OpenAI President Greg Brockman on GPT-5.5 “Spud,” AI Model Moats, and a 'Compute Powered Economy'

OpenAI released its long-awaited GPT-5.5, or “Spud,” model on Thursday. The model is “a beginning point,” OpenAI President Greg Brockman told me shortly after release, setting up a series built to take action on your behalf and accomplish your goals. GPT-5.5, Brockman said, is adept at coding and computer control, able to act without arduous amounts of instruction.

“In many ways, it is a step towards a new way of getting work done with a computer,” Brockman said. “Being very proactive and really being able to solve problems end to end with a little instruction.”

In an ‘emergency’ episode of Big Technology Podcast, I spoke with Brockman about GPT-5.5’s capabilities, whether model commoditization is coming, how Spud plays in the ongoing cybersecurity conversation, and what he means when he says we’re heading toward a “Compute-powered economy.”

You can read our conversation, edited lightly for length and clarity, below. Or listen on Apple Podcasts, Spotify, or your podcast app of choice.

Alex Kantrowitz: Welcome Greg. First off, can you confirm GPT-5.5 is indeed ‘Spud?’

Greg Brockman: Yes

Okay, so what is GPT-5.5?

I think in many ways, it is a step towards a new way of getting work done with a computer. It’s a new class of intelligence. It’s extremely useful at things like programming, and all the different aspects of debugging and solving very hard and gnarly problems, just being very proactive and really being able to solve problems end to end with a little instruction.

But the thing that’s most remarkable is not necessarily the fact that it got better at coding. That, I think, is what everyone kind of expects.. But the fact that it’s now really crossed the threshold of usefulness for general kinds of applications. And so it’s much better at creating slides, spreadsheets, much better at computer use, using your browser, being able to kind of click through applications that are otherwise hard to have an AI operate. And so I think that we’re really seeing the emergence of this new way of using a computer, and it starts with this kind of intelligence at the core.

When we spoke last you mentioned that this was effectively the culmination of a two-year research process. This was planned two-years ago?

Yes, we do have very long horizons for how we plan now. One note is that we stack together many research ideas and bets on a variety of timescales. And so the way to think about it is that we are making constant progress across every single part of the stack.

And so what GPT-5.5 represents is not an endpoint in many ways. It’s a beginning point. It’s really a step towards the kinds of models that we see coming over even just upcoming months. And I think that you should expect that we are going to have even larger improvements in the capability across a wide variety of these aspects of what the model can do, and that’s something that I think will be very exciting.

Can you share specifically what those aspects are? If this is the beginning, what is it the beginning of?

Think about the models as the brain. You can think about the systems and the harnesses like Codex and the applications, like the Super App, as almost the body around it to make it into a useful AI. And that’s really what’s happening, is a shift from language models being the thing that is produced by labs like ourselves to an AI that’s useful, that’s an assistant that’s out there trying to solve your goal, that’s really operating according to your instruction.

And you can see now, Codex is becoming this app that’s not just for the coders. It’s really for anyone using a computer. And it’s not perfect. There are still some tasks where that it should be able to do it and it doesn’t quite get it right. Sometimes the personality isn’t quite what you wanted. It’s extremely powerful and out there doing a lot of really amazing things, but the way it communicates back to you — you have to still spend some time really trying to read through, okay, exactly how did it solve this problem?

And so these aspects, we know exactly how to make them much better. And we’ve already had a pretty remarkable improvement from 5.4 to 5.5. I think we’re going to have even more remarkable improvements across every single aspect of what makes these models useful.

Internally, we think a lot about the end application. That is one thing that changed for us over the past 12, 18 months, something like that, is that we used to really just be focused on — Let’s improve on the benchmarks. Let’s make these models more cerebrally capable. But we now are really focused on — Let’s bring them to real-world application. Let’s think about finance, sales, marketing, every single function that someone uses a computer. How can we help with their computer work? How can we make the model have not just theoretical capability to help but is experienced? Those kinds of tasks, that’s been able to see what good looks like.

And I think that the place we’re going is one where you as a person doing work, that you are the overseer, you are the CEO of almost this autonomous corporation, or, this fleet of agents, perhaps, is more, is the way to say it, and that they are operating according to your goals. Now you are still accountable You’re still in the driver’s seat. You’re still the person who thinks about, well, is this what I wanted, was this work up to standard, but that the details of exactly what buttons were clicked and exactly the code that was written, or exactly how the formula and the spreadsheet works, that you can abstract yourself from those if they’re not important to the evaluation of whether or not something was what you wanted. And so I think it’s like increasing leverage for every worker.

If this is a culmination of two-years of work, is it right to think of it as the moment reinforcement learning training methods overtook pretraining in importance?

I would say it a little differently. There’s many steps to the pipeline. There’s pre-training, mid-training, reinforcement learning, there’s data collection, there’s a lot of these different things that all come together to produce the end result. And the way in which is connected to the world, that’s also very key to making it useful.

The thing that I’m really saying is, we have been investing on every single one of these. We have a team that’s not just about individuals working on these pieces, but a team that really comes together and looks across the whole stack to say, how do we make this more useful for real-world applications?

And so it’s not really any one thing that we do. If you’re building a car— it’s not just about, do you have a better engine. You can build a great engine, but if the rest of the car is not up to the quality level of the engine, it’s not going to matter. And so I think that is the real innovation. It’s really the end-to-end, co-design and all coming together in a repeatable fashion to make these models better and better for our users.

You’ve said the model more intuitively knows what you want. How do you get it to know that? And does that mean prompt engineering is dead?

Number one, I think it really comes down to when we say there’s a new class of capability, a new class of intelligence, that’s really what we mean.. The models are becoming much more intuitive to use because they have deeper understanding of what it is you’re asking of them. They really look at the context, try to understand and puzzle out what am I being asked to do.

And it really makes you realize, to the second part is prompt engineering dead, which I think that prompt engineering, in some ways, may be even more vibrant than before, but you spend so much time now, trying to explain to your computer what you even want. You try to pack in this context and be well, here’s what’s going on, here’s the situation, here’s the thing I want from you. And you’re like — Why do I have to explain this to my computer? Right? The whole thing is the computer should be doing the work to help me. I don’t want to have to be, breaking down the task, trying to explain to it step-by-step how to do things. I want to point it in a direction, and I want it to be able to take care of the details and to get me the result again in a way that I can observe and provide feedback along the way. But I want it to be the driver of the of those like low level execution.

And so I think that in some ways, where prompt engineering is going to go is — it’s going to be about you can get so much more out of these models with so much less effort, but with the same amount of effort, you still have a multiplier.

Okay, let me briefly speak with you about the economics of building a model like this. There’s been this pattern where these big, massive models getting distilled by open source model makers, and then open source is just a couple months behind the leading foundational models. How is this defensible in the long term?

It’s not as simple as you can take the output to these models and distill and you have exactly the model the same capability, it’s just smaller and can run fast. If that were the case, we would just do that, and then we would also have a model that would be much more easy to serve in many ways. And of course, there’s a lot of art behind distillation.

Now the at the deployment side, we think a lot about safeguards. We think a lot about mitigations, and we do that for many, different aspects of how these models could be misused in real situations. And that’s something that we have been investing in for many years, and we think about that across areas like cyber, or thinking about that areas like bio, that we have a long standing effort that you can see in our preparedness framework, which is public, about how we approach these kinds of uses of the model, and how we try to make maximize the benefits, mitigate the risks.

And so I think it’s a real motion that every piece of what we do needs to connect to the question of, how do we continue to make progress, but also, how do we make these models broadly available? Because that’s something that we really believe in, that we believe this technology empowers people, and that we want it to benefit people and lift everyone up.

The pricing on this model is, double the last model, GPT-5.4. And so, from an economics or the business standpoint, if open source can deliver not as good performance, but almost as good, and do it cheaper, how do you handle that threat?

First of all, if you look at our history, which really is not driven by anything in competition. It’s just like our own progress and desire. We have dropped prices on the same level of intelligence year over year, sometimes by literally a factor of 100. It’s like at least an order of magnitude year over year, sometimes literally 100.

The thing that keeps happening, it’s real Jevons paradox, where it’s you lower the cost of something and way more activity happens And I think that what we keep seeing is that there are returns to intelligence that for the kinds of tasks that these models are now capable of doing, that a little bit more intelligence goes a long way.

In some ways, you can almost look at it as oh, there’s just an incremental improvement in intelligence, but I think there’s going to be a massive improvement in terms of what people use it for. And by the way, I think that incremental is very much an understatement for this model relative to five, it’s a point one improvement, in some ways, but I think that really undersells the magic that we see within this model, and that our early testers have really seen in their practical work.

So if people see these numbers and they say — there’s IPO pressure on, OpenAI, and therefore, we’ve been getting a great deal on intelligence, and the free ride is over. You would argue against that?

We have a very simple business in some ways. We rent build by compute, and we resell it with some positive margin, and as long as it’s, positive operating margin, and as long as there’s scalable demand for intelligence, which I think is true, as long as there’s problems to solve, then we can scale that compute all day.

That’s the main directive that I that I ask of the team, is just think about we need to add value onto. Up at the raw compute and make sure that we are at positive operating margin on it, and that is something where it’s not even about the different competition in the marketplace. It’s just a question of, can you have compute that gets turned into intelligence, and that’s just how that it does that at a slightly improved value coming out relative to the cost going in.

And I think that is something where, again, we’re always trying to make more efficient models, but then we just want more of them, and then we want the more intelligent models. And regardless of where they’re coming from, it’s all the same compute that’s going in. And so I think that it’s a great like competition in this marketplace has been great for innovation, but I think that it’s something where it’s driving more usage and more overall spend in the ecosystem, and you can see that in the revenue numbers of us and others in this industry.

Regarding cybersecurity, there are two very different approaches between OpenAI and Anthropic: Anthropic’s latest, massive model, Mythos, is not released to the public. This one, Spud, or GPT-5.5 is released to the public. Is there a chance that releasing this powerful model could lead to some major cyber attacks?

The thing to understand is that we have been investing in cyber safeguards and cybersecurity as a part of our preparedness framework for years. This is something we have invested in far ahead of having the kinds of capabilities we see coming.

And so we have been taking a very deliberate step-by-step approach. You can see, even just over the past couple of weeks, where we’ve expanded our trusted access for cyber program. And, in general, we believe in ecosystem resilience. We think that you do want to go step-by-step, that these models are getting continuously better. We have line of sights and even more capable ones, and that you want to be able to put these models in the hands of defenders to make sure that you’re able to protect critical infrastructure.

And we believe in that resilience of as you can bring these models into people’s hands, that then they’re able to explore in ways that you would not be able to without that access. And so you want this graduated approach and to make sure that you are moving down that pipeline as you can bring in additional safeguards in order to make sure that you can maximize the benefits and mitigate the risks.

And so we’ve really taken a deliberate approach. I think our team has been working incredibly hard to think through the cyber implications of this model. We also believe in iterative deployment. That’s part of this really bringing the models as they continuously get better. And we believe in democratic access. And we believe that ultimately, the goal of creating this technology is to empower people to ensure that it does benefit all of humanity. And so we are constantly trying to solve for how do we safely and responsibly bring this technology to bear in the world in a broad way.

Let me talk through the other case and then get your response. The other case would be, there are you can’t account for everything, and there are clearly going to be some vulnerabilities that will only be found by people or entities deploying this and looking for them. So maybe it makes sense to start with a trusted group of testers before you deploy it, what do you think?

Well, I believe the correct answer here is subtle, and I think it is rooted in the technical specifics of what you have in front of you and many factors. You need to think about, how are the models progressing not just your own capabilities, but others in the ecosystem. You need to think about, what benefit do you get from having a small group that has access and are they able to have high leverage by being able to find and produce patches? But then, how do you coordinate the disclosure of those across an industry?

And so there’s a lot of factors that go into it. I think that the true answer is if either extreme is not quite there are tools that can be applied to. Specific situation, and I think that this is not the first time we’ve had to think about this problem. It’s not the last time we will have to think about it.

But one thing to note is that we have had our model in the hands of defenders for some time, that we’ve been building up our trusted access program, that the model that we’re releasing is not cyber permissive that it has a number of safeguards built into it, and that you can then have a gap between what you’re privately sharing, testing those kinds of things.

And so I think my short answer is, it’s there’s definitely these different schools of thought in terms of values — of is the value that you want to get these models into people’s hands and empower them, or is the value that you want them to be centralized and controlled, and that you don’t want them in people’s hands. That is something that is a maybe underlying tension in some of these debates. But I think that the tactics that those almost flow from the details, and that they can be informed by these values, but either extreme, reflexively, I don’t think will yield the best outcome for the world.

Back to agents. These agents work the best if you let them have a high degree of autonomy. But that’s risky. What is the proper amount of trust to put into agents now?

So I think that now, agents tend to be quite reliable, and even things like prompt injections, I think that there’s still holes there, but that we’re patching them, and that the models are becoming much more resilient.

But I also think that the flip side is that — just like if you have employees, if you have a team of five employees, they’re all trustworthy, fine, but if you have 500,000 of the same employees, that there’s the law of large numbers that you start to worry about, okay, how do I have good governance and oversight.

So this is something where as we’re investing in these capabilities and making the Super App more accessible, not just to coders, but to any person doing work with a computer, we’re also investing in governance and oversight. And you can see this very concretely in workspace agents, which we released recently, that’s within your enterprise, you can now define agents so you get a hosted Codex harness in the cloud. You can hook up tools. You can hook it up to your Slack, and it’s doing work. It’s awesome. A lot of people use it. It’s been very cool to see how viral it goes within an organization. When you see someone else’s agent, you’re like wait, I can build one of these too, and you can just fork it and do your own thing, and then that’s an opportunity to have great governance and that’s baked into the product.

So I think that the short answer is you want to ramp the responsibility entrusted with the agent, and the diversity of things that agents are doing together, with security, safety, observability, oversight, and if you’re not doing those hand in hand, then I think that that’s a little bit out of balance. And I think it’s important to think about both sides.

Go ahead, but be careful?

And really lean in. I think it’s as you scale, you can prototype. That starts to bring in — Do you still have the ability to oversee what’s going on? So you need to make sure at each step. Do you feel like you’re calibrated, you understand what your teams are up to?

Greg, let’s end with this. You’ve called this a “compute powered economy.” What does that mean?

Well, I think we are heading to a world where the more compute is poured into a problem, the faster that problem will be solved, and that the ceiling of problem that can be solved depends on how much compute is available. And you think about things like drug discovery — being able to solve complex diseases like Alzheimer’s is outside of humanity’s reach. Right now. But imagine a world where you can take a gigawatt data center and have it just think about how to solve Alzheimer’s for a month, for a year, however long it takes.

And it may not be literally just cerebrally solving this problem, you may have to consult with world experts. Maybe it has to suggest experiments that get run in a wet lab. But if you can solve such a problem, that would be such a transformatively positive thing for humanity, and I think we were heading to a world where that is how important problems get solved, and that is how tasks in your daily life can also be solved, whether it’s having an agent that knows you, that has your personal context, that is trustworthy, that you can ask for advice on health, and you get back trustworthy information, and that’s just a thing. That’s a smartphone that’s in your pocket You can just talk to and it’ll be out there doing things and proactively knows, what are your goals? What are your interests? And how it can help you.

Compute is going to be the resource that shows how much computers can be used to help people to do work on behalf of people. We’re heading to that world, and it’s one that we’re all building collectively.

That would explain the massive investments that you’ve led making these big infrastructure bets…

Still not enough. We’re gonna feel the scarcity. We’re gonna feel it. We’re feeling it already. You can sense it now on people who are trying to use these agents, and just simply cannot, hitting the rate limits. So we’re working on behalf of our customers, on behalf of everyone who wants to use these agents to ensure that there is enough. And I don’t think we’re gonna get there. We’re gonna do our best, but I think that we are headed to a world of compute scarcity. And again, I think this is something where we can all contribute to trying to help there just be more availability of this in the world.

Greg, thanks again for coming on.

Likewise, great chatting.