A16Z: Latest Episode
Newest A16Z episode: Can Open Models Keep Up With OpenAI... with Nathan Lambert (October 14, 2026). Key takeaway, summary and previous episodes from 1% Better.
Page updated by 1% Better. New episodes are summarised within hours of release.
Can Open Models Keep Up With OpenAI and Anthropic? | Nathan Lambert
A16Z with Nathan Lambert · Published · 58m
Key takeaway
The biggest lever separating frontier AI labs from open-source isn't talent - it's compute. An average OpenAI researcher has access to 5,000 H100-equivalent GPUs at any time, a number that makes top Chinese researchers' 'jaws fall to the floor.' If you're building with AI, understanding this compute gap helps you make smarter bets on which tools and models will realistically close the capability gap for your use case.
Episode overview
A16Z partner Lisha Li interviews Nathan Lambert, AI researcher and founder of the non-profit lab Trillium, about the race between open and closed AI models. They unpack why OpenAI and Anthropic maintain structural advantages in compute and capital, why distillation's role in Chinese labs' progress is overstated by the scientific literature, and how post-training via RL environments has become the dominant frontier for capability gains. Nathan also explains Trillium's mission to openly document how AI is being used to improve AI itself, arguing that transparency around recursive self-improvement is more valuable - and more honest - than the fear-driven discourse currently dominating AI policy.
Read the full A16Z episode summary for key insights, quotes and action items.
Previous A16Z episodes
About this page
This page always shows the newest A16Z episode in the 1% Better library. We summarise each episode from its audio or captions within hours of release, so bookmark it to check what is new. Browse every A16Z summary and guide or all episode summaries.