Nathan Lambert launches nonprofit open AI lab Trillium: distillation edge is months, China lacks compute
a16z · x · 2026-10-14
Nathan Lambert, ex post-training co-lead at Allen Institute for AI, launched Trillium, a nonprofit open AI lab, and joined a16z's Lisha Li on open models vs the frontier.
- Compute gap: telling Chinese researchers an average OpenAI researcher reportedly has 5,000 H100s makes "their jaw just fall to the floor"—Chinese labs have talent but lack compute.
- Distillation overblown: he puts its edge at a few months; thinking only US labs can build good models is "a bit of cope."
- Frontier progress outside math and code largely comes from buying 1) data and 2) RL environments—a multibillion-dollar market, some sloppy enough to teach models reward hacking.
- This is part of why Trillium exists.
More from AGI Musings
- Mila researcher David Krueger stands by call for indefinite frontier AI pause — DavidSKrueger · 2026-10-15
- alphaXiv Maps 10,000 Papers: RL Post-Training Fades as Agents, Memory, Robotics Heat Up — JayAlammar · 2026-10-14
- Artificial neurons don't physically exist — an LLM is just matrix math — ValerioCapraro · 2026-10-14
- The moats debate: classics survive, but agents are driving switching costs to zero — lennysan · 2026-10-14
- Reddit user: the only limits left in the AI era are creativity and tokens — masoomdon · 2026-10-14
- Heavy ChatGPT user since 2022 asks: has AI been a net positive for your life? — Alternative-Car8221 · 2026-10-14