Y Combinator's Garry Tan Urges US Open-Weight Labs to Distill Frontier Models

Distillation is a training technique where a smaller 'student' model learns to mimic a larger 'teacher' model, gaining similar capabilities at lower cost.

Tan argues that letting US open-weight labs distill American frontier models creates strong domestic alternatives to Chinese models like DeepSeek and Qwen.

No. Teams can evaluate current open models now, but should track licensing and legal terms for any distilled model before production use.