Can Open Models Keep Up With OpenAI and Anthropic? | Nathan Lambert
- Source
- youtube.com
- Author
- a16z
- Date
Context
Closely following the frontier is different from leading it. In an a16z interview, Nathan Lambert, founder of the nonprofit research lab Trillium, argues that OpenAI and Anthropic have structural advantages in capital, computing capacity and the ability to pioneer new training methods, while open-model labs, particularly in China, have proven able to follow close behind.
The distinction is between reaching a capability level and financing the research that first makes it possible. A narrow performance gap does not by itself show that those underlying advantages disappeared. The conversation weighs how much of the open labs' progress comes from , meaning training a model on a stronger model's outputs, versus technical talent. Lambert's argument is a way to frame competition, rather than a measured causal explanation for every open-model release.
- Lambert argues OpenAI and Anthropic hold structural advantages in compute, capital and the ability to pioneer new training paradigms, while open-model labs, particularly in China, have followed remarkably close behind.
- The conversation weighs how much open-model progress comes from distillation versus technical talent, and how data, RL environments and post-training increasingly shape model progress while compute remains a major constraint.
- Lambert's new nonprofit lab, Trillium, plans to release RL environments and document how AI is used to automate AI research, aiming to show researchers and policymakers what frontier systems can and cannot do.
- distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.
Checking sign-in…
Loading comments…