01:44:23 · John Coogan, Thibault Sottiaux
[01:44:23] S1: Mhmm. Talk about progress in computer use. What is actually driving progress there? Is this just something that sort of comes for free with scale and and model advances, or is there deliberate data collection that's happening and some sort of flywheel that's unlocking new capabilities in computer use? [01:44:44] S4: Yeah. We've done a lot of efforts bespoke effort on Windows, Mac, and like mobile computer use, also like phone use as well. And so there's an entire team working on this. It doesn't just come for free, but what does come for free is like every time we push the efficiency frontier and the model gets like, you know, more efficient like at thinking and acting and it costs less tokens and it gets compressed in time, it also gets better at computer use because it reduces the latency and reduces the cost. And so the two compounds, we have a lot of gains that we're getting from also visual understanding and every time it improves, model just gets more precise so it doesn't have to correct itself. And it was like maybe it misclosed the button and they're just like, oh, it's just like, wait, I have to redo that. So every time it's like more accurate and more token efficient, computer just definitely benefits from it. And when you compare it to 5.5, it's like, you know, it's just really like three times faster. Know, we're not at all hitting a wall here in, like, how fast we can do computers.
01:45:39 · John Coogan, Thibault Sottiaux
[01:45:39] S1: Yeah. Can you talk about how the role of member of technical staff is evolving? Because you're you're you're talking about Soul Ultra going off and working for days at a time. And at a certain point, your job sort of evolves to if you have a launch tomorrow, don't kick off a task that takes four days. Even if the model's capable of it and will deliver something great in four days, you need it tomorrow. And so you have to size your workloads appropriately. How is the how how are you thinking about sizing work and and actually delegating the right the right chunk of work at this stage? [01:46:19] S4: Yeah. I find your question very interesting because, it actually highlights, like, a shift in our thinking over you know, since we had five sixes, you don't really instruct necessarily, you know, for a task that's gonna take four days. It's like, you know, you tell it all the information that you have. So, know, you're like, Hey, I have a launch tomorrow. Yeah. Right? And then it's like, keep track of the time and like understand that, you know, the PR needs to land, like, you know, I'm in that or like by 2AM, it will just like reason over it. Yep. You're not the one that needs to manage like, you know, all of that extraneous like complexity. Yeah. And so that's it. What you're seeing as well, like, you know, your relationship with the agent like changes over time as it gets more intelligent and you're just like, oh yeah, can just talk to you like, you know, another like, you know, super smart human.
01:46:19 · Thibault Sottiaux, Jordi Hays
[01:46:19] S4: Yeah. I find your question very interesting because, it actually highlights, like, a shift in our thinking over you know, since we had five sixes, you don't really instruct necessarily, you know, for a task that's gonna take four days. It's like, you know, you tell it all the information that you have. So, know, you're like, Hey, I have a launch tomorrow. Yeah. Right? And then it's like, keep track of the time and like understand that, you know, the PR needs to land, like, you know, I'm in that or like by 2AM, it will just like reason over it. Yep. You're not the one that needs to manage like, you know, all of that extraneous like complexity. Yeah. And so that's it. What you're seeing as well, like, you know, your relationship with the agent like changes over time as it gets more intelligent and you're just like, oh yeah, can just talk to you like, you know, another like, you know, super smart human. [01:46:58] S0: Yeah. Yeah. Talk about the efficiency of the model, what work went into that, why why it matters, you know, what kind of conversations you're having with, you know, big customers, all that stuff.
01:46:58 · Jordi Hays, Thibault Sottiaux
[01:46:58] S0: Yeah. Yeah. Talk about the efficiency of the model, what work went into that, why why it matters, you know, what kind of conversations you're having with, you know, big customers, all that stuff. [01:47:10] S4: Yeah. So what matters a lot right now is sitting at the frontier and you know, getting the max capability when you want it, but also for your normal average day to day task is like, you know, being super efficient and not just for latency, just because also we're seeing, you know, so like we had this era of token maxing and then, you know, we've been talking a lot with, you know, all the companies and enterprises that we're working super closely with and then they were like, oh, it's just a little bit, you know, maybe out of control. It's like, you know, what we want is like, you know, were in a highly efficient model that is steerable, controllable. We want the right, you know, spend control dashboards. And so we also have all of that. You can look at your spend, understand the ROI, but also you can rest knowing that, you know, this is actually like a super, super efficient model. And so it gets the job done with, you know, way way way fewer tokens, which you know, to you, you know, means that, know, you have to pay less for the same results, which is super important. And this is really the theme I think of the year is like, you know, that being on the efficiency of like, you know, performance Mhmm. And and and cost.
01:47:10 · Thibault Sottiaux, John Coogan
[01:47:10] S4: Yeah. So what matters a lot right now is sitting at the frontier and you know, getting the max capability when you want it, but also for your normal average day to day task is like, you know, being super efficient and not just for latency, just because also we're seeing, you know, so like we had this era of token maxing and then, you know, we've been talking a lot with, you know, all the companies and enterprises that we're working super closely with and then they were like, oh, it's just a little bit, you know, maybe out of control. It's like, you know, what we want is like, you know, were in a highly efficient model that is steerable, controllable. We want the right, you know, spend control dashboards. And so we also have all of that. You can look at your spend, understand the ROI, but also you can rest knowing that, you know, this is actually like a super, super efficient model. And so it gets the job done with, you know, way way way fewer tokens, which you know, to you, you know, means that, know, you have to pay less for the same results, which is super important. And this is really the theme I think of the year is like, you know, that being on the efficiency of like, you know, performance Mhmm. And and and cost. [01:48:11] S1: What about if you wanna spend more for faster, performance? What does the future of, either ASIC enabled, Cerebras enabled, Spark, and and fast mode, What do you wanna see develop there, either immediately or over the next couple weeks?
01:48:11 · John Coogan, Thibault Sottiaux, Bernt Øivind Børnich
[01:48:11] S1: What about if you wanna spend more for faster, performance? What does the future of, either ASIC enabled, Cerebras enabled, Spark, and and fast mode, What do you wanna see develop there, either immediately or over the next couple weeks? [01:48:27] S4: Yeah. I think what we are, truly working towards, like, it's a buffet of options. Right? So Mhmm. For for your normal, like, you know, interactive task is, you know, you're going to use five, six sole on medium or on high and you're going to have an amazing time. [01:48:41] S5: If you [01:48:41] S4: have a really hard problem and you're trying to, for example, find a cyber vulnerability in something, you're going to run ultra and you're going to run it for two days and it's just going leave no stone unturned and like, you know, invent novel techniques and you know, you're going to be like absolutely blown away by what it comes up with. But a lot of times you also just need speed, you know, like for some of the steps that we're dealing with is like, you know, we love working off of the Cerberus version of this, which is like at, you know, about like, you know, seven fifty tokens per second, which is an order of magnitude faster than the default version that we have on the API and in the product today. And this is just really situational or if you just want the very best usually absolutely no compromise. It does come at a cost. [01:49:27] S1: Of course. A lot of