Nathan Lambert's written Congressional testimony on the state of open models - Chinese open-weight downloads now 2x America's, >80% of OpenRouter open-model usage

Worth a read if you track open weights. Nathan Lambert (interconnects) posted the written version of his Congressional testimony on where open models stand. The numbers that got me: Hugging Face downloads of Chinese open-weight models are about 3.2B vs 1.6B American since he started tracking in Aug 2025, OpenRouter open-model traffic went from ~1T tokens a week to ~80T in a year with Chinese models over 80% of it, and his estimate is Chinese open weights sit 2-5 months behind the closed frontier while American open weights are 6-9 months back. The gap is smallest in agentic coding, which matches what I see running evals - Qwen, GLM and Kimi are the default local choices for tool-use work now. Link: https://www.interconnects.ai/p/the-current-balance-of-power-in-open