O One

“And I think actually one of the most underrated developments of the past few years was the O-one model from OpenAI in late 2024, where if you recall, there was a chart that showed, okay, test time, compute, or in amount of inference done, amount of thinking ou…”

“Even at a company like OpenAI, you would have people ask naturally, why do something when you have a machine that works? And fundamentally, you know, it's to the credit of, you know, Jakob, Ilya, many of the people who really had conviction and vision in this …”

“The cost of a million tokens on an O-one equivalent model in December of 24 was about 26 bucks. And then in November of 25, it was 30 cents. So we saw another 88 X drop, not 88% or 80, you know, 88, 88 times cheaper in 11 months for that next generation of mod…”

“So what O-one showed is if you spend test time compute, you can get better results. So that was very exciting to me because there was one way of investing resources that was beyond the training set.”

“And I think OpenAI deserves a lot of credit for you know, releasing the first, like, serious RL plus LLMs release with O-one. And I think this really kicked off a pretty, you know, substantial change because it opened up a new axis of scaling, right? There was…”

“For a long time, OpenAI was charging more per token for the reasoning model, right, O-one and O-three than they were for GPT-Four-O, even though the architecture is, like, basically the same. It's just the weights are different.”

“The cost of intelligence for a given set of intelligence, let's say GPT-IV, let's say O-one, whatever, it is literally falling a hundred X over the course of one year.”

“The model outperforming expert labelers is, is possible. The model learning, like, self-direction is, like, expected. And yeah, we've seen, like, kind of cool emergent behaviors with, like, you know, like, O-one, O-three, R-one, kind of, like, these, like, thi…”

“Pure LLM systems are scoring like zero percent now again on on arc B two single COT systems like R one and O one score like one percent.”

“The R-one model is, is basically comparable to O-one, which OpenAI released four months ago and was training on internally, call it nine or 10 months ago. So OpenAI is on O-three now. Its frontier is ahead of where R-one is.”

“Like O-one is not a chat model. And I think this is both from a usage perspective, but also ties back to some of the training stuff and post training that we already talked about. Like the previous models were so focused on early chat based on, especially on c…”

“Anything where it just seems simple and like, you can do one off and you don't need to bring a ton of context into it. You can typically use like sonnet or GPT for that. Anything where I feel like if I was going to try to implement it myself and I would need t…”

“One way of thinking about what we built is like O-one for search because, Oh, one is all about like, you know, some questions require more compute than others, and we'll put as much compute into the question as we need to solve it. So similarly with our search…”

“When I do this with O-one, right, because it's doing that thinking phase of 10,000... It spends a lot of memory on generating this KV cache and reading this KV cache constantly. Now the maximum batch size, i.e. Concurrent users I can have, is a fraction of tha…”

“If you're interested in you know, your, Open replication of what OpenAI's O-one is you're gonna be on the 10 K spectrum of our GPUs.”

“So you may have heard of O-one from OpenAI, and there is kind of similar work at Meta and other places where this sort of very basic forms of reasoning that consists in having an LLM produce lots of different sequences of words, and then having a way of search…”

“It sounds like basically with O-one, the chain of thoughts will replace the workflow. So you might not need to break it down into steps yourself, but the evals are still really important.”

“There's a huge amount of infrastructure and scaffolding to build for sure, but I think O-one points the way to a model that is capable of doing great agentic tasks.”

“It seems like, or, 15% of the batch are already using O-one, even though it's not, like, fully available yet.”

“And I think O-one is going to do that to agentic frameworks as well. Hey, I think To me, it seems very unlikely that the, you know, you and me sort of like sipping an espresso and thinking about how, like, different personified roles of people should interact …”

“I think we clearly got to level two, or we clearly got to level two with O-one.”

“And O-one is obviously our most capable model ever, but it's also our most aligned model ever by a lot.”

“And then they'll have the actual O-one production build probably in the next couple of months, which will be probably pretty spectacular.”