Everyday AI Podcast – An AI and ChatGPT Podcast - Ep 809: OpenAI’s New GPT-5.6 Release: What's New, Why the Feds Are Blocking it and What's Next

Starting point is 00:00:00 This is the Everyday AI Show, the Everyday Podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Normally when OpenAI releases a new model inside of ChadjBT, it has the potential to instantly change how millions of people work. Think new capabilities, support for new inputs, and new artifacts as outputs. You can see why one billion people getting access to something as straightforward as a model update could actually be a huge deal. But with OpenAI's latest model update, GPT 5-6, we didn't really get any of that. That's because for the first time, we got a GPT upgrade that was announced but not released. That's because ever since Anthropic escalated its feud with the federal government, Frontier Labs in the U.S.

Starting point is 00:01:01 have been under increased scrutiny to ensure its models are safe. So on Friday, when Open AI. unveiled GPD 5.6, all but like 99.9% of the world just got a blog post into TBD. Yet even though only a select few trusted companies in the world can use GPD5.6 right now, that doesn't mean you should sit idly by. Actually, I'd argue the opposite, because with this new government, government AI permission slip era that we're all living in now, the smartest thing you can do is actually understand the new rules before everyone else does. Because the leaders who figure out how to operate in this world right now are the ones who will have a massive head start when

Starting point is 00:01:46 the doors finally open for everyone. So today, I'm going to show you exactly what the new GPT-5-6 is, why the feds wanted a look first, and what it all means for you and your business. So here is the big picture. Well, there's a powerful new model, but it's kind of locked up at launch. So opening I announced GPD 5.6 on Friday, but they blocked normal chat GPD access to it. So normally when opening I releases a new model, well, everyone just goes to chat gbt.com and they start using it and making changes to the workflow. Not the case now. It's a limited release. Only a couple dozen companies in the world reportedly have access to it now. Now, there's also a new naming system, which we're going to get to.

Starting point is 00:02:31 So it is called GP56, Seoul, Tara, and Luna, and it packages frontier capabilities into what Open AI is now calling durable product tiers. So opening I previewed the model to the U.S. government first. We're going to get into that, then limited the launch at the government's request. And this whole situation right now mirrors the harsher federal move against anthropic weeks earlier as AI now becomes controlled. infrastructure. So on today's show, stick with me for the next 20-ish minutes. Here's what you're going to learn. You're going to know what Open AI actually built and why even its cheapest model now rates as dangerous in the new 5-6 series. You're going to know the buried admission from Open AI's own tests that show that this model kind of cheated and overstepped its own instructions. You're going to know why the federal government gated access and how this echoes the in traffic shutdown weeks earlier. And I'm going to give you the operator playbook for, or when the best model is, well, one you can't actually use and what you should do about it in that case. All right. I'm excited for today's show. Hope you are too. This is Everyday AI. Welcome. My name's Jordan and we do this every day. That's the name Everyday AI. So this is your daily

Starting point is 00:03:46 live stream, podcast, and free daily newsletter helping everyday business leaders like you and me not just keep up with the AI developments, but I tell you what it actually means in plain technical, non-technical terms. I give you the non-b-S version. You take that information to grow your company and your career. So it starts here with the unedited, unscripted live stream podcast. But make sure you go to our website at your everyday AI.com. Sign up for the free daily news that are going to be recapping today's show and everything

Starting point is 00:04:14 else that you need to know in the world of AI. There's a ton of late breaking news over the past couple of hours. So make sure you check out today's newsletter. All right. who GPD 5.6. We don't have it, right? Very few people do. And I think aside for them,

Starting point is 00:04:33 we're going to go over, you know, kind of opening eyes blog post a little bit. I'm going to tell you some of my big takeaways from looking at the model card that they released that I don't think a lot of people read or look at because not everyone's a dork like me. And I'm going to give you some of the reaction as well

Starting point is 00:04:47 from some of the very few people who did get access and actually talked about it. But here's the stats. So this is now a three-tier model family. So GPD 5.6 is now one generation split into three lasting tiers that Open AI alluded that they're going to keep naming models this. So we don't know if we're going to get a 5.7 or a, you know, GPT6 next. But it seems like they're going to keep using these naming mechanisms. So Sol, Tara, and Luna in the easiest way because, yeah, we're having to.

Starting point is 00:05:22 learn a bunch of new names now. I would just map this to opus, sonnet, and haiku. Right? So soul is the most powerful. Terra is the middle. And then Luna would be the one that's fastest and cheapest, but not the smartest. So in the same way that Anthropic has kind of had their three classes. Now we're going to have those three dedicated classes. And to make it even a little bit more confusing, it looks like there's multiple intelligence tiers within those classes. But we'll get to that later when we all actually have access to this. But soul is the main event, right? And if you're Spanish isn't that great, you know, soul, Tara, Luna, that's sun, Earth, moon. So maybe that'll make it easier to remember, right? Third rock from the sun. So soul, that is the big dog,

Starting point is 00:06:12 so to speak. So that handles the hardest reasoning and coding. Terra in the middle balances costs with performance. And then Luna can just run faster and cheaper. So soul is the flagship. and its strength set in the entire situation in motion. So here's the rollout. So right now, the companies that do have it, reportedly are just using it via the API and Codex. So OpenAI hasn't said that when and if this rolls out in a couple of weeks because they said that they plan to make GPD 56,

Starting point is 00:06:47 Sol Terra and Luna generally available in the coming weeks. They didn't say where. So right now, it's only available on the API and Codex. So we're not sure if this will be released inside of Chadshap. My assumption is they will. But I'm guessing they're using it right now with the trusted testers only via the API and codex because it's a little bit easier to have those guardrails in place. So right now, it's just a few.

Starting point is 00:07:15 So reports have said it's like 20 something companies that have this. And I'm sure there's a lot of people at those. company, so don't know exactly how many users, but at least with the one billion weekly active users that Open AI has, it's safe to say that 99.9% of us do not have access. And if we do get access in two weeks, we're still not even sure if it will be inside of Chad GPT or if it will only be in the API and in Codex. So if nothing else, if you're a power Chad GPT user and you're like, yeah, I want to get my hands on GPT 5,6, well, maybe start using Codex a little bit. So the good thing is pricing didn't change, which is great.

Starting point is 00:07:52 Anytime you see a nice jump up in capabilities and pricing stays the same, that's sweet, right? So this on the API side, right? This is just if you're paying by the token, it is still, Seoul is still $5 per million input and $30 per million output, which is the same as GPT55. And then Terra, about half that cost with 250 in per million, 15 out per million. And then Luna, very affordable at $1 per million. per million input and $6 per million output, which is interesting, right? Because we talked about, like, as an example, GPD 3.5 Flash being a much more expensive

Starting point is 00:08:34 model than previously. And I know that Google had historically been cleaning up on the Flash series. So we'll see maybe here the story that a lot of people might not be talking about. Yeah, wait and we'll check back in. But I don't know. I think maybe a sleeper here is Luna being used by developers for simple things. I think that, you know, one thing that we're going to talk about as we talked or as we alluded to on our recent start here series episode, where we talked about going from token maxing to token efficiency.

Starting point is 00:09:09 So go check out episode 789. But I think it's going to become very common in the third and fourth quarter in the enterprise first. I think everyone else is going to catch up to this in early 20. 2027, but essentially using a mixture of models. Weird. Something I said two years ago in my bold predictions. Because it just doesn't, isn't going to make sense, especially if you're paying per token on the API side, like so many enterprise companies are.

Starting point is 00:09:33 It doesn't make sense anymore to be paying the same, you know, price per token for something that is just going to help you rewrite an email versus something that's going to go do a 30-hour autonomous project that would take a team refactoring. a code base, you know, two weeks of human hours, right? It doesn't make sense to use the same, you know, $5 input, $30 model as it would for something that's, you know, 20% of the cost. So that's, I would just say, you know, it's not out yet. We'll see benchmarks, all that.

Starting point is 00:10:07 But I don't know, to me, with Google increasing the price of their flash model, Anthropics models not getting cheaper anytime sooner. Keep an eye on Luna, actually. So let's keep going. This is from OpenAI's blog post. I'm just going to read from their words and then give you some other people's words and my words as well. So they said, we're beginning a limited preview of the GPT 56 series,

Starting point is 00:10:32 Seoul, our flagship model, Tara, a balanced model for everyday work. In Luna, a fast and affordable model. Terra has competitive performance to GPT 5.5 while being two times cheaper. and Luna brings strong capabilities at our lowest cost. GPD 56 Seoul launches with our most robust safety stack to date. We strengthen protections for higher risk activity, sensitive cyber requests, and repeated misuse, and spent multiple weeks finding weaknesses, pressure testing our system, and hardening it against real world attacks.

Starting point is 00:11:08 All right. And then they go on to say, we believe in broad access. and we plan to make GVD-56 Seoul, Terra and Luna generally available in the coming weeks. As part of our ongoing engagement with the U.S. government, we previewed our plans and the model's capabilities ahead of today's launch. At their request, they're alluding to the U.S. government, we are starting with a limited preview for a small group of trusted partners whose participation has been shared with the government before releasing more broadly. During this preview, we will continue testing and coordinating closely with partners as we work toward broader availability.

Starting point is 00:11:47 We don't believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them. We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks. while we work with the administration to develop the cyber executive order framework and a repeatable process for future model releases. So a couple of things to point out there, right? We don't have benchmarks on all of these not models, but they did say that the middle model, Terra, it is competitive to GPD-5-5 while being twice as cheap or half the cost. So that's one call out. So, you know, GPD 5.5 is a great model.

Starting point is 00:12:39 I use still, even when we had Fable 5 access from Anthropic, for a lot of my non-coding tasks, GPD 5.5 was pro was much better. So it's going to be interesting to see. The other thing to call out in Open AI's release, obviously it is this piece of like, yes, definitely showing that they're working with the government, but in the same hand saying they definitely don't agree with this approach being. the long-term sustainability piece to it. And we're going to get to that later. And I think it's very important. So first, a couple quick benchmarks here.

Starting point is 00:13:13 So again, Open AI didn't share everything, but what they did share. And there's always a good amount of cherry picking. Open AI, Anthropic, Google. I mean, they all do it, right? They're not going to just put out their worst benchmarks. So I'm guessing it'll still be at least, I don't know, two, three, four weeks before we see how this model truly perform. against competitors.

Starting point is 00:13:36 But on some of the most important benchmarks, it beats Mythos 5, not just Mythos 5, but obviously Fable 5, which is the version of Mythos that's available generally to public consumers with guardrails. So as an example,

Starting point is 00:13:54 Terminal 2.1 bench, which is one of the more important benchmarks, I'd say. So to give you an idea, terminal bench 2.1 is a verified e-val suite used to measure in AI agents, ability to complete complex multi-step tasks in isolated CLI environments. So essentially, it's models ability to do long-running work via an agentic harness, right? So one of the more important benchmarks of 2026 in GPT-5-6 Soul Ultra,

Starting point is 00:14:32 has a 91.9 score on that. Mythos had an 88, so not necessarily super close. And then Fable 5 had an 84. So in that benchmark, very far ahead. And then Open AI, to their credit, did share some other benchmarks where they weren't in the lead. So as an example, they shared exploit bench. All right.

Starting point is 00:14:54 So exploit bench, a little bit different in terms of what it shows. But it just measures how fall. AI agents can climb from reaching vulnerable code to triggering the bug to exploiting primitive. So this is more of a model's ability to work in these cyber situations to take part in, well, cyber attacks. So this is one of those benchmarks. People were talking a lot about online. So GP5.6 Seoul, it did about the same as Mythos pre-enched. view, but for like a third of the cost.

Starting point is 00:15:36 So it didn't quite hit the full capabilities of Mythos 5, which was the release version. But the previous version, Mythos preview, GPD 56, soul is on the same path, much better than, you know, Opus 4.8. But the great thing that a lot of people are talking about is, well, hey, it's the same, roughly the same just under Mythos 5, but the same as Mythos Preview for a third of the cost. So yeah, as we talk about, you know, so many companies now are already cutting back their token spend, cutting back their AI budgets because the token maxing phase, you know, companies were spending millions of dollars unchecked. And they're like, yeah,

Starting point is 00:16:15 maybe we should, you know, start paying attention to how good these tokens are versus just burning as many as possible. So even on important benches where maybe Mythos 5, the production version, edged out this new GPD 56 sole, open AI showing, they did it at about a third of the cost. All right. So I didn't see a lot of people that actually had access to this model. And I did see a couple people tweet about it and then later deleted their tweets. But one that still loved. So I feel fine sharing it from a Shopify CTO, uh, Mikhail Parkin, who said, I had the opportunity. And this is on Twitter. I had the opportunity to deeply test both fable five and GPD 5.6 max. 5.6. 5.6.

Starting point is 00:17:01 is clearly better than Opus 4.8 at everything, slightly faster too, though that depends on the load. Then saying versus Fable, it is clearly worse. So 5.6 is worse on coding, but better on agenetic workloads. And then someone else asked on just how does it work on non-technical, non-coding tasks, to which McHale responded that he did. So saying, I did test, right? I did test it. then saying on agentic non-coding taking actions, I found 5-6 clearly better.

Starting point is 00:17:38 So if you're someone that's writing code, you know, software development, it looks like maybe 5-6 may not be as good as fable 5. But for non-technical users, again, this is the CTO of Shopify, who I would guess really knows what he's talking about, you know, saying. that on non-coding agenic tasks, stuff I use, you know, uh, models for every single day, he found five, six clearly better. And so we'll see, right, how that actually plays out across benchmarks and real world usage, but a pretty big side.

Starting point is 00:18:20 So let's talk a little bit more. What's new under the hood? And then we'll get to all the drama and unfold that story. So the new max mode is something a lot of people are talking about that. That lets Seoul reason longer, while the Ultra mode, very similar to what Claude has, runs subagents in parallel. So Open AI says Sol set a coding record scoring 88.8 on the terminal bench test. But these are right now, Open AI's own numbers.

Starting point is 00:18:48 So we'll see when we get it in LM Arena, when we get it in artificial analysis, et cetera. But here's where it started to get not concerning, but maybe. concerning on paper. That's because GPT 56, like a lot of more models of the last year or so that start showing emergent capabilities, well, it acted kind of beyond its order. So opening eye in its model card talked about this, how its own safety tests found sole, more willing to act beyond what users actually asked. So in logged cases, it deleted the wrong files, faked verified results, and copied credentials unprompted. So this is a very important. So this is

Starting point is 00:19:30 during their initial testing and then they go through and they try to patch all these things up. But, you know, Open AI does now urge human supervision and outside evaluators found that same pattern. And a lot of this, you know, going through the government lens, well, it's because of the cyber and bio ratings. So open AI rated Seoul, Terra, and even Luna high for cybersecurity capabilities, which is the first time that their models have hit that mark in mass. So opening I also rated all three high for biological chemical capabilities as well. So that rating means that even the cheapest tiers are going to change risk planning. Right.

Starting point is 00:20:15 And you might think, okay, like what does that matter? Well, a lot of the hoopla over the mythos and fable back and forth came to cyber security and exploits. Right. So when even a change. cheaper model like Luna has a higher capability for bio and chemical wrongdoing, more or less, right? Because here's the reality. When you release this thing to the masses, you know, I think most people are maybe listening to this or reading about it are probably thinking of, well,

Starting point is 00:20:49 using the model for good, right? Using the model for your work, right? But good chunk of the population uses it for other purposes, right? They use it for bad reasons. And that's why some of these, you know, warnings and hitting these thresholds on, on bio and chem, you know, really are then forcing the U.S. government to have a greater involvement, whereas they didn't have that involvement before. But the reality is this. No one has access. Hardly no one. It didn't even drop in ChadGBT. Individual users can't get access to it.

Starting point is 00:21:27 And according to Axios, it was only about 20 organizations. that originally had access. So here's why. Let's get to the, the details on why there's this, apparently extremely capable new model that the few people that have used both this and Anthropics best model that is now off the shelves,

Starting point is 00:21:48 but could be back on the shelves in Mythos and Fable, right? A lot of people are saying and benchmarks are showing, it's going to be, it's going to be comparable, you know, I do ultimately think, we'll see, this is a point model update. Right. So what that means is they went from 5.5 to 5.6. So presumably this is on the same pre-training run as 5.5, the rumored spud pre-training round.

Starting point is 00:22:17 So I wouldn't expect a huge update in capabilities at all, right, with a 5-6. So if this was like a GPD6, I would say, yeah, we're going to, it's going to probably beat So my guess is once all the benchmarks come out, I don't think five, six is going to beat Mythos on all benchmarks. It will on many, right? But there's dozens. But I think if you look at the artificial, artificial analysis intelligence index, I would assume that Mythos would probably still be ahead.

Starting point is 00:22:51 It'll likely be close, right? But the reality is Mythos, when it first came out and, you know, Anthropic, you could say to their credit or discredit spent months, you know, trumpeting how this was a weapon and how it's so powerful and the public candy can get access to it, right? And they were essentially begging the government to intervene. And they actually, Anthropic even said, hey, we should slow down model development, right? Anyways, eventually the Trump administration paid attention. And on June 2nd, they signed an executive order asking AI labs to hand over their top models for a 30, day preview. So then researchers get 30 days to test a model before it's a public release. So it's

Starting point is 00:23:36 voluntary. It's not an actual rule or law or anything like that. But essentially, you know, all the tech companies, they want to be on the Trump administration's good side. So they're going through, you know, this process. So, but the reality is, well, it had already hit Anthropic. So it looks like opening I took a slightly different approach. here. Anthropic, according to reports, kind of rushed some certain things through. There was maybe some talk from the government that Anthropic, according to reports, did not take seriously. And that's one of the reasons when reportedly Andy Jassy, the CEO of Amazon, reached out to someone in the administration. It's one of the reasons they pulled it because allegedly Anthropic was maybe not

Starting point is 00:24:28 taking everything the U.S. government said seriously because they had been in a little bit of a of a spat with the U.S. government over the last few months. So let's talk about that. So Anthropic launched its Fable 5 in Mythos 5 models to the public a couple of weeks ago on June 9th, but three days later that it was a federal export order over foreign access made Anthropic pulled that worldwide. So essentially when someone from Amazon reportedly found what they said was a jail break or a vulnerability of Fable 5 and Mythos 5. And that Anthropic reportedly did not take it seriously. The federal government just said, okay, well, you can't let any foreign nationals use this.

Starting point is 00:25:12 Obviously, Anthropics said that's a bizarre request. We can't do anything with that, right? They didn't actually say bizarre request. But they essentially said you can't enforce that. So the only thing we can do to comply with this is to pull this model worldwide, right? You can't, you know, obviously Anthropic has for national, as all companies do, working in their organization. So they had to pull it off the shelves. And that order reportedly gave Anthropic just 90 minutes to comply with no room to negotiate, according to a Fortune report.

Starting point is 00:25:46 So that led to this forced shutdown that Anthropic led with. So seemingly, Open AI maybe learned from that. and instead took a much different route, probably a slower and more, according to reports anyways, a much more thoughtful approach in this because Anthropic faced a sudden no-notice shutdown and they've been offline never since. We did get word Friday just hours after the GPD 5.6 limited rollout that some organizations regained access to mythos, not to fete. But OpenAI instead negotiated the limited preview that we have, so it's been controlled from the very start.

Starting point is 00:26:34 But both restrictions trace back to Washington fueling this fierce debate over fairness and control and ultimately what this means between the race between U.S. labs in China. So why does that part matter? And it does, and we're going to end with this here in a minute. AI moves too fast to follow, but you're expected to keep up. Otherwise, your career or company might lag behind while AI native competitors leap ahead. But you don't have 10 hours a day to understand it all. That's what I do for you.

Starting point is 00:27:13 But after 700 plus episodes of everyday AI, the most common questions I get is, where do I start? That's why we created the Start Here series, an ongoing podcast series of more than a dozen episodes you can listen to in order. It covers the AI basics for beginners and sharpens the skills of AI champions pushing their companies forward. In the ongoing series, we explain complex trends in simple language that you can turn into action. There's three ways to jump in. Number one, go scroll back to the first one in episode 691. Number two, tap the link in your show notes at any time for the start here series. Or you can just go to start here series.com, which also gives you free access to our inner circle community

Starting point is 00:27:58 where you can connect with other business leaders doing the same. The start here series will slow down the pace of AI so you can get ahead. It's no secret. Most Chinese labs, right? Anthropic just came out and accused Alibaba of the biggest distillation yet. So more or less, these Chinese labs are taking the work of US labs using their outputs to train their own models instead of doing it themselves because they can do that for a fraction of the cost, right? So that is the four dummies version of what's

Starting point is 00:28:41 going on in AI right now. And you might say, okay, why does it matter if Chinese labs are stealing, you know, US labs work, you know, and distilling models off of their big models. Well, ultimately, as we've seen here from this kind of situation now that's happening in Washington, but impacting companies worldwide is the cyber capabilities. I've been saying this for literally years, right? AI will become the new military. I would assume that most military actions of the future will be first launched in some way, shape or form with AI, whether that's using AI on the battlefield. And AI will be viewed as the

Starting point is 00:29:27 ultimate weapon or whether it is cyber. So that's why this is such a huge deal when it comes to geopolitical tensions. And specifically, it is the U.S. developing their closed proprietary models versus China finding all these loopholes to exploit. You know, it seems like Anthropic is the one that's getting hit the hardest. You know, so all these Chinese companies distilling anthropics work and putting it out there for a fraction of the costs. So, but critics are warning that gating frontier AI right now shuts out students, independent builders, smaller companies, and maybe ultimately, well, U.S. enterprise, right?

Starting point is 00:30:12 Because there's so many enterprises right now that refuse to touch Chinese open source models or open weight models. And many have valid reasons that make sense. Some maybe don't have valid reasons and it's more of just a stigma, right? Because you're not, you want more control or more security or promise in what a model, what went into a model. So you can trust its outputs a little bit more because large language models are already this, you know, very gray, unexplainable, you know, mystery box.

Starting point is 00:30:48 So many enterprises, and they, many of them have legal reasons that they can't use these Chinese models. But regardless, it does create this divide in the U.S. enterprise where it's saying, okay, what are we going to do here? You know, are we going to still, you know, let's just use Anthropic as the example because I think Anthropic on the API side, I think it's safe to say that's their bread and butter, right? It is open AI's bread and butter. They're driving subscriptions with millions of paying business customers. That's not Anthropics. Anthropics is fewer customers who are paying way higher API bills. So something like this really does threaten Anthropics sustainability.

Starting point is 00:31:33 And that's one of the reasons they've been, especially over the past month, the loudest about some of these accusations about the Chinese distilling their models. But ultimately, this. slow down. Now, this permission slip AI that open AI and Anthropic are having to play in is creating maybe an unfair advantage for the open source counterparts because, you know, as innovation gets delayed, even if it's for 30 days or, you know, if it's 30 days and then another 30 days to address whatever concerns come up in the first 30 days and then having this tiered rollout, right? even if it's a 30, 60, 90 day delay where maybe the world won't get mythos level or fable,

Starting point is 00:32:20 you know, model that we could have gotten back in May. Maybe we won't all get it till July or August, right? And that 30, 60, 90 day delay, if the Chinese labs are distilling whatever models are available, that's maybe going to start to close that gap. We've already seen the gap close. from open source or open weight models to proprietary. Like I said, I think it used to be six months. Now we're probably down to two with the new with CAI's new GLM 5.2.

Starting point is 00:32:56 So with containment in doubt and key facts still hidden, leaders have to plan for several outcomes. So this is not just talking about the new GBT 5.6 models. This is also talking about the new Fable 5 in Mythos 5. And I'm sure soon we'll be seeing rumors of, you know, Mythos 51 and Fable 51. And eventually any day now, we should be hearing from Google with Google Gemini 5. Or sorry, 3.5 Pro. Regardless, this is the new reality. Your company, I'm very likely, right?

Starting point is 00:33:33 Very sure when I say this. The overwhelming majority of enterprises in the U.S. are running their businesses now, not their AI strategy, running their businesses on frontier AI from the U.S. So what happens when that gets delayed now, 30, 60, 90 days? Or if it's just always going to be two tiers, if AI is no longer democratized, maybe it's only going to be the 1% of companies

Starting point is 00:34:01 that are going to get access to these models. That is a potential reality that I think everyone out there has to prepare for. And we don't get to sit here and say, oh, it's not fair. It doesn't matter. Right. If your biggest competitor has access to, you know, Mythos 5 and GPT5, 6 soul and you don't, there may be nothing you can do about it.

Starting point is 00:34:25 So you have to plan accordingly. So model access is now a vendor and supply chain risk, not just a simple subscription that you approve or disprove. So you have to be able to build model fallbacks in routing. Routing is so important because, like I said, you may not need a mythos level model for 99% of your company's work or a GPD 5.6 sole level model for the majority of what your company is trying to produce. So you have to be able to start breaking down. What are those day-to-day pivotal work tasks? Which models can they be mapped to?

Starting point is 00:35:04 So it's all these things I've been talking about for years. now you can see why I've been blabbing on about them because they are now incredibly more important than ever if there is no longer this true democratization of frontier AI, right? Because you might have to start making a choice if your company may never get access to a GPD 5.6 soul, right? Open AI did say we are planning to release this to everyone. Anthropic has not been very clear on what they plan to do with Mythos 5. previously when they announced Fable 5, they only said that they were going to include it in subscriptions for a couple of weeks. And, you know, so there's no guarantee that your company can have access to all frontier options, right? You might get some from OpenAI. We'll see

Starting point is 00:35:54 what Google says. Previously, Anthropic had kind of said they're not going to include it in subscriptions. So you need to do this. Model access is not something that we can take for granted anymore. And you have to go through reverse engineer every single thing that's important to your company. Map the model that does it the best right now that's available right now. And don't just say we're going to have a frontier model and we're going to pay whatever price and we're going to run all of our workflows through that because not only is that not sustainable, but if you do that, if you put too many of your workflows in one bucket that may

Starting point is 00:36:35 disappear, so too may your company's progress overnight. And you don't want that to happen. So yes, you need to build those model fallbacks and routing. So one gated tool can't hold your operations, but you got to pay closer attention than ever. The fine prints of these model releases, the restrictions, everything that's happening. And that's what we're going to continue to do every single day on everyday AI. So I know this one went in slightly different direction, but that's the reality because we don't have a model to play with. When it comes to GPD 5.6 right now, we can't do it hands-on anymore with Fable 5 either. So I think it's important to not just know what's new, but why the U.S. government is blocking it and what your business should be doing next.

Starting point is 00:37:21 So I hope that we covered at least most of that on today's show. And I hope this was helpful. And as we do get access, you better believe whether it's GPD 5.6 Seoul, or mythos. We're going to continue to get more hands on so your company can get ahead. So if this was helpful, please do me a favor and subscribe on the podcast. If you're listening on Spotify or Apple Podcast, then make sure to go to your EverydayAI.com. We're going to be recapping the highlights of today's show, as well as all of the other AI news you need to know to grow your company and career. Thanks for tuning in. Hope to see you back tomorrow and every day for more Everyday AI. Thanks y'all.

Starting point is 00:37:59 And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.

Everyday AI Podcast – An AI and ChatGPT Podcast - Ep 809: OpenAI’s New GPT-5.6 Release: What's New, Why the Feds Are Blocking it and What's Next

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.