Teaching Everyone to Fish for Tokens
Nvidia wants you building your own model, not buying from Anthropic/OpenAI.
by Nathan Lambert
Intel · Source
Interconnects
Everything the radar picked up from Interconnects, newest first.
20 items collected
GLM-5.3: How Chinese labs keep stride with the frontier
Hint: It’s really not a distillation story.
by Nathan LambertI wrote an AI textbook — how long until AI can do it better?
Reflections on AI's writing ability and how AI models get more capable.
by Nathan Lambert5 useful things you'll learn in my new post-training textbook (shipping now!)
After a few long years of finding time to document my lessons from training open models, my post-training book is done!
by Nathan LambertLessons from the hacks
Musings on model alignment, what determines safety, and where we go from here.
by Nathan LambertIntroducing our Artifacts Hub and Adoption Dashboard
Scaling our curation and measurement of the open ecosystem.
by Nathan LambertLatest open artifacts (#23): Laguna S2.1, Inkling, & Kimi K3 show the utility of open models on the Pareto frontier
Capacity to train strong models is proliferating.
by Florian BrandOpen models recap: more on Kimi K3, Qwen 3.8, Xi's WAIC speech, distillation, the open-closed gap, and what's next
A podcast with Florian Brand.
by Nathan LambertKimi K3: The open-weights escalation
The global implications on the AI ecosystem.
by Nathan Lambert6 months to live for open models
The most serious test to date of open source AI’s viability is happening right now.
by Nathan LambertLatest open artifacts (#22): Zyphra, Cohere, and Poolside are expanding the breadth of the ecosystem
An assessment of the open ecosystem and the motivations behind releasing models
by Florian BrandGLM-5.2 is the step change for open agents
A capability threshold I've been carefully monitoring.
by Nathan LambertBanning Open Source AI Would Be A Mistake
This post was originally an op-ed co-authored with Kevin Xu of Interconnected for a general, non-technical audience.
by Nathan LambertState of the blog, mid-2026
About 3 years since I started writing weekly.
by Nathan LambertFrontier post-training recipe review with Finbarr Timbers
"Interview" #18
by Nathan LambertWelcome to the AGI era of AI governance
It's a one-way door and we weren't ready for it.
by Nathan LambertClaude Fable 5 and new AI safety fables
One step further into the power politics of frontier AI systems.
by Nathan LambertFarewell Ai2
This was my last week at the Allen Institute for AI (Ai2), where I got the great privilege to work on the Olmo models, to grow, to learn, and to have broad lasting impacts.
by Nathan LambertOpen and closed models are on different exponentials
Where marginally higher intelligence drives value, and where it doesn't.
by Nathan LambertSome ideas for what comes next, May 2026
Gemini Flash 3.5, Mythos, open-closed balance, America's open-source surge, emerging power struggles and more.
by Nathan Lambert