“pretty meaningfully overblown”

ConceptResearch

“another example of the narrative being fairly overblown”

ConceptResearch

“GPT-3 was smarter than me. I mean, it wasn't maybe as capable. I'll give myself that.”

ConceptResearch

“It was smarter than me.”

ConceptResearch

“China's latest model, Kimi K3, is out and it is inching ever closer to the frontier of American AI.”

ConceptResearch

“And so, you know, a bunch of that observability, this is a big part of, I think, where that sort of AI-centric work that sort of, you know, we believe our secret sauce, but certainly everybody's working on that sort of like AI-centric focus on sort of being able to detect thos...”

ConceptResearch

“Marketing's agent turns competitor moves into landing pages. Sales' agent enriches leads, drafts emails, and updates the CRM. Ops' agent chases the paperwork and tracks the budget. Every agent has access to shared context and follows your rules about scope and approvals. It's ...”

ConceptResearch

“Now for context, at 975 billion parameters, Inkling is quite a bit smaller than the 1.6 billion parameters of DeepSeek V4 Pro, but a little larger than the 750 billion parameters of ZAI's GLM-5.2. Importantly, TML pre-trained Inkling from scratch. And aside from a small bootst...”

ConceptResearch

“On coding, Inkling scored 54.3% on SweBench Pro and 63.8% on TerminalBench 2.1, again putting it ahead of NeMoTrON and K2.5, but a little behind K2.6 and at a significant shortfall compared to GLM-5.2, 5.6 SOL, and Fable-5. On GDPVal, which measures real-world tasks and is a s...”

ConceptResearch

“those are the types of like serendipity things that we'd actually like to automate”

ConceptResearch

“We have lots of funny, like, RL fails that are not explicitly reward hacking. Well, I mean, one is when we trained— one of the early things we did was like, can you just like make a plate map? Like, can you like lay the experimental conditions out on a plate. And it got annoye...”

ConceptResearch

“We are all in on the bitter lesson and scale. We think that methods that scale and that are general beat those that are not. You know, as Ilya said at NeurIPS last year, we have but one internet. It's the fossil fuel we fracked. We got every ounce of data that we could out of ...”

ConceptResearch

“So it's causal. It is sometimes referred to as causal.”

ConceptResearch

“Yeah, it's pretty interesting when you look at like the broad fingerprint of capabilities that's advertised here. So the first thing, by the way, from a You can tell what part of the model announcements I rush to immediately when I see these cyber capabilities, because that's ...”

ConceptResearch

“Yeah. And you know, how much does superintelligence help you with stuff building nuclear safely anyways? Onto projects in open source. And we do have a couple of interesting new models that are open source, starting with Nimotron-Labs-Diffusion, a tri-mode language model unify...”

ConceptResearch