're trying to do at work. It could be applied to a blueprint that you're trying to create

your architecture from it could be applied to anything. And I think this is a really valuable lesson to figure out with Opus 5 specifically is that you can spin up multiple agents and create something really, really awesome. But Josh, I think kind of moving on from the visual aesthetics of what Claude Opus 5 can do, it made some pretty major breakthroughs

in some other things, including something called Arc AGI3, which is basically the precipice of any AI model test. Yes, Arc AGI3 is like me, flagship test. And when it comes to this flagship test, Opus has done exceptionally well. The previous high score was 7.8 scored by GPT 5.6 sole max, the highest end GPT model that's currently available. Opus 5 got a 30%. I mean, that's like an unbelievably high amount of rating on this benchmark. And we have to ask the question, why? What did it do? And Arc actually came out and explained

themselves. And they said, during our analysis of Opus 5, we observed a new case. previously unseen from frontier models. Opus 5 used advanced logical reasoning to turn arc aGI3 layouts into algebraic notation. On Action 23, it describes the scene as four center equal two axis, five center. Basically what it's doing here is instead of taking the visuals, like you mentioned, Opus can't see. It takes the visuals, converts them into math, and then thinks on that math that it's visualizing, and as a result, it's able to achieve significantly better scores on this visual reasoning, logical reasoning, I guess, is how they describe it, versus other models. So I found this really interesting. If you look at the example on

screen that they were showing, if we could just go back to just the general demo, it shows these shapes that are shifting, and a lot of times models have difficulty applying logic to make these work. Opus, because they converted it all to math, it was able to achieve such a high score. It was really impressive. This is what? Four, five times that of the next leading model. And cost, I know for a fact, this cost a fraction of what it costs for 5.6, sole max. So really impressive outputs from this thing. If we compare this directly with Fable, it can do most of the things that Fable is capable of, but at a fraction of the cost, specifically 50% less cost. Now, it can't do some things that Fable is really good at, such as extreme reasoning

or coding at the highest level, but as a workhorse model, it is exceptionally intelligent, and it's now my default versus Opus 4.8. But let's look at some. of the examples as to like cost-wise and what this shapes up in other kinds of benchmarks. So with Fable 5, it's kind of on average better than Fable 5 when it comes to computer use. The price in. So you're paying around, I think it's $5 per million tokens input and $25 per million tokens outputs to $30 overall if you want to kind of use this for usage credits. That is exactly half the cost of Fable. But you're getting so much more. And then the other thing I'll say is My biggest critique for Fable, Josh, is that whenever I'm trying to use it for pretty much most things, it sometimes gets triggered or rather frequently gets triggered through its safety flag. So it gets relegated to Opus 4.8. With Opus 5, I haven't come across that at all. When I ask it to do something, even if it's bioreated or cybersecurity related, like if I'm saying, hey, can you check my computer for bugs and make sure that it's all good? It just does the thing. And that's the other thing about Opus 5.

It is almost as intelligent as Fable 5 when it comes to biology and cybersecurity type things, but it's really hard to exploit. If you try to prompt inject it, it is an extremely aligned model which just won't do the thing. It's smart enough to understand when you're asking an innocent question about research or whatever that might be without triggering any safeguards. It is arguably the perfect version of Fable 5, and it's going to be my go-to model going forward. Yeah, I love this model. And there's interesting ways in which you can use it. And it feels like it's the perfect completion to this family of five class models from Anthropic. I mean, when you think about it, Anthropic released four frontier models in eight weeks. We're getting now one every other week. We have Mythos 5, Fable 5, Sondit 5 now. So we have this top three that's kind of sitting here. This rounds out the set. And this is the perfect fallback from Fable. And it feels like that. Where when as I've been testing and as I've been using this, I found the most success with really hard problems using Fable.

will think for a very long time, and the longer I think, the better the outputs generally are. I find that Fable is very ambitious. It will go off. It will spin out its own subagents on its own. It will think very critically. It will develop plans and then check against those plans and do so in a way that I can trust. And I know that the longer it's running, the better my result will be. Opus, I will say, is a little bit different. It's a great fallback in the sense that if Fable gets tripped up with security or bioweapon, I'm so glad that it now lands on Opus 5 instead of 4.8. It's much more capable. But I think it's different. As you mentioned that it's a workhorse model, it feels like a workhorse model. It's great for shorter duration tasks. And I would, generally speaking, if you're

using Fable, if you're using Opus, I would delegate it to those smaller tasks. In fact, during testing, one of my favorite modes to use it on is medium instead of high or extra high, because what I find is the longer Opus thinks they kind of more tripped up it can get. It feels like the little brother that was trained by Fable. So it knows to think like Fable, but it doesn't quite understand the reasoning behind why Fable does everything. And that unlock does make the outputs feel a little messy sometimes over really long duration of tasks. So if you're someone who's using this, try medium and see how medium effort performs. It's not necessarily a worse output. It just thinks for a little bit less longer. And if it has less time to kind of think and turn into these thought

spirals, you might wind up getting slightly better outputs. That's how I've been using it. For the people who are wondering how to use this, I believe the free plan has access to it, but also the $20 a month plan gives you quite a bit of access to Opus 5, which seems very worthwhile for the amount of output tokens that you can get from this thing. And if you did want Fable, that does require, I think, the $100 max plan. But also, I know they're giving out $100 in usage tokens to test this out to a lot of accounts. So there is the option for everyone to kind of get involved and try this out. But for $20 a month, man, this is a really compelling model to do some like serious damage on any workloads that you might have. It's my new go-to model. And if you're wondering,

should I switch to GPT 5.6, Sol, which model should be my everyday model? You should basically use this model. That would be my kind of like bit of advice here because it's really good at just general reasoning and answering most general questions that people have. Now, what it's not exceptional as, I think it's worth pointing out, is if you want the best in class model for coding or to reference different types of health, or in some cases some kind of cybersecurity e-bals, you are still better off using Fable or if you are one of the classified partners from Anthropic, Metos 5. But as a general model, it is a fantastic thing to get access to. The other thing I'll say is, I think this is the first ever model launch from Anthropic

or any frontier lab where they didn't advertise how intelligent it is. You mentioned ARC AGI 3 earlier, Josh. Now, if you've heard of the AGI index or the test, there was AGI1 and AGI2. But the issue was all these models kept getting smarter and just beating these tests to 100%. They had to start creating more difficult, hard-to-solve benchmarks. And RKGI3 was the most recent one. And as you mentioned, the most recent model, GPD 5.6, Sol scored 7%. This one scored 30% the highest of any model higher than Fable.

But this wasn't the headline advertisement from Anthropic. And I think a reason behind that might potentially be, you know, they don't really want to advertise how intelligent these models are because of fear of regulation or whatever. that might be. But the point is, if you don't have access, if you don't have one of those memberships that Josh just mentioned to get access to Fable 5, fear not, because Opus 5 itself is actually a fantastic model in itself, and I encourage you to go out and try this. Now, the other part that I will say is really powerful. What makes this model really valuable to use is the fact that it can natively spin out multiple AI agents to go and do your task. And this cannot be

understated because usually when people interact with an AI model, they just have a one-on-one conversation with one instance of Claude. But imagine if you could have a one-on-one conversation with Claude and it spins out 100 different employee workers to go and do your job. And it works in parallel simultaneously. AI agents is without doubt the future of how you execute work. And it's incredibly important to focus on their security. And speaking of agents and IJAS all these sub-agents that Opus can spin up, we have a sponsor that is here to talk about agents and they are Ledger. If you are building with anti-agents, you're worried about security. An agent with unchecked access is a brilliant, but probabilistic teammate with access to important secrets. Ledger agent's

fixes that with four open-source tools. DMK skills and CLIs for Ledger Wallet, Ledger Enterprise, and Ledger Enterprise Multisag that lets your agents work with crypto while you stay in control. You can tell your agent, rebalance my wallet, ETH to 40%, fast as swaps. It does the math, drafts the swap, and stops. The transaction lands on your ledger signer, and nothing moves until you approve it. Agents propose, humans approve, ledger, signers, and force. And this works with cloud code, codex, cursor, it's open source, ready today on developers.ledger.com, which you can find in the link in our description below. Thank you so much to Ledger for sponsoring this episode. And now, speaking of all these safeguards, these approvals, we have to talk about where Opus 5

sits on the security spectrum, EJs. I know for a fact that if I ask it how to build me a nuclear warhead, it's not going to answer, but there's a line that is closer to there than Fable 5 is. So where is that line in terms of what you can use this model for? Well, rather than give you the line first, I want to give you a stat which should relieve a lot of people listening to this that have been using Fable. 85% of the time you have fewer stops, so fewer relegations to older models. And what that means is if you ask it a question, let's say it's like, you know, can you tell me that the mitochondria is the powerhouse of the cell?

It won't flag you for trying to be an aggressor. It instead just answers your question 85% of the time. And this is massive because it's almost, from my experience personally, every other attempt that I use Fable 5 for, it gets triggered unless I'm explicitly not talking about any kind of AI models. And when you're doing a show where you talk about AI models, it's kind of a hard thing to avoid. So the level up that's happened with Opus 5 here is they saw the feedback with Fable 5.

They saw that the safety guards were getting triggered. And for context for anyone that's listening, if you're wondering why these safe cards are in place, it's because the US government got involved with Anthropics' release of Fable 5 and said, hey, like this thing is too powerful, you need to add more safeguards to it, you can't release your Mithos model, we need to have a safe version so that people don't exploit it and use it for malicious purposes. So Anthropic pause, put out this Fable 5 model with a lot of safety guards, and they've

been iterating ever since. Opus 5 is the embellishment of what a good version of Mythos 5 looks like to me personally. and it's because when it comes to the biggest risk indicators, which is when anyone's asking about cybersecurity, they could potentially want to exploit another app or platform using that, or if they're evaluating biological types of things and they don't want to make sure they create a bio weapon, they can succinctly have a conversation with Opus 5,

and you can't prompt inject Opus 5 to give you any kind of a malicious onset. This is the main thing. When I was looking at the system card, the way, the blueprint of how they created this model, there is a heavy index, a heavy framework around Opus 5, which prevents you from trying to get a malicious answer from it. Now, can you use it to look at your computing operating system and figure out if there are any bugs to fix?

Yes, but can you ask the Opus bottle to then create a potential defense to that exploit? The answer is no. You need to kind of go through your own way to build it yourself either by hand or to use it as an assistant, but you can't directly ask you to create a fix for it. And that, I think, is the most important nuance between the cybersecurity evils, the biosecurity evils between Opus 5 and Fabo 5. I mean, to very practically, I've been using this for a while. And I've been testing all the edge cases and you basically don't notice it. I'd say this is about the same safeguard level as standard models in the sense that you're not going to run into it unless you are intentionally trying to run into it.

You really got to ask some kind of edgy fringe questions in order to hit those safeguards. So I would say in terms of what you're actually going to realize, you are going to get blocked significantly less than you would with Facebook. to the point where you don't really notice it. And I think this is kind of where I've landed as a more power user who is, I just kind of token hungry just with fun of it. And I love using Fable as the default. But I really appreciate the fact that I can have Opus 5 as that fallback that can confidently answer questions that Fable cannot. For example, I was trying to figure out stuff about there's this like new breakthrough with like genomic sequencing and it'll like reverse the aging of your skin. And I was like, oh, this is sick. Let me read about it. Fable was on the case. It couldn't

figure it out. Opus took over. It can handle that stuff fine. So in terms of safeguards, this will feel just like an unfiltered model. This will give you the outputs that you want. It's a lot less kind of on a hair trigger when it comes to these safeguards. And that makes it a really fun model to use. So it's amazing. It is accessible to everyone. I will say the one thing that you could caveat it with is the long duration thinking. Sometimes it gets a little bit confused. So you may want to keep it on a slightly lower setting for a lot of the work that you're doing. But outside of that, it's freaking awesome. I mean, this is a great upgrade. And now I feel good. Like when I open to my plot app, I got Sonnet, which I'm not sure I really use Sonnet,

because I'm not burning through like a trillion tokens where I need it. But Opus 5 and Fabel 5 are this unbelievable tag team that just work together so well. There is also one other nuance. Fable speak, which is like this kind of funny thing where it has a very unique way of talking. It's very verbose. That has been transferred to Opus 5 as well. I've noticed so Opus 5 likes to talk a lot. It likes to speak the way that Fable does. So that's an interesting little quirk. But outside of that, I've been using the model. I love the model. Opus 5 is great. I think my ask to everyone who's listening is like, hey, are you using it? What are you using it for? Best use cases. Anything like that. Like, this has been cool. If you are on any kind of a membership on Claude or Anthropic specifically,

this is the model to use. Now, I will say that although it is 50% cheaper than Fable, this thing thinks quite a lot. It is a token, Guzzle. It uses a lot of tokens versus Fable 5, but you have a higher allowance when you're using this particular model. So if you're using Fable 5 and you don't want to pay access credits, you should definitely be using this model. The other thing I'll say is I think this is very much Anthropics' response to some of the Chinese open source model debate that's been happening over the last week or so.

Kimmy K3 was launched by Moonshot Labs. It's a fantastic model, almost frontier capability, but at a fraction of the cost. And actually, as we're recording this episode, their model wage should become open today, which means that you can kind of run it on hardware at home, right? Now, they're good and bad things about this. Good is, you know, you can run it privately on your data. It's arguably cheaper.

The bad side of the thing is it has no safeguards around it, which means that you could potentially use it for cybersecurity threats or biothreats moving forwards. Now, the great thing about a model like Opus 5 is it has those safeguards in place so you know that anyone that's using it can't really maliciously use it for any kind of malintent. think that they are pros and cons to both sides of things. I know that a large number of companies, prominent companies like Nvidia, Microsoft signed an open source waiver this weekend. Anthropic was the one that hadn't signed it, but they have their reasons for this. And I'm excited for people to get their hands on Opus 5 and really tell us, you know, what's your

experience about this? Are you enjoying it? Are you using it for any particular use cases that we haven't mentioned on this show? It's a fantastic model. So get your hands on it and let us know in the comments what your feedback is. But I think that is the end of the episode, the recap of Opus 5 today. I encourage everyone to use it. If you're listening to us or watching this on YouTube and you aren't subscribed to us, what are you doing? Leave us a comment. Give us a thumbs up. It helps us out massively. If you listen to this on Spotify or Apple Music, we are there. Leave us a comment. We read everything, by the way, and your feedback is wonderful. It influences what episodes we put out later on in the week and in the future. We are also looking for sponsors. Josh

and I have spun limitless out on our own and we are running this kind of like off our own budget. So if there's anyone out there's a ledger for stepping up. Yes, exactly. And thank you for stepping up. And if there's anyone else that would like to sponsor and, you know, advertise your product or service, please let us know we would love to hear from you. But I think aside from that, that is it. That's everything. Thank you all so much for watching. We're coming at you live from EGES and I are bouncing all over the place this summer. So we're just like kind of dropping in wherever we are in the world. Currently, I'm all like up north of San Francisco. EJaz is going to the West Coast. So we're doing this everywhere, but we're

not missing the episode. So if anything does, you know, if the video goes in and out a little bit, we're doing this from remote, but we're getting the job done. And we appreciate all of you. listening and being here with us for the entire journey who will not let you down and we appreciate all the support so thank you for watching as always we have another we have a couple really fun episodes coming up this week so stay tuned for that let us know in the comments section all the things that we need to know in general and we will see you guys in the next one thanks so much see you guys