July 28, 2026 • 26 mins
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: info@limitless.fm
NEWSLETTER: https://limitlessft.substack.com/
FOLLOW ON X: https://x.com/LimitlessFT
SPOTIFY: https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQ
APPLE: https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890
RSS FEED: https://limitlessft.substack.com/
------
TIMESTAMPS
0:00 Claude Opus 5 Launch
2:41 Visual Demo Breakthroughs
5:28 From Games to Workflows
9:00 Arc AGI 3 Leap
13:30 Choosing Opus vs Fable
16:44 Agent Security and Safety
18:12 Where Safety Lines Sit
23:23 Open Models and Wrap-Up
------
RESOURCES
Josh: https://x.com/JoshKale
Ejaaz: https://x.com/cryptopunk7213
------
Not financial or tax advice. See our investment disclosures here:
https://www.bankless.com/disclosures
Josh 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.
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Josh:
So Anthropic just released a brand new model. It's not the best model. (00:00):
undefined
Josh: It's not the smartest model. (00:03):
undefined
Josh: It's not the cheapest or the fastest model. And yet this is possibly our most (00:04):
undefined
Josh: favorite model release this entire year. (00:09):
undefined
Josh: So what gives here? What's actually happening with Claude Opus 5? (00:11):
undefined
Josh: This is the new iteration after Opus 4.8 and it sits just under Claude Fable 5. (00:15):
undefined
Josh: Now a lot of people see this, they're like, hmm, I'm not that excited about (00:19):
undefined
Josh: this model. It's just an iteration on another model that we've already had. (00:22):
undefined
Josh: I think you're wrong. And I have some demos here. Ejaz, I know you've been playing (00:26):
undefined
Josh: with this since the weekend, basically, talking to Opus 5, figuring out where (00:30):
undefined
Josh: the edge cases are, figuring out how all of this works. (00:35):
undefined
Josh: And the review so far is pretty great, right? Like Opus is a really fun model. (00:38):
undefined
Josh: And I think that's what we're going to spend this episode unpacking, (00:42):
undefined
Josh: is just how fun it is and how to get the best out of this model if you are someone (00:44):
undefined
Josh: who's new coming to Opus 5.0. (00:48):
undefined
Ejaaz: First and foremost, this model, as you mentioned, is not the cheapest, (00:50):
undefined
Ejaaz: but it's as cheap as Opus 4.8. (00:54):
undefined
Ejaaz: But it's a massively different type of model. And the first characteristic that (00:57):
undefined
Ejaaz: it's very good at is visual intelligence. (01:00):
undefined
Ejaaz: So what I mean by this is no clawed models can actually kind of see things. (01:03):
undefined
Ejaaz: It kind of like when you send it an image, it kind of breaks down the pixels (01:07):
undefined
Ejaaz: and reinvents what it's looking at from a code perspective. (01:11):
undefined
Ejaaz: Now, what I'm about to show you on my screen, Josh, was one-shotted over one and a half hours, okay? (01:14):
undefined
Ejaaz: It guzzled around 150,000 tokens. (01:20):
undefined
Ejaaz: I didn't click anything after I sent the prompt and all of this is one shot. Are you ready? (01:24):
undefined
Ejaaz: Let me tell you if you, let me ask you if you recognize what you're looking at right now. (01:30):
undefined
Josh: Oh my goodness, this looks like home. I'm seeing the Williamsburg Bridge. (01:34):
undefined
Josh: Oh my God, this is the East River. (01:37):
undefined
Josh: Do you see this, Pat Boy? Oh, wait, dude, this looks awesome. (01:40):
undefined
Ejaaz: This is a structurally accurate, of course, graphically rendered version of (01:44):
undefined
Ejaaz: the Brooklyn Bridge live with trucks at sunset. (01:50):
undefined
Ejaaz: All I said to Claude was, I want you to recreate the Brooklyn Bridge, (01:54):
undefined
Ejaaz: dimensionally accurate, pull the architectural renderings, everything on your own. (01:59):
undefined
Ejaaz: And then I want you to span out a bunch of sub agents. (02:05):
undefined
Ejaaz: And I want you to get these agents to build different components of the Brooklyn (02:09):
undefined
Ejaaz: Bridge and then review it. (02:13):
undefined
Ejaaz: So take a screenshot, review it, and only feed me back the end result once you (02:15):
undefined
Ejaaz: are satisfied with it. But it needs to look like something out of a really cool computer game. (02:20):
undefined
Ejaaz: And this is what we ended up with. Here, let me show you the sunset that wide (02:26):
undefined
Ejaaz: shot, if you want to have a look. (02:29):
undefined
Ejaaz: Let me show you a tower over here. (02:30):
undefined
Josh: Oh, that's fun. Okay, so it is pre-made shots for it as well. (02:32):
undefined
Ejaaz: Yeah, maybe I'll show you the promenade if you notice the bridge is swaying, (02:35):
undefined
Ejaaz: but maybe you want to look underneath the deck, make sure it's all structurally sound. (02:38):
undefined
Ejaaz: Now, imagine the different use cases for this. Obviously, there's simulations, (02:42):
undefined
Ejaaz: there's games, but if you're an architect that is coming up with a brand new (02:45):
undefined
Ejaaz: idea and you want to kind of put things out in the wild, you can now have a (02:49):
undefined
Ejaaz: physically and structurally accurate rendering of this model. (02:53):
undefined
Ejaaz: Now, I know a lot of folks listening to this might be thinking, (02:56):
undefined
Ejaaz: well, okay, maybe I don't want to be creating games and stuff like that. (02:58):
undefined
Ejaaz: We'll get to the other good stuff. But what I want to point out is cold Opus (03:01):
undefined
Ejaaz: can't see anything. I just want to reiterate that. (03:05):
undefined
Ejaaz: It generated every single asset, every single visual that you're seeing on screen (03:07):
undefined
Ejaaz: right now from scratch. It created a file on my computer. (03:11):
undefined
Ejaaz: It did so with one single prompt in one and a half hours, and it only cost 100,000 (03:15):
undefined
Ejaaz: tokens. It's so, so impressive. (03:19):
undefined
Josh: I think this is the thing that Opus 5 is really exceptional at that a lot of (03:21):
undefined
Josh: people are figuring out is when we went to Fable we always loved the visual demos. (03:25):
undefined
Josh: Fable was excellent but it's incredibly expensive. It takes a long time. (03:28):
undefined
Josh: It is that leading flagship model if you want the best. Opus 5 gives you (03:32):
undefined
Josh: Almost all of that, but with a fraction of the cost, a fraction of the time, (03:36):
undefined
Josh: and other really unique capabilities. So we have a few other demos, (03:40):
undefined
Josh: some that we found online that we didn't make ourselves. (03:43):
undefined
Josh: And they blew my mind with the first one being this video that we're seeing (03:45):
undefined
Josh: now. So this guy on X, he said he one-shotted this game. (03:49):
undefined
Josh: And this game is a demo made of custom code without a single external asset. (03:54):
undefined
Josh: And when you look at this game, it looks like you're watching a game of Call of Duty. (03:59):
undefined
Josh: It's really impressive. This is Call of Duty. It's what it looks like, (04:02):
undefined
Josh: right? I mean, the hands are a little funny and like there's definitely a lot of weird quirks here. (04:05):
undefined
Josh: But for a one shot prompt on a model that is not a flagship model, (04:08):
undefined
Josh: this is incredibly impressive. (04:12):
undefined
Josh: And what I really appreciated about this model and what we're going to do right (04:14):
undefined
Josh: now for the people who are watching is you could actually go and play this model (04:17):
undefined
Josh: right now in the browser. (04:20):
undefined
Josh: So please, if you would, I have the browser and let's see what your skills are (04:21):
undefined
Josh: like. Now, for people who don't know, I've spent my entire life playing Call (04:26):
undefined
Josh: of Duty, very familiar with the game. (04:29):
undefined
Josh: And I'm here to criticize Ejaz on how well you are able to play this one-shot (04:31):
undefined
Josh: demo from Opus 5. And we have the game loaded up here, Ejaz. Let's see what you got. (04:36):
undefined
Josh: It's funny, this looks like a map from Call of Duty Modern Warfare. (04:40):
undefined
Josh: I could see you're getting kills, you know, reasonably well. (04:43):
undefined
Ejaaz: Oh my God, I'm getting cooked. (04:45):
undefined
Josh: Honestly, of all the players, if I had to team up with, I'm not sure you're (04:47):
undefined
Josh: going to be the one for Call of Duty. (04:51):
undefined
Josh: I got to be honest with you, I'm not super impressed with the movement or the shots. (04:53):
undefined
Josh: Fair enough. But this is like a really impressive demo. I mean, (04:56):
undefined
Josh: the fact that this was created with one shot on opus 5 this is a game that costs (04:58):
undefined
Josh: studios billions of dollars to develop and you have a lot of the textures like (05:03):
undefined
Josh: the textures of the map look good (05:06):
undefined
Josh: the design of the weapon looks good the hands are a little funky but it has (05:08):
undefined
Josh: the skybox it has all these things (05:12):
undefined
Josh: that you would look for if we had audio turned on you can actually hear (05:13):
undefined
Josh: the sound of footsteps you can hear the hit boxes when someone gets hit by a (05:16):
undefined
Josh: bullet you can hear the weapon firing it's really robust and it clearly has (05:19):
undefined
Josh: this very deep understanding of what these games look like and that's what i (05:23):
undefined
Josh: found and super interesting with this particular demo. (05:26):
undefined
Josh: We have a couple other demos that I wanted to show before we got too deep into this. (05:28):
undefined
Josh: One of which being a snowboarder game. So one of my favorite video games when (05:32):
undefined
Josh: I was younger, SSX Tricky. (05:36):
undefined
Josh: I'm not sure you've ever played that, Ejaz. It is a snowboarding game where (05:37):
undefined
Josh: you go down a slope and you do tricks and you get points for doing these crazy tricks. (05:41):
undefined
Josh: There is another one just like that. It doesn't have the tricks per se, (05:46):
undefined
Josh: but it is a snowboard game indistinguishable from a game like SSX Tricky. (05:49):
undefined
Josh: In fact, far more impressive in terms of visual integrity than a game like SSX (05:53):
undefined
Josh: Tricky would be. And there's a quote post that Signal wrote. (05:59):
undefined
Josh: I always love reading Signal's post on X. (06:02):
undefined
Josh: And he had some commentary about this specifically. And he says, (06:04):
undefined
Josh: if you put this game in the App Store in 2012, it would have potentially grossed (06:07):
undefined
Josh: $5 to $10 million, at least over its lifespan, maybe even more. (06:11):
undefined
Josh: Today, it can be one-shotted by roughly anyone with the correct physics, (06:15):
undefined
Josh: graphics, gameplay, etc. (06:18):
undefined
Josh: It's like $10 million of economic value wiped out in 15 years on only one game. (06:20):
undefined
Josh: And I think that's kind of true. It's very impressive that we're able to get (06:25):
undefined
Josh: things indistinguishable from leading AAA games from a single prompt on a model (06:28):
undefined
Josh: that isn't even Frontier. And that's kind of the essence of Opus 5 as we've (06:33):
undefined
Josh: been playing it. It's like the best (06:38):
undefined
Josh: bang for your buck as it relates to getting high quality tokens for what you're able to pay for them. (06:40):
undefined
Ejaaz: Before we go into what else Opus 5 is good at, and trust me, (06:45):
undefined
Ejaaz: there are many things, I just want to talk about what it takes to create something (06:49):
undefined
Ejaaz: that is as compelling visually as what you're seeing on screen as the games (06:53):
undefined
Ejaaz: that we've showed you in one shot. (06:57):
undefined
Ejaaz: Like, I am not exaggerating. I will show you the prompt that Matt Schumer used (06:58):
undefined
Ejaaz: to create that FBS game that we just played badly. (07:02):
undefined
Ejaaz: He goes, I want you to build a first person shooter at the level of the most (07:06):
undefined
Ejaaz: recent Call of Duty games, it should be utterly perfect, visually beautiful, (07:09):
undefined
Ejaaz: with every single thing done at AAA quality, from textures to physics to anything you can think of. (07:13):
undefined
Ejaaz: Then he goes on to say, and this is the key part, so if you're listening to (07:19):
undefined
Ejaaz: this and you're wondering, how can I do this for a game or for any other task (07:22):
undefined
Ejaaz: that you're trying to do using your AI model, he says the following, (07:25):
undefined
Ejaaz: fan out sub-agents and have sub-agents tackle each one individually so that (07:29):
undefined
Ejaaz: the game is utterly perfect. (07:34):
undefined
Ejaaz: You should slash loop on each item and have a separate sub-agent check it visually (07:35):
undefined
Ejaaz: to ensure it looks AAA. That separate sub-agent should be a really harsh critic. (07:41):
undefined
Ejaaz: And if it doesn't look AAA, it should keep going. (07:45):
undefined
Ejaaz: Now, let me unpack why that's so important for not just this gaming prompt, (07:48):
undefined
Ejaaz: but any prompt that you do for whatever you're trying to build using AI. (07:51):
undefined
Ejaaz: He, number one, said, spin up multiple versions of yourself, (07:55):
undefined
Ejaaz: Claude, so that you can work separately at the same time on different components of product. (07:59):
undefined
Ejaaz: Now, if you think about a traditional company, they hire a ton of employees (08:05):
undefined
Ejaaz: and they have different teams that all work together to build the same product. (08:08):
undefined
Ejaaz: In this instance, it's the same thing, except there are various versions of (08:11):
undefined
Ejaaz: Claude that tackle different tasks. (08:15):
undefined
Ejaaz: Number two, he asked it specifically to spin up a manager role, (08:17):
undefined
Ejaaz: which basically monitors these agents that are building these different things (08:21):
undefined
Ejaaz: to make sure that they do a good job. (08:24):
undefined
Ejaaz: And then number three, he said, use this loop function. The loop function is (08:25):
undefined
Ejaaz: basically, hey, if you're not happy with the work that you just created, (08:29):
undefined
Ejaaz: don't give it to me to give you feedback and say like, hey, improve this. (08:33):
undefined
Ejaaz: Just use another agent to do this. And this beautiful. (08:36):
undefined
Ejaaz: Kind of synonymous loop can be applied to anything, not just a game. (08:40):
undefined
Ejaaz: It could be applied to your Excel tasks that you're trying to do at work. (08:43):
undefined
Ejaaz: It could be applied to a blueprint that you're trying to create your architecture (08:46):
undefined
Ejaaz: from. It could be applied to anything. (08:50):
undefined
Ejaaz: And I think this is a really valuable lesson to figure out with Opus 5 specifically, (08:51):
undefined
Ejaaz: is that you can spin up multiple agents and create something really, really awesome. (08:56):
undefined
Ejaaz: But Josh, I think kind of moving on from the visual aesthetics of what Claude (09:00):
undefined
Ejaaz: Opus 5 can do, it made some pretty major breakthroughs in some other things, (09:04):
undefined
Ejaaz: including something called Arc AGI 3, which is basically the precipice of any AI model test. (09:09):
undefined
Josh: Yes, Arc AGI 3 is like the flagship test. And when it comes to this flagship (09:15):
undefined
Josh: test, Opus has done exceptionally well. (09:20):
undefined
Josh: The previous high score was 7.8, scored by GPT 5.6 Sol Max, the highest end (09:23):
undefined
Josh: GPT model that's currently available. Opus 5 got a 30%. (09:30):
undefined
Josh: I mean, that's like an unbelievably high amount of rating on this benchmark (09:33):
undefined
Josh: and we have to ask the question why what did it do and (09:38):
undefined
Josh: arc actually came out and explained themselves and they said during our analysis (09:42):
undefined
Josh: of opus 5 we observed a new capability previously unseen from frontier models (09:45):
undefined
Josh: opus 5 used advanced logical reasoning to turn arc agi 3 layouts into algebraic notation (09:50):
undefined
Josh: on action 23 it describes the seen as four center equal two axis five center (09:56):
undefined
Josh: basically what it's doing here is instead of taking the visuals you just like (10:01):
undefined
Josh: you mentioned opus can't see (10:05):
undefined
Josh: it takes the visuals converts them into math and then thinks on that math that it's (10:07):
undefined
Josh: visualizing and as a result it's able to achieve significantly better scores (10:12):
undefined
Josh: on this visual reasoning (10:16):
undefined
Josh: logical reasoning i guess is how they describe it versus other models so i find (10:17):
undefined
Josh: this really interesting if you look at the example on screen that they were (10:21):
undefined
Josh: showing if we could just go back to just the general demo it shows these shapes that are shifting (10:24):
undefined
Josh: and a lot of times models have difficulty (10:29):
undefined
Josh: Applying logic to make these work. Opus, because they converted it all to math, (10:32):
undefined
Josh: it was able to achieve such a high score. It was really impressive. (10:36):
undefined
Josh: This is what, four or five times that of the next leading model? (10:38):
undefined
Josh: And cost, I know for a fact, this cost a fraction of what it costs for 5.6 Sol max. (10:42):
undefined
Josh: So really impressive outputs from this thing. (10:47):
undefined
Ejaaz: If we compare this directly with Fable, it can do most of the things that Fable (10:49):
undefined
Ejaaz: is capable of, but at a fraction of the cost, specifically 50% less cost. (10:54):
undefined
Ejaaz: Now, it can't do some things that Fable is really good at, such as extreme reasoning (10:59):
undefined
Ejaaz: or coding at the highest level. (11:05):
undefined
Ejaaz: But as a workhorse model, it is exceptionally intelligent. (11:07):
undefined
Ejaaz: And it's now my default versus Opus 4.8. But let's look at some of the examples (11:11):
undefined
Ejaaz: as to like cost wise and what this shapes up in other kinds of benchmarks. (11:16):
undefined
Ejaaz: So with Fable 5, it's kind of on average better than Fable 5 when it comes to computer use. (11:20):
undefined
Ejaaz: The price ends. So you're paying around, I think it's $5 per million tokens (11:26):
undefined
Ejaaz: input and $25 per million tokens output. So $30 overall, if you want to kind (11:30):
undefined
Ejaaz: of use this for usage credits, that is exactly half the cost of Fable, (11:35):
undefined
Ejaaz: but you're getting so much more. (11:38):
undefined
Ejaaz: And then the other thing I'll say is my biggest critique for Fable, (11:40):
undefined
Ejaaz: Josh, is that whenever I'm trying to use it for pretty much most things, (11:44):
undefined
Ejaaz: it sometimes gets triggered or rather frequently gets triggered through its safety flag. (11:49):
undefined
Ejaaz: So it gets relegated to Opus 4.8. (11:53):
undefined
Ejaaz: With Opus 5, I haven't come across that at all. (11:56):
undefined
Ejaaz: When I ask it to do something, even if it's bio-related or cybersecurity-related, (12:00):
undefined
Ejaaz: like if I'm saying, hey, can you check my computer for bugs and make sure that it's all good? (12:03):
undefined
Ejaaz: It just does the thing. And that's the other thing about Opus 5. (12:07):
undefined
Ejaaz: It is almost as intelligent as Fable 5 when it comes to biology and cybersecurity-type (12:11):
undefined
Ejaaz: things, but it's really hard to exploit. (12:17):
undefined
Ejaaz: If you try to prompt inject it, it is an extremely aligned model, (12:21):
undefined
Ejaaz: which just won't do the thing. It's smart enough to understand when you're asking (12:25):
undefined
Ejaaz: an innocent question about research (12:29):
undefined
Ejaaz: or whatever that might be without triggering any safeguards. It is, (12:30):
undefined
Ejaaz: arguably the perfect version of fable five and it's going to be my go-to model going forwards. (12:34):
undefined
Josh: Yeah i love this model and there's interesting ways in which you can use it (12:39):
undefined
Josh: and it feels like it's the perfect (12:43):
undefined
Josh: completion to this family of five class models from anthropic i mean when you think about it (12:45):
undefined
Josh: anthropic released four frontier models in eight weeks we're getting now one (12:50):
undefined
Josh: every other week we have mythos five fable five sonnet five now so we have the (12:53):
undefined
Josh: this top three that's kind of sitting here this rounds out the set and this is the perfect fallback (12:58):
undefined
Josh: from fable and it feels like that, where when, (13:02):
undefined
Josh: as I've been testing it, as I've been using this, I found the most success with (13:05):
undefined
Josh: really hard problems using Fable. (13:09):
undefined
Josh: Fable will think for a very long time. And the longer I think, (13:11):
undefined
Josh: the better the outputs generally are. (13:13):
undefined
Josh: I find that Fable is very ambitious. It will go off. It will spin out its own subagents on its own. (13:16):
undefined
Josh: It will think very critically. It will develop plans and then check against (13:21):
undefined
Josh: those plans and do so in a way that I can trust. (13:24):
undefined
Josh: And I know that the longer it's running, the better my result will be. (13:27):
undefined
Josh: Opus, I will say, is a little bit different. It's a great fallback in the sense (13:30):
undefined
Josh: that if Fable gets tripped up with security or bioweapon, I'm so glad that it (13:34):
undefined
Josh: now lands on Opus 5 instead of 4.8. It's much more capable. (13:38):
undefined
Josh: But I think it's different. As you mentioned that it's a workhorse model, (13:41):
undefined
Josh: it feels like a workhorse model. It's great for shorter duration tasks. (13:44):
undefined
Josh: And I would, generally speaking, if you're using Fable, if you're using Opus, (13:48):
undefined
Josh: I would delegate it to those smaller tasks. In fact, during testing, (13:52):
undefined
Josh: one of my favorite modes to use it on is medium instead of high or extra high. (13:56):
undefined
Josh: Because what I find is the longer Opus thinks the kind of more tripped up it (14:00):
undefined
Josh: can get it feels like the little brother that was trained by fable so it knows (14:04):
undefined
Josh: to think like fable but it doesn't quite understand the reasoning behind why fable does everything (14:08):
undefined
Josh: and that unlock does make the outputs feel a little messy sometimes over really (14:14):
undefined
Josh: long duration tasks so if you're someone who's using this (14:18):
undefined
Josh: Try medium and see how medium effort performs. It's not necessarily a worse (14:21):
undefined
Josh: output. It just thinks for a little bit less longer. (14:25):
undefined
Josh: And if it has less time to kind of think and turn into these thought spirals, (14:27):
undefined
Josh: you might wind up getting slightly better outputs. That's how I've been using it. (14:31):
undefined
Josh: For the people who are wondering how to use this, I believe the free plan has (14:34):
undefined
Josh: access to it, but also the $20 a month plan gives you quite a bit of access (14:37):
undefined
Josh: to Opus 5, which seems very worthwhile for the amount of output tokens that (14:41):
undefined
Josh: you can get from this thing. (14:45):
undefined
Josh: And if you did want fable that does require i think the hundred dollar max plan (14:46):
undefined
Josh: but also i know they're giving out a hundred dollars in usage tokens to test (14:49):
undefined
Josh: this out to a lot of accounts so there is the option for everyone to kind of get involved (14:53):
undefined
Josh: and try this out but for twenty dollars a month man this is a really compelling (14:56):
undefined
Josh: model to do some like serious damage on any workloads that you might have (15:00):
undefined
Ejaaz: It's my new go-to model and if if you're wondering hmm should i switch to gpt 5.6 sol (15:04):
undefined
Ejaaz: which model should be my everyday model you should basically use this model (15:10):
undefined
Ejaaz: that that would be my kind of like bit of advice here because it's really good (15:14):
undefined
Ejaaz: at just general reasoning and answering most general questions that people have. (15:18):
undefined
Ejaaz: Now, what it's not exceptional as, I think it's worth pointing out, (15:21):
undefined
Ejaaz: is if you want the best-in-class model for coding or to reference different (15:25):
undefined
Ejaaz: types of health or, in some cases, some kind of cybersecurity e-bals, (15:30):
undefined
Ejaaz: you are still better off using. (15:35):
undefined
Ejaaz: Fable or if you are one of the classified partners from Anthropic, Mythos 5. (15:37):
undefined
Ejaaz: But as a general model, it is a fantastic thing to get access to. (15:41):
undefined
Ejaaz: The other thing I'll say is, I think this is the first ever model launch from (15:45):
undefined
Ejaaz: Anthropoc or any Frontier Lab, where they didn't advertise. (15:50):
undefined
Ejaaz: How intelligent it is. You mentioned ARC AGI 3 earlier, Josh. (15:54):
undefined
Ejaaz: Now, if you've heard of the AGI index or the test, there was AGI 1 and AGI 2. (15:58):
undefined
Ejaaz: But the issue was all these models kept getting smarter and just beating these tests to 100%. (16:03):
undefined
Ejaaz: They had to start creating more difficult, hard-to-solve benchmarks. (16:08):
undefined
Ejaaz: And ARC AGI 3 was the most recent one. And as you mentioned, (16:12):
undefined
Ejaaz: the most recent model, GPT 5.6 SOL, scored 7%. This one scored 30%, (16:15):
undefined
Ejaaz: the highest of any model higher than Fable, but this wasn't the headline advertisement from Anthropic. (16:19):
undefined
Ejaaz: And I think a reason behind that might potentially be, you know, (16:26):
undefined
Ejaaz: they don't really want to advertise how intelligent these models are because (16:29):
undefined
Ejaaz: of fear of regulation or whatever that might be. (16:32):
undefined
Ejaaz: But the point is, if you don't have access, if you don't have one of those memberships (16:34):
undefined
Ejaaz: that Josh just mentioned to get access to Fable 5, fear not, (16:37):
undefined
Ejaaz: because Opus 5 itself is actually a fantastic model in itself. (16:41):
undefined
Ejaaz: And I encourage you to go out and try this. Now, the other part that I will (16:44):
undefined
Ejaaz: say is really powerful, what makes this model like really valuable to use is (16:48):
undefined
Ejaaz: the fact that it can natively spin out multiple AI agents to go and do your task. (16:52):
undefined
Ejaaz: And this cannot be understated because usually when people interact with an (16:58):
undefined
Ejaaz: AI model, they just have a one-on-one conversation with one instance of Claude. (17:02):
undefined
Ejaaz: But imagine if you could have a one-on-one conversation with Claude and it spins (17:05):
undefined
Ejaaz: out a hundred different employee workers to go and do your job and it works (17:10):
undefined
Ejaaz: in parallel simultaneously. AI agents is without doubt the future of how you execute work. (17:13):
undefined
Ejaaz: And it's incredibly important to focus on their security. (17:19):
undefined
Josh: And speaking of agents and just all the sub-agents that Opus can spin up, (17:21):
undefined
Josh: we have a sponsor that is here to talk about agents and they are Ledger. (17:25):
undefined
Josh: If you are building with AI agents, you're worried about security. (17:29):
undefined
Josh: An agent with unchecked access is a brilliant but probabilistic teammate with (17:31):
undefined
Josh: access to important secrets. (17:36):
undefined
Josh: Ledger Agent Stack fixes that with four open source tools, DMK Skills and CLIs (17:37):
undefined
Josh: for Ledger Wallet, Ledger Enterprise and Ledger Enterprise Multisig that lets (17:41):
undefined
Josh: your agents work with crypto while you stay in control. You can tell your agent, rebalance my wallet. (17:45):
undefined
Josh: ETH to 40%, fastest swaps. It does the math, drafts the swap, and stops. (17:50):
undefined
Josh: The transaction lands on your ledger signer and nothing moves until you approve it. (17:55):
undefined
Josh: Agents Propose, Humans Approve, Ledger Signers Enforce. And this works with (17:58):
undefined
Josh: Cloud Code, Codex, Cursor, it's open source, ready today on developers.ledger.com, (18:03):
undefined
Josh: which you can find in the link in our description below. Thank you so much to (18:08):
undefined
Josh: Ledger for sponsoring this episode. (18:11):
undefined
Josh: And now, speaking of all these safeguards, these approvals, we have to talk (18:13):
undefined
Josh: about where Opus 5 sits on the security spectrum. EJS, I know for a fact that (18:16):
undefined
Josh: if I ask it, how to build me a nuclear warhead, it's not going to answer. (18:21):
undefined
Josh: But there's a line that is closer to there than Fable 5 is. So where is that (18:24):
undefined
Josh: line in terms of what you can use this model for? (18:29):
undefined
Ejaaz: Well, rather than give you the line first, I want to give you a stat, (18:31):
undefined
Ejaaz: which should relieve a lot of people listening to this that have been using Fable. (18:36):
undefined
Ejaaz: 85% of the time, you have fewer stops, so fewer relegations to older models. (18:40):
undefined
Ejaaz: And what that means is, if you ask it a question, let's say it's like, (18:47):
undefined
Ejaaz: you know, can you tell me that the mitochondria is the powerhouse of the cell? (18:50):
undefined
Ejaaz: It won't flag you for trying to be an aggressor. (18:53):
undefined
Ejaaz: It instead just answers your question 85% of the time. (18:56):
undefined
Ejaaz: And this is massive because it's almost, from my experience personally, (19:01):
undefined
Ejaaz: every other attempt that I use Fable 5 for, it gets triggered unless I'm explicitly (19:04):
undefined
Ejaaz: not talking about any kind of AI models. And when you're doing a show where (19:09):
undefined
Ejaaz: you talk about AI models, it's kind of a hard thing to avoid. (19:12):
undefined
Ejaaz: So the level up that's happened with Opus 5 here is they saw the feedback with Fable 5. (19:15):
undefined
Ejaaz: They saw that the safety guards were getting triggered. And for context, (19:20):
undefined
Ejaaz: for anyone that's listening, if you're wondering why these safeguards are in (19:23):
undefined
Ejaaz: place, it's because the US government got involved with Anthropic's release (19:25):
undefined
Ejaaz: of Fable 5 and said, hey, like this thing is too powerful. You need to add more (19:29):
undefined
Ejaaz: safeguards to it. You can't release your Mythos model. (19:32):
undefined
Ejaaz: We need to have a safe version so that people don't exploit it and use it for malicious purposes. (19:35):
undefined
Ejaaz: So Anthropic, pause, put out this Fable 5 model with a lot of safety guards, (19:39):
undefined
Ejaaz: and they've been iterating ever since. Opus 5 is the embellishment of what a (19:43):
undefined
Ejaaz: good version of Mythos 5 looks like to me personally. (19:47):
undefined
Ejaaz: And it's because when it comes to the biggest risk indicators, (19:51):
undefined
Ejaaz: which is when anyone's asking about cybersecurity, they could potentially want (19:55):
undefined
Ejaaz: to exploit another app or platform using that. (19:59):
undefined
Ejaaz: Or if they're evaluating biological types of things and they don't want to make (20:02):
undefined
Ejaaz: sure they create a bioweapon, they can succinctly have a conversation with Opus (20:06):
undefined
Ejaaz: 5 and you can't prompt inject Opus 5 to give you any kind of a malicious answer. (20:11):
undefined
Ejaaz: And this is the main thing. when I was looking at the system card, (20:16):
undefined
Ejaaz: the way, the blueprint of how they created this model, there is a heavy index, (20:18):
undefined
Ejaaz: a heavy framework around Opus 5, which prevents you from trying to get a malicious answer from it. (20:23):
undefined
Ejaaz: Now, can you use it to look at your computing operating system and figure out (20:29):
undefined
Ejaaz: if there are any bugs to fix, (20:33):
undefined
Ejaaz: Yes. But can you ask the Opus model to then create a potential defense to that exploit? (20:35):
undefined
Ejaaz: The answer is no. You need to kind of like go through your own way to build (20:42):
undefined
Ejaaz: it yourself, either by hand or to use it as an assistant. (20:45):
undefined
Ejaaz: But you can't directly ask it to create a fix for it. And that, (20:47):
undefined
Ejaaz: I think, is the most important nuance between the cybersecurity evals, (20:50):
undefined
Ejaaz: the biosecurity evals between Opus 5 and Fable 5. (20:54):
undefined
Josh: I mean, to frame this pretty practically, I've been using this for a while and (20:57):
undefined
Josh: I've been testing all the edge cases and you basically don't notice it. (21:00):
undefined
Josh: I'd say this is about the same safeguard level as standard models in the sense (21:03):
undefined
Josh: that you're not going to run into it unless you are intentionally trying to run into it (21:06):
undefined
Josh: you really got to ask some kind of edgy fringe questions in order to hit those (21:10):
undefined
Josh: safeguards so i would say in terms of what you're actually going to realize (21:13):
undefined
Josh: you are going to get blocked significantly less than you would with fable to (21:18):
undefined
Josh: the point where you don't really notice it (21:20):
undefined
Josh: and i think this is kind of where i've landed as a more power user who is (21:23):
undefined
Josh: i just kind of token hungry just for the fun of it and i love using fable as (21:27):
undefined
Josh: the default but i really appreciate the fact that I can have Opus 5 as that (21:30):
undefined
Josh: fallback that can confidently answer questions that Fable cannot. (21:34):
undefined
Josh: For example, I was trying to figure out stuff about there's this like new breakthrough (21:38):
undefined
Josh: with like genomic sequencing and it'll like diverse the aging of your skin. (21:42):
undefined
Josh: I was like, oh, this is sick. Let me read about it. (21:45):
undefined
Josh: Fable was on the case. It couldn't figure it out. Opus took over. (21:47):
undefined
Josh: It can handle that stuff fine. So in terms of safeguards, this will feel just (21:50):
undefined
Josh: like an unfiltered model. This will give you the outputs that you want. (21:53):
undefined
Josh: It's a lot less kind of on a hair trigger when it comes to these safeguards. (21:56):
undefined
Josh: And that makes it a really fun model to use. So it's amazing. (22:01):
undefined
Josh: It is accessible to everyone. (22:04):
undefined
Josh: I will say that the one thing that you could caveat it with is the long duration thinking. (22:05):
undefined
Josh: Sometimes it gets a little bit confused. So you may want to keep it on a slightly (22:10):
undefined
Josh: lower setting for a lot of the work that you're doing. (22:13):
undefined
Josh: But outside of that, it's freaking awesome. I mean, this is a great upgrade. (22:15):
undefined
Josh: And now I feel good. Like when I opened up my Cloud app, I got Sonnet, (22:18):
undefined
Josh: which I'm not sure I really use Sonnet because I'm not burning through like (22:21):
undefined
Josh: a trillion tokens where I need it. (22:25):
undefined
Josh: But opus 5 and fable 5 are this unbelievable tag team that just work together (22:27):
undefined
Josh: so well there is also one one other nuance um fable speak which is like this (22:31):
undefined
Josh: kind of funny thing where (22:35):
undefined
Josh: it has a very unique way of talking it's very verbose that has been transferred (22:36):
undefined
Josh: to opus 5 as well i've noticed so opus 5 likes to talk a lot it likes to speak (22:40):
undefined
Josh: the way that fable does so that's an interesting little quirk um but outside (22:45):
undefined
Josh: of that i've been using the model i love the model opus 5 is great (22:48):
undefined
Josh: i think my ask to everyone who's listening is like hey are you using it what (22:51):
undefined
Josh: are you using it for best use cases, anything like that. Like this has been cool. (22:54):
undefined
Ejaaz: If you are on any kind of a membership on Claude or Anthropic specifically, this is the model to use. (22:58):
undefined
Ejaaz: Now I will say that although it is 50% cheaper than Fable, this thing thinks (23:04):
undefined
Ejaaz: quite a lot. It is a token guzzler. It uses a lot of tokens versus Fable 5. (23:09):
undefined
Ejaaz: But you have a higher allowance when you're using this particular model. (23:15):
undefined
Ejaaz: So if you're using Fable 5 and you don't want to pay access credits, (23:18):
undefined
Ejaaz: you should definitely be using this model. (23:22):
undefined
Ejaaz: The other thing I'll say is, I think this is very much Anthropik's response (23:24):
undefined
Ejaaz: to some of the Chinese open source model debate that's been happening over the last week or so. (23:28):
undefined
Ejaaz: Kimi K3 was launched by Moonshot Labs. It's a fantastic model, (23:34):
undefined
Ejaaz: almost frontier capability, but at a fraction of the cost. (23:37):
undefined
Ejaaz: And actually, as we're recording this episode, their model weight should become (23:40):
undefined
Ejaaz: open to day which means that you can kind of run it on hardware at home right (23:44):
undefined
Ejaaz: now they're good and bad things about this good is. (23:48):
undefined
Ejaaz: You know you can run it privately on your data it's arguably cheaper the bad (23:52):
undefined
Ejaaz: side of the thing is it has no safeguards around it which means that you could (23:55):
undefined
Ejaaz: potentially use it for cyber security threats or bio threats (23:59):
undefined
Ejaaz: moving forwards now the great thing about a model like opus 5 is it has those (24:02):
undefined
Ejaaz: safeguards in place so you know that anyone that's using it can't really maliciously (24:07):
undefined
Ejaaz: use it for any kind of mal intent and i think that there are pros and cons at both sides of things. (24:10):
undefined
Ejaaz: I know that a large number of companies, prominent companies like NVIDIA, (24:15):
undefined
Ejaaz: Microsoft signed an open source waiver this weekend. Anthropic was the one that (24:18):
undefined
Ejaaz: hadn't signed it, but they have their reasons for this. (24:23):
undefined
Ejaaz: And I'm excited for people to get their hands on Opus 5 and really tell us, (24:25):
undefined
Ejaaz: you know, what's your experience about this? Are you enjoying it? (24:28):
undefined
Ejaaz: Are you using it for any particular use cases that we haven't mentioned on this show? (24:31):
undefined
Ejaaz: It's a fantastic model. So get your hands on it and let us know in the comments (24:35):
undefined
Ejaaz: what your feedback is. But I think that is the end of the episode, (24:38):
undefined
Ejaaz: the recap of Opus 5 today. I encourage everyone to use it. If you're listening (24:41):
undefined
Ejaaz: to us or watching this on YouTube and you aren't subscribed to us, what are you doing? (24:45):
undefined
Ejaaz: Leave us a comment, give us a thumbs up. It helps us out massively. (24:48):
undefined
Ejaaz: If you're listening to this on Spotify or Apple Music, we are there. (24:51):
undefined
Ejaaz: Leave us a comment. We read everything, by the way, and your feedback is wonderful. (24:54):
undefined
Ejaaz: It influences what episodes we put out later on in the week and in the future. (24:57):
undefined
Ejaaz: We are also looking for sponsors. Josh and I have spun Limitless out on our (25:01):
undefined
Ejaaz: own and we are running this kind of like off our own budget. So if there's. (25:04):
undefined
Josh: Anyone out there- Thank you to Ledger for stepping up. (25:08):
undefined
Ejaaz: Yes, exactly. And thank you to Ledger for stepping up. and if there's anyone (25:10):
undefined
Ejaaz: else that would like to sponsor and advertise your product or service, (25:14):
undefined
Ejaaz: please let us know. We would love to hear from you. But I think aside from that, (25:17):
undefined
Ejaaz: That is it. (25:20):
undefined
Josh: That's everything. Thank you all so much for watching. We're coming at you live from... (25:21):
undefined
Josh: Ejaz and I are bouncing all over the place this summer. So we're just like kind (25:25):
undefined
Josh: of dropping in wherever we are in the world. (25:28):
undefined
Josh: Currently, I'm like up north of San Francisco. Ejaz is going to the West Coast. (25:29):
undefined
Josh: So we're doing this everywhere, but we're not missing the episode. (25:33):
undefined
Josh: So if anything does, you know, if the video goes in and out a little bit, (25:35):
undefined
Josh: we're doing this from remote, but we're getting the job done. (25:38):
undefined
Josh: And we appreciate all of you listening and being here with us for the entire (25:40):
undefined
Josh: journey. We will not let you down. And we appreciate all the support. (25:44):
undefined
Josh: So thank you for watching as always. We have another... (25:46):
undefined
Josh: We have a couple of really fun episodes coming up this week, (25:49):
undefined
Josh: so stay tuned for that. Let us know in the comments section all the things that (25:50):
undefined
Josh: we need to know in general, and we will see you guys in the next one. Thanks so much. (25:53):
undefined
Ejaaz: See you guys. (25:56):
undefined
undefined
Josh: It's not the smartest model. (00:03):
undefined
Josh: It's not the cheapest or the fastest model. And yet this is possibly our most (00:04):
undefined
Josh: favorite model release this entire year. (00:09):
undefined
Josh: So what gives here? What's actually happening with Claude Opus 5? (00:11):
undefined
Josh: This is the new iteration after Opus 4.8 and it sits just under Claude Fable 5. (00:15):
undefined
Josh: Now a lot of people see this, they're like, hmm, I'm not that excited about (00:19):
undefined
Josh: this model. It's just an iteration on another model that we've already had. (00:22):
undefined
Josh: I think you're wrong. And I have some demos here. Ejaz, I know you've been playing (00:26):
undefined
Josh: with this since the weekend, basically, talking to Opus 5, figuring out where (00:30):
undefined
Josh: the edge cases are, figuring out how all of this works. (00:35):
undefined
Josh: And the review so far is pretty great, right? Like Opus is a really fun model. (00:38):
undefined
Josh: And I think that's what we're going to spend this episode unpacking, (00:42):
undefined
Josh: is just how fun it is and how to get the best out of this model if you are someone (00:44):
undefined
Josh: who's new coming to Opus 5.0. (00:48):
undefined
Ejaaz: First and foremost, this model, as you mentioned, is not the cheapest, (00:50):
undefined
Ejaaz: but it's as cheap as Opus 4.8. (00:54):
undefined
Ejaaz: But it's a massively different type of model. And the first characteristic that (00:57):
undefined
Ejaaz: it's very good at is visual intelligence. (01:00):
undefined
Ejaaz: So what I mean by this is no clawed models can actually kind of see things. (01:03):
undefined
Ejaaz: It kind of like when you send it an image, it kind of breaks down the pixels (01:07):
undefined
Ejaaz: and reinvents what it's looking at from a code perspective. (01:11):
undefined
Ejaaz: Now, what I'm about to show you on my screen, Josh, was one-shotted over one and a half hours, okay? (01:14):
undefined
Ejaaz: It guzzled around 150,000 tokens. (01:20):
undefined
Ejaaz: I didn't click anything after I sent the prompt and all of this is one shot. Are you ready? (01:24):
undefined
Ejaaz: Let me tell you if you, let me ask you if you recognize what you're looking at right now. (01:30):
undefined
Josh: Oh my goodness, this looks like home. I'm seeing the Williamsburg Bridge. (01:34):
undefined
Josh: Oh my God, this is the East River. (01:37):
undefined
Josh: Do you see this, Pat Boy? Oh, wait, dude, this looks awesome. (01:40):
undefined
Ejaaz: This is a structurally accurate, of course, graphically rendered version of (01:44):
undefined
Ejaaz: the Brooklyn Bridge live with trucks at sunset. (01:50):
undefined
Ejaaz: All I said to Claude was, I want you to recreate the Brooklyn Bridge, (01:54):
undefined
Ejaaz: dimensionally accurate, pull the architectural renderings, everything on your own. (01:59):
undefined
Ejaaz: And then I want you to span out a bunch of sub agents. (02:05):
undefined
Ejaaz: And I want you to get these agents to build different components of the Brooklyn (02:09):
undefined
Ejaaz: Bridge and then review it. (02:13):
undefined
Ejaaz: So take a screenshot, review it, and only feed me back the end result once you (02:15):
undefined
Ejaaz: are satisfied with it. But it needs to look like something out of a really cool computer game. (02:20):
undefined
Ejaaz: And this is what we ended up with. Here, let me show you the sunset that wide (02:26):
undefined
Ejaaz: shot, if you want to have a look. (02:29):
undefined
Ejaaz: Let me show you a tower over here. (02:30):
undefined
Josh: Oh, that's fun. Okay, so it is pre-made shots for it as well. (02:32):
undefined
Ejaaz: Yeah, maybe I'll show you the promenade if you notice the bridge is swaying, (02:35):
undefined
Ejaaz: but maybe you want to look underneath the deck, make sure it's all structurally sound. (02:38):
undefined
Ejaaz: Now, imagine the different use cases for this. Obviously, there's simulations, (02:42):
undefined
Ejaaz: there's games, but if you're an architect that is coming up with a brand new (02:45):
undefined
Ejaaz: idea and you want to kind of put things out in the wild, you can now have a (02:49):
undefined
Ejaaz: physically and structurally accurate rendering of this model. (02:53):
undefined
Ejaaz: Now, I know a lot of folks listening to this might be thinking, (02:56):
undefined
Ejaaz: well, okay, maybe I don't want to be creating games and stuff like that. (02:58):
undefined
Ejaaz: We'll get to the other good stuff. But what I want to point out is cold Opus (03:01):
undefined
Ejaaz: can't see anything. I just want to reiterate that. (03:05):
undefined
Ejaaz: It generated every single asset, every single visual that you're seeing on screen (03:07):
undefined
Ejaaz: right now from scratch. It created a file on my computer. (03:11):
undefined
Ejaaz: It did so with one single prompt in one and a half hours, and it only cost 100,000 (03:15):
undefined
Ejaaz: tokens. It's so, so impressive. (03:19):
undefined
Josh: I think this is the thing that Opus 5 is really exceptional at that a lot of (03:21):
undefined
Josh: people are figuring out is when we went to Fable we always loved the visual demos. (03:25):
undefined
Josh: Fable was excellent but it's incredibly expensive. It takes a long time. (03:28):
undefined
Josh: It is that leading flagship model if you want the best. Opus 5 gives you (03:32):
undefined
Josh: Almost all of that, but with a fraction of the cost, a fraction of the time, (03:36):
undefined
Josh: and other really unique capabilities. So we have a few other demos, (03:40):
undefined
Josh: some that we found online that we didn't make ourselves. (03:43):
undefined
Josh: And they blew my mind with the first one being this video that we're seeing (03:45):
undefined
Josh: now. So this guy on X, he said he one-shotted this game. (03:49):
undefined
Josh: And this game is a demo made of custom code without a single external asset. (03:54):
undefined
Josh: And when you look at this game, it looks like you're watching a game of Call of Duty. (03:59):
undefined
Josh: It's really impressive. This is Call of Duty. It's what it looks like, (04:02):
undefined
Josh: right? I mean, the hands are a little funny and like there's definitely a lot of weird quirks here. (04:05):
undefined
Josh: But for a one shot prompt on a model that is not a flagship model, (04:08):
undefined
Josh: this is incredibly impressive. (04:12):
undefined
Josh: And what I really appreciated about this model and what we're going to do right (04:14):
undefined
Josh: now for the people who are watching is you could actually go and play this model (04:17):
undefined
Josh: right now in the browser. (04:20):
undefined
Josh: So please, if you would, I have the browser and let's see what your skills are (04:21):
undefined
Josh: like. Now, for people who don't know, I've spent my entire life playing Call (04:26):
undefined
Josh: of Duty, very familiar with the game. (04:29):
undefined
Josh: And I'm here to criticize Ejaz on how well you are able to play this one-shot (04:31):
undefined
Josh: demo from Opus 5. And we have the game loaded up here, Ejaz. Let's see what you got. (04:36):
undefined
Josh: It's funny, this looks like a map from Call of Duty Modern Warfare. (04:40):
undefined
Josh: I could see you're getting kills, you know, reasonably well. (04:43):
undefined
Ejaaz: Oh my God, I'm getting cooked. (04:45):
undefined
Josh: Honestly, of all the players, if I had to team up with, I'm not sure you're (04:47):
undefined
Josh: going to be the one for Call of Duty. (04:51):
undefined
Josh: I got to be honest with you, I'm not super impressed with the movement or the shots. (04:53):
undefined
Josh: Fair enough. But this is like a really impressive demo. I mean, (04:56):
undefined
Josh: the fact that this was created with one shot on opus 5 this is a game that costs (04:58):
undefined
Josh: studios billions of dollars to develop and you have a lot of the textures like (05:03):
undefined
Josh: the textures of the map look good (05:06):
undefined
Josh: the design of the weapon looks good the hands are a little funky but it has (05:08):
undefined
Josh: the skybox it has all these things (05:12):
undefined
Josh: that you would look for if we had audio turned on you can actually hear (05:13):
undefined
Josh: the sound of footsteps you can hear the hit boxes when someone gets hit by a (05:16):
undefined
Josh: bullet you can hear the weapon firing it's really robust and it clearly has (05:19):
undefined
Josh: this very deep understanding of what these games look like and that's what i (05:23):
undefined
Josh: found and super interesting with this particular demo. (05:26):
undefined
Josh: We have a couple other demos that I wanted to show before we got too deep into this. (05:28):
undefined
Josh: One of which being a snowboarder game. So one of my favorite video games when (05:32):
undefined
Josh: I was younger, SSX Tricky. (05:36):
undefined
Josh: I'm not sure you've ever played that, Ejaz. It is a snowboarding game where (05:37):
undefined
Josh: you go down a slope and you do tricks and you get points for doing these crazy tricks. (05:41):
undefined
Josh: There is another one just like that. It doesn't have the tricks per se, (05:46):
undefined
Josh: but it is a snowboard game indistinguishable from a game like SSX Tricky. (05:49):
undefined
Josh: In fact, far more impressive in terms of visual integrity than a game like SSX (05:53):
undefined
Josh: Tricky would be. And there's a quote post that Signal wrote. (05:59):
undefined
Josh: I always love reading Signal's post on X. (06:02):
undefined
Josh: And he had some commentary about this specifically. And he says, (06:04):
undefined
Josh: if you put this game in the App Store in 2012, it would have potentially grossed (06:07):
undefined
Josh: $5 to $10 million, at least over its lifespan, maybe even more. (06:11):
undefined
Josh: Today, it can be one-shotted by roughly anyone with the correct physics, (06:15):
undefined
Josh: graphics, gameplay, etc. (06:18):
undefined
Josh: It's like $10 million of economic value wiped out in 15 years on only one game. (06:20):
undefined
Josh: And I think that's kind of true. It's very impressive that we're able to get (06:25):
undefined
Josh: things indistinguishable from leading AAA games from a single prompt on a model (06:28):
undefined
Josh: that isn't even Frontier. And that's kind of the essence of Opus 5 as we've (06:33):
undefined
Josh: been playing it. It's like the best (06:38):
undefined
Josh: bang for your buck as it relates to getting high quality tokens for what you're able to pay for them. (06:40):
undefined
Ejaaz: Before we go into what else Opus 5 is good at, and trust me, (06:45):
undefined
Ejaaz: there are many things, I just want to talk about what it takes to create something (06:49):
undefined
Ejaaz: that is as compelling visually as what you're seeing on screen as the games (06:53):
undefined
Ejaaz: that we've showed you in one shot. (06:57):
undefined
Ejaaz: Like, I am not exaggerating. I will show you the prompt that Matt Schumer used (06:58):
undefined
Ejaaz: to create that FBS game that we just played badly. (07:02):
undefined
Ejaaz: He goes, I want you to build a first person shooter at the level of the most (07:06):
undefined
Ejaaz: recent Call of Duty games, it should be utterly perfect, visually beautiful, (07:09):
undefined
Ejaaz: with every single thing done at AAA quality, from textures to physics to anything you can think of. (07:13):
undefined
Ejaaz: Then he goes on to say, and this is the key part, so if you're listening to (07:19):
undefined
Ejaaz: this and you're wondering, how can I do this for a game or for any other task (07:22):
undefined
Ejaaz: that you're trying to do using your AI model, he says the following, (07:25):
undefined
Ejaaz: fan out sub-agents and have sub-agents tackle each one individually so that (07:29):
undefined
Ejaaz: the game is utterly perfect. (07:34):
undefined
Ejaaz: You should slash loop on each item and have a separate sub-agent check it visually (07:35):
undefined
Ejaaz: to ensure it looks AAA. That separate sub-agent should be a really harsh critic. (07:41):
undefined
Ejaaz: And if it doesn't look AAA, it should keep going. (07:45):
undefined
Ejaaz: Now, let me unpack why that's so important for not just this gaming prompt, (07:48):
undefined
Ejaaz: but any prompt that you do for whatever you're trying to build using AI. (07:51):
undefined
Ejaaz: He, number one, said, spin up multiple versions of yourself, (07:55):
undefined
Ejaaz: Claude, so that you can work separately at the same time on different components of product. (07:59):
undefined
Ejaaz: Now, if you think about a traditional company, they hire a ton of employees (08:05):
undefined
Ejaaz: and they have different teams that all work together to build the same product. (08:08):
undefined
Ejaaz: In this instance, it's the same thing, except there are various versions of (08:11):
undefined
Ejaaz: Claude that tackle different tasks. (08:15):
undefined
Ejaaz: Number two, he asked it specifically to spin up a manager role, (08:17):
undefined
Ejaaz: which basically monitors these agents that are building these different things (08:21):
undefined
Ejaaz: to make sure that they do a good job. (08:24):
undefined
Ejaaz: And then number three, he said, use this loop function. The loop function is (08:25):
undefined
Ejaaz: basically, hey, if you're not happy with the work that you just created, (08:29):
undefined
Ejaaz: don't give it to me to give you feedback and say like, hey, improve this. (08:33):
undefined
Ejaaz: Just use another agent to do this. And this beautiful. (08:36):
undefined
Ejaaz: Kind of synonymous loop can be applied to anything, not just a game. (08:40):
undefined
Ejaaz: It could be applied to your Excel tasks that you're trying to do at work. (08:43):
undefined
Ejaaz: It could be applied to a blueprint that you're trying to create your architecture (08:46):
undefined
Ejaaz: from. It could be applied to anything. (08:50):
undefined
Ejaaz: And I think this is a really valuable lesson to figure out with Opus 5 specifically, (08:51):
undefined
Ejaaz: is that you can spin up multiple agents and create something really, really awesome. (08:56):
undefined
Ejaaz: But Josh, I think kind of moving on from the visual aesthetics of what Claude (09:00):
undefined
Ejaaz: Opus 5 can do, it made some pretty major breakthroughs in some other things, (09:04):
undefined
Ejaaz: including something called Arc AGI 3, which is basically the precipice of any AI model test. (09:09):
undefined
Josh: Yes, Arc AGI 3 is like the flagship test. And when it comes to this flagship (09:15):
undefined
Josh: test, Opus has done exceptionally well. (09:20):
undefined
Josh: The previous high score was 7.8, scored by GPT 5.6 Sol Max, the highest end (09:23):
undefined
Josh: GPT model that's currently available. Opus 5 got a 30%. (09:30):
undefined
Josh: I mean, that's like an unbelievably high amount of rating on this benchmark (09:33):
undefined
Josh: and we have to ask the question why what did it do and (09:38):
undefined
Josh: arc actually came out and explained themselves and they said during our analysis (09:42):
undefined
Josh: of opus 5 we observed a new capability previously unseen from frontier models (09:45):
undefined
Josh: opus 5 used advanced logical reasoning to turn arc agi 3 layouts into algebraic notation (09:50):
undefined
Josh: on action 23 it describes the seen as four center equal two axis five center (09:56):
undefined
Josh: basically what it's doing here is instead of taking the visuals you just like (10:01):
undefined
Josh: you mentioned opus can't see (10:05):
undefined
Josh: it takes the visuals converts them into math and then thinks on that math that it's (10:07):
undefined
Josh: visualizing and as a result it's able to achieve significantly better scores (10:12):
undefined
Josh: on this visual reasoning (10:16):
undefined
Josh: logical reasoning i guess is how they describe it versus other models so i find (10:17):
undefined
Josh: this really interesting if you look at the example on screen that they were (10:21):
undefined
Josh: showing if we could just go back to just the general demo it shows these shapes that are shifting (10:24):
undefined
Josh: and a lot of times models have difficulty (10:29):
undefined
Josh: Applying logic to make these work. Opus, because they converted it all to math, (10:32):
undefined
Josh: it was able to achieve such a high score. It was really impressive. (10:36):
undefined
Josh: This is what, four or five times that of the next leading model? (10:38):
undefined
Josh: And cost, I know for a fact, this cost a fraction of what it costs for 5.6 Sol max. (10:42):
undefined
Josh: So really impressive outputs from this thing. (10:47):
undefined
Ejaaz: If we compare this directly with Fable, it can do most of the things that Fable (10:49):
undefined
Ejaaz: is capable of, but at a fraction of the cost, specifically 50% less cost. (10:54):
undefined
Ejaaz: Now, it can't do some things that Fable is really good at, such as extreme reasoning (10:59):
undefined
Ejaaz: or coding at the highest level. (11:05):
undefined
Ejaaz: But as a workhorse model, it is exceptionally intelligent. (11:07):
undefined
Ejaaz: And it's now my default versus Opus 4.8. But let's look at some of the examples (11:11):
undefined
Ejaaz: as to like cost wise and what this shapes up in other kinds of benchmarks. (11:16):
undefined
Ejaaz: So with Fable 5, it's kind of on average better than Fable 5 when it comes to computer use. (11:20):
undefined
Ejaaz: The price ends. So you're paying around, I think it's $5 per million tokens (11:26):
undefined
Ejaaz: input and $25 per million tokens output. So $30 overall, if you want to kind (11:30):
undefined
Ejaaz: of use this for usage credits, that is exactly half the cost of Fable, (11:35):
undefined
Ejaaz: but you're getting so much more. (11:38):
undefined
Ejaaz: And then the other thing I'll say is my biggest critique for Fable, (11:40):
undefined
Ejaaz: Josh, is that whenever I'm trying to use it for pretty much most things, (11:44):
undefined
Ejaaz: it sometimes gets triggered or rather frequently gets triggered through its safety flag. (11:49):
undefined
Ejaaz: So it gets relegated to Opus 4.8. (11:53):
undefined
Ejaaz: With Opus 5, I haven't come across that at all. (11:56):
undefined
Ejaaz: When I ask it to do something, even if it's bio-related or cybersecurity-related, (12:00):
undefined
Ejaaz: like if I'm saying, hey, can you check my computer for bugs and make sure that it's all good? (12:03):
undefined
Ejaaz: It just does the thing. And that's the other thing about Opus 5. (12:07):
undefined
Ejaaz: It is almost as intelligent as Fable 5 when it comes to biology and cybersecurity-type (12:11):
undefined
Ejaaz: things, but it's really hard to exploit. (12:17):
undefined
Ejaaz: If you try to prompt inject it, it is an extremely aligned model, (12:21):
undefined
Ejaaz: which just won't do the thing. It's smart enough to understand when you're asking (12:25):
undefined
Ejaaz: an innocent question about research (12:29):
undefined
Ejaaz: or whatever that might be without triggering any safeguards. It is, (12:30):
undefined
Ejaaz: arguably the perfect version of fable five and it's going to be my go-to model going forwards. (12:34):
undefined
Josh: Yeah i love this model and there's interesting ways in which you can use it (12:39):
undefined
Josh: and it feels like it's the perfect (12:43):
undefined
Josh: completion to this family of five class models from anthropic i mean when you think about it (12:45):
undefined
Josh: anthropic released four frontier models in eight weeks we're getting now one (12:50):
undefined
Josh: every other week we have mythos five fable five sonnet five now so we have the (12:53):
undefined
Josh: this top three that's kind of sitting here this rounds out the set and this is the perfect fallback (12:58):
undefined
Josh: from fable and it feels like that, where when, (13:02):
undefined
Josh: as I've been testing it, as I've been using this, I found the most success with (13:05):
undefined
Josh: really hard problems using Fable. (13:09):
undefined
Josh: Fable will think for a very long time. And the longer I think, (13:11):
undefined
Josh: the better the outputs generally are. (13:13):
undefined
Josh: I find that Fable is very ambitious. It will go off. It will spin out its own subagents on its own. (13:16):
undefined
Josh: It will think very critically. It will develop plans and then check against (13:21):
undefined
Josh: those plans and do so in a way that I can trust. (13:24):
undefined
Josh: And I know that the longer it's running, the better my result will be. (13:27):
undefined
Josh: Opus, I will say, is a little bit different. It's a great fallback in the sense (13:30):
undefined
Josh: that if Fable gets tripped up with security or bioweapon, I'm so glad that it (13:34):
undefined
Josh: now lands on Opus 5 instead of 4.8. It's much more capable. (13:38):
undefined
Josh: But I think it's different. As you mentioned that it's a workhorse model, (13:41):
undefined
Josh: it feels like a workhorse model. It's great for shorter duration tasks. (13:44):
undefined
Josh: And I would, generally speaking, if you're using Fable, if you're using Opus, (13:48):
undefined
Josh: I would delegate it to those smaller tasks. In fact, during testing, (13:52):
undefined
Josh: one of my favorite modes to use it on is medium instead of high or extra high. (13:56):
undefined
Josh: Because what I find is the longer Opus thinks the kind of more tripped up it (14:00):
undefined
Josh: can get it feels like the little brother that was trained by fable so it knows (14:04):
undefined
Josh: to think like fable but it doesn't quite understand the reasoning behind why fable does everything (14:08):
undefined
Josh: and that unlock does make the outputs feel a little messy sometimes over really (14:14):
undefined
Josh: long duration tasks so if you're someone who's using this (14:18):
undefined
Josh: Try medium and see how medium effort performs. It's not necessarily a worse (14:21):
undefined
Josh: output. It just thinks for a little bit less longer. (14:25):
undefined
Josh: And if it has less time to kind of think and turn into these thought spirals, (14:27):
undefined
Josh: you might wind up getting slightly better outputs. That's how I've been using it. (14:31):
undefined
Josh: For the people who are wondering how to use this, I believe the free plan has (14:34):
undefined
Josh: access to it, but also the $20 a month plan gives you quite a bit of access (14:37):
undefined
Josh: to Opus 5, which seems very worthwhile for the amount of output tokens that (14:41):
undefined
Josh: you can get from this thing. (14:45):
undefined
Josh: And if you did want fable that does require i think the hundred dollar max plan (14:46):
undefined
Josh: but also i know they're giving out a hundred dollars in usage tokens to test (14:49):
undefined
Josh: this out to a lot of accounts so there is the option for everyone to kind of get involved (14:53):
undefined
Josh: and try this out but for twenty dollars a month man this is a really compelling (14:56):
undefined
Josh: model to do some like serious damage on any workloads that you might have (15:00):
undefined
Ejaaz: It's my new go-to model and if if you're wondering hmm should i switch to gpt 5.6 sol (15:04):
undefined
Ejaaz: which model should be my everyday model you should basically use this model (15:10):
undefined
Ejaaz: that that would be my kind of like bit of advice here because it's really good (15:14):
undefined
Ejaaz: at just general reasoning and answering most general questions that people have. (15:18):
undefined
Ejaaz: Now, what it's not exceptional as, I think it's worth pointing out, (15:21):
undefined
Ejaaz: is if you want the best-in-class model for coding or to reference different (15:25):
undefined
Ejaaz: types of health or, in some cases, some kind of cybersecurity e-bals, (15:30):
undefined
Ejaaz: you are still better off using. (15:35):
undefined
Ejaaz: Fable or if you are one of the classified partners from Anthropic, Mythos 5. (15:37):
undefined
Ejaaz: But as a general model, it is a fantastic thing to get access to. (15:41):
undefined
Ejaaz: The other thing I'll say is, I think this is the first ever model launch from (15:45):
undefined
Ejaaz: Anthropoc or any Frontier Lab, where they didn't advertise. (15:50):
undefined
Ejaaz: How intelligent it is. You mentioned ARC AGI 3 earlier, Josh. (15:54):
undefined
Ejaaz: Now, if you've heard of the AGI index or the test, there was AGI 1 and AGI 2. (15:58):
undefined
Ejaaz: But the issue was all these models kept getting smarter and just beating these tests to 100%. (16:03):
undefined
Ejaaz: They had to start creating more difficult, hard-to-solve benchmarks. (16:08):
undefined
Ejaaz: And ARC AGI 3 was the most recent one. And as you mentioned, (16:12):
undefined
Ejaaz: the most recent model, GPT 5.6 SOL, scored 7%. This one scored 30%, (16:15):
undefined
Ejaaz: the highest of any model higher than Fable, but this wasn't the headline advertisement from Anthropic. (16:19):
undefined
Ejaaz: And I think a reason behind that might potentially be, you know, (16:26):
undefined
Ejaaz: they don't really want to advertise how intelligent these models are because (16:29):
undefined
Ejaaz: of fear of regulation or whatever that might be. (16:32):
undefined
Ejaaz: But the point is, if you don't have access, if you don't have one of those memberships (16:34):
undefined
Ejaaz: that Josh just mentioned to get access to Fable 5, fear not, (16:37):
undefined
Ejaaz: because Opus 5 itself is actually a fantastic model in itself. (16:41):
undefined
Ejaaz: And I encourage you to go out and try this. Now, the other part that I will (16:44):
undefined
Ejaaz: say is really powerful, what makes this model like really valuable to use is (16:48):
undefined
Ejaaz: the fact that it can natively spin out multiple AI agents to go and do your task. (16:52):
undefined
Ejaaz: And this cannot be understated because usually when people interact with an (16:58):
undefined
Ejaaz: AI model, they just have a one-on-one conversation with one instance of Claude. (17:02):
undefined
Ejaaz: But imagine if you could have a one-on-one conversation with Claude and it spins (17:05):
undefined
Ejaaz: out a hundred different employee workers to go and do your job and it works (17:10):
undefined
Ejaaz: in parallel simultaneously. AI agents is without doubt the future of how you execute work. (17:13):
undefined
Ejaaz: And it's incredibly important to focus on their security. (17:19):
undefined
Josh: And speaking of agents and just all the sub-agents that Opus can spin up, (17:21):
undefined
Josh: we have a sponsor that is here to talk about agents and they are Ledger. (17:25):
undefined
Josh: If you are building with AI agents, you're worried about security. (17:29):
undefined
Josh: An agent with unchecked access is a brilliant but probabilistic teammate with (17:31):
undefined
Josh: access to important secrets. (17:36):
undefined
Josh: Ledger Agent Stack fixes that with four open source tools, DMK Skills and CLIs (17:37):
undefined
Josh: for Ledger Wallet, Ledger Enterprise and Ledger Enterprise Multisig that lets (17:41):
undefined
Josh: your agents work with crypto while you stay in control. You can tell your agent, rebalance my wallet. (17:45):
undefined
Josh: ETH to 40%, fastest swaps. It does the math, drafts the swap, and stops. (17:50):
undefined
Josh: The transaction lands on your ledger signer and nothing moves until you approve it. (17:55):
undefined
Josh: Agents Propose, Humans Approve, Ledger Signers Enforce. And this works with (17:58):
undefined
Josh: Cloud Code, Codex, Cursor, it's open source, ready today on developers.ledger.com, (18:03):
undefined
Josh: which you can find in the link in our description below. Thank you so much to (18:08):
undefined
Josh: Ledger for sponsoring this episode. (18:11):
undefined
Josh: And now, speaking of all these safeguards, these approvals, we have to talk (18:13):
undefined
Josh: about where Opus 5 sits on the security spectrum. EJS, I know for a fact that (18:16):
undefined
Josh: if I ask it, how to build me a nuclear warhead, it's not going to answer. (18:21):
undefined
Josh: But there's a line that is closer to there than Fable 5 is. So where is that (18:24):
undefined
Josh: line in terms of what you can use this model for? (18:29):
undefined
Ejaaz: Well, rather than give you the line first, I want to give you a stat, (18:31):
undefined
Ejaaz: which should relieve a lot of people listening to this that have been using Fable. (18:36):
undefined
Ejaaz: 85% of the time, you have fewer stops, so fewer relegations to older models. (18:40):
undefined
Ejaaz: And what that means is, if you ask it a question, let's say it's like, (18:47):
undefined
Ejaaz: you know, can you tell me that the mitochondria is the powerhouse of the cell? (18:50):
undefined
Ejaaz: It won't flag you for trying to be an aggressor. (18:53):
undefined
Ejaaz: It instead just answers your question 85% of the time. (18:56):
undefined
Ejaaz: And this is massive because it's almost, from my experience personally, (19:01):
undefined
Ejaaz: every other attempt that I use Fable 5 for, it gets triggered unless I'm explicitly (19:04):
undefined
Ejaaz: not talking about any kind of AI models. And when you're doing a show where (19:09):
undefined
Ejaaz: you talk about AI models, it's kind of a hard thing to avoid. (19:12):
undefined
Ejaaz: So the level up that's happened with Opus 5 here is they saw the feedback with Fable 5. (19:15):
undefined
Ejaaz: They saw that the safety guards were getting triggered. And for context, (19:20):
undefined
Ejaaz: for anyone that's listening, if you're wondering why these safeguards are in (19:23):
undefined
Ejaaz: place, it's because the US government got involved with Anthropic's release (19:25):
undefined
Ejaaz: of Fable 5 and said, hey, like this thing is too powerful. You need to add more (19:29):
undefined
Ejaaz: safeguards to it. You can't release your Mythos model. (19:32):
undefined
Ejaaz: We need to have a safe version so that people don't exploit it and use it for malicious purposes. (19:35):
undefined
Ejaaz: So Anthropic, pause, put out this Fable 5 model with a lot of safety guards, (19:39):
undefined
Ejaaz: and they've been iterating ever since. Opus 5 is the embellishment of what a (19:43):
undefined
Ejaaz: good version of Mythos 5 looks like to me personally. (19:47):
undefined
Ejaaz: And it's because when it comes to the biggest risk indicators, (19:51):
undefined
Ejaaz: which is when anyone's asking about cybersecurity, they could potentially want (19:55):
undefined
Ejaaz: to exploit another app or platform using that. (19:59):
undefined
Ejaaz: Or if they're evaluating biological types of things and they don't want to make (20:02):
undefined
Ejaaz: sure they create a bioweapon, they can succinctly have a conversation with Opus (20:06):
undefined
Ejaaz: 5 and you can't prompt inject Opus 5 to give you any kind of a malicious answer. (20:11):
undefined
Ejaaz: And this is the main thing. when I was looking at the system card, (20:16):
undefined
Ejaaz: the way, the blueprint of how they created this model, there is a heavy index, (20:18):
undefined
Ejaaz: a heavy framework around Opus 5, which prevents you from trying to get a malicious answer from it. (20:23):
undefined
Ejaaz: Now, can you use it to look at your computing operating system and figure out (20:29):
undefined
Ejaaz: if there are any bugs to fix, (20:33):
undefined
Ejaaz: Yes. But can you ask the Opus model to then create a potential defense to that exploit? (20:35):
undefined
Ejaaz: The answer is no. You need to kind of like go through your own way to build (20:42):
undefined
Ejaaz: it yourself, either by hand or to use it as an assistant. (20:45):
undefined
Ejaaz: But you can't directly ask it to create a fix for it. And that, (20:47):
undefined
Ejaaz: I think, is the most important nuance between the cybersecurity evals, (20:50):
undefined
Ejaaz: the biosecurity evals between Opus 5 and Fable 5. (20:54):
undefined
Josh: I mean, to frame this pretty practically, I've been using this for a while and (20:57):
undefined
Josh: I've been testing all the edge cases and you basically don't notice it. (21:00):
undefined
Josh: I'd say this is about the same safeguard level as standard models in the sense (21:03):
undefined
Josh: that you're not going to run into it unless you are intentionally trying to run into it (21:06):
undefined
Josh: you really got to ask some kind of edgy fringe questions in order to hit those (21:10):
undefined
Josh: safeguards so i would say in terms of what you're actually going to realize (21:13):
undefined
Josh: you are going to get blocked significantly less than you would with fable to (21:18):
undefined
Josh: the point where you don't really notice it (21:20):
undefined
Josh: and i think this is kind of where i've landed as a more power user who is (21:23):
undefined
Josh: i just kind of token hungry just for the fun of it and i love using fable as (21:27):
undefined
Josh: the default but i really appreciate the fact that I can have Opus 5 as that (21:30):
undefined
Josh: fallback that can confidently answer questions that Fable cannot. (21:34):
undefined
Josh: For example, I was trying to figure out stuff about there's this like new breakthrough (21:38):
undefined
Josh: with like genomic sequencing and it'll like diverse the aging of your skin. (21:42):
undefined
Josh: I was like, oh, this is sick. Let me read about it. (21:45):
undefined
Josh: Fable was on the case. It couldn't figure it out. Opus took over. (21:47):
undefined
Josh: It can handle that stuff fine. So in terms of safeguards, this will feel just (21:50):
undefined
Josh: like an unfiltered model. This will give you the outputs that you want. (21:53):
undefined
Josh: It's a lot less kind of on a hair trigger when it comes to these safeguards. (21:56):
undefined
Josh: And that makes it a really fun model to use. So it's amazing. (22:01):
undefined
Josh: It is accessible to everyone. (22:04):
undefined
Josh: I will say that the one thing that you could caveat it with is the long duration thinking. (22:05):
undefined
Josh: Sometimes it gets a little bit confused. So you may want to keep it on a slightly (22:10):
undefined
Josh: lower setting for a lot of the work that you're doing. (22:13):
undefined
Josh: But outside of that, it's freaking awesome. I mean, this is a great upgrade. (22:15):
undefined
Josh: And now I feel good. Like when I opened up my Cloud app, I got Sonnet, (22:18):
undefined
Josh: which I'm not sure I really use Sonnet because I'm not burning through like (22:21):
undefined
Josh: a trillion tokens where I need it. (22:25):
undefined
Josh: But opus 5 and fable 5 are this unbelievable tag team that just work together (22:27):
undefined
Josh: so well there is also one one other nuance um fable speak which is like this (22:31):
undefined
Josh: kind of funny thing where (22:35):
undefined
Josh: it has a very unique way of talking it's very verbose that has been transferred (22:36):
undefined
Josh: to opus 5 as well i've noticed so opus 5 likes to talk a lot it likes to speak (22:40):
undefined
Josh: the way that fable does so that's an interesting little quirk um but outside (22:45):
undefined
Josh: of that i've been using the model i love the model opus 5 is great (22:48):
undefined
Josh: i think my ask to everyone who's listening is like hey are you using it what (22:51):
undefined
Josh: are you using it for best use cases, anything like that. Like this has been cool. (22:54):
undefined
Ejaaz: If you are on any kind of a membership on Claude or Anthropic specifically, this is the model to use. (22:58):
undefined
Ejaaz: Now I will say that although it is 50% cheaper than Fable, this thing thinks (23:04):
undefined
Ejaaz: quite a lot. It is a token guzzler. It uses a lot of tokens versus Fable 5. (23:09):
undefined
Ejaaz: But you have a higher allowance when you're using this particular model. (23:15):
undefined
Ejaaz: So if you're using Fable 5 and you don't want to pay access credits, (23:18):
undefined
Ejaaz: you should definitely be using this model. (23:22):
undefined
Ejaaz: The other thing I'll say is, I think this is very much Anthropik's response (23:24):
undefined
Ejaaz: to some of the Chinese open source model debate that's been happening over the last week or so. (23:28):
undefined
Ejaaz: Kimi K3 was launched by Moonshot Labs. It's a fantastic model, (23:34):
undefined
Ejaaz: almost frontier capability, but at a fraction of the cost. (23:37):
undefined
Ejaaz: And actually, as we're recording this episode, their model weight should become (23:40):
undefined
Ejaaz: open to day which means that you can kind of run it on hardware at home right (23:44):
undefined
Ejaaz: now they're good and bad things about this good is. (23:48):
undefined
Ejaaz: You know you can run it privately on your data it's arguably cheaper the bad (23:52):
undefined
Ejaaz: side of the thing is it has no safeguards around it which means that you could (23:55):
undefined
Ejaaz: potentially use it for cyber security threats or bio threats (23:59):
undefined
Ejaaz: moving forwards now the great thing about a model like opus 5 is it has those (24:02):
undefined
Ejaaz: safeguards in place so you know that anyone that's using it can't really maliciously (24:07):
undefined
Ejaaz: use it for any kind of mal intent and i think that there are pros and cons at both sides of things. (24:10):
undefined
Ejaaz: I know that a large number of companies, prominent companies like NVIDIA, (24:15):
undefined
Ejaaz: Microsoft signed an open source waiver this weekend. Anthropic was the one that (24:18):
undefined
Ejaaz: hadn't signed it, but they have their reasons for this. (24:23):
undefined
Ejaaz: And I'm excited for people to get their hands on Opus 5 and really tell us, (24:25):
undefined
Ejaaz: you know, what's your experience about this? Are you enjoying it? (24:28):
undefined
Ejaaz: Are you using it for any particular use cases that we haven't mentioned on this show? (24:31):
undefined
Ejaaz: It's a fantastic model. So get your hands on it and let us know in the comments (24:35):
undefined
Ejaaz: what your feedback is. But I think that is the end of the episode, (24:38):
undefined
Ejaaz: the recap of Opus 5 today. I encourage everyone to use it. If you're listening (24:41):
undefined
Ejaaz: to us or watching this on YouTube and you aren't subscribed to us, what are you doing? (24:45):
undefined
Ejaaz: Leave us a comment, give us a thumbs up. It helps us out massively. (24:48):
undefined
Ejaaz: If you're listening to this on Spotify or Apple Music, we are there. (24:51):
undefined
Ejaaz: Leave us a comment. We read everything, by the way, and your feedback is wonderful. (24:54):
undefined
Ejaaz: It influences what episodes we put out later on in the week and in the future. (24:57):
undefined
Ejaaz: We are also looking for sponsors. Josh and I have spun Limitless out on our (25:01):
undefined
Ejaaz: own and we are running this kind of like off our own budget. So if there's. (25:04):
undefined
Josh: Anyone out there- Thank you to Ledger for stepping up. (25:08):
undefined
Ejaaz: Yes, exactly. And thank you to Ledger for stepping up. and if there's anyone (25:10):
undefined
Ejaaz: else that would like to sponsor and advertise your product or service, (25:14):
undefined
Ejaaz: please let us know. We would love to hear from you. But I think aside from that, (25:17):
undefined
Ejaaz: That is it. (25:20):
undefined
Josh: That's everything. Thank you all so much for watching. We're coming at you live from... (25:21):
undefined
Josh: Ejaz and I are bouncing all over the place this summer. So we're just like kind (25:25):
undefined
Josh: of dropping in wherever we are in the world. (25:28):
undefined
Josh: Currently, I'm like up north of San Francisco. Ejaz is going to the West Coast. (25:29):
undefined
Josh: So we're doing this everywhere, but we're not missing the episode. (25:33):
undefined
Josh: So if anything does, you know, if the video goes in and out a little bit, (25:35):
undefined
Josh: we're doing this from remote, but we're getting the job done. (25:38):
undefined
Josh: And we appreciate all of you listening and being here with us for the entire (25:40):
undefined
Josh: journey. We will not let you down. And we appreciate all the support. (25:44):
undefined
Josh: So thank you for watching as always. We have another... (25:46):
undefined
Josh: We have a couple of really fun episodes coming up this week, (25:49):
undefined
Josh: so stay tuned for that. Let us know in the comments section all the things that (25:50):
undefined
Josh: we need to know in general, and we will see you guys in the next one. Thanks so much. (25:53):
undefined
Ejaaz: See you guys. (25:56):
undefined
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Bobby Bones Show
Listen to 'The Bobby Bones Show' by downloading the daily full replay.