Reproduce “The ATOM Report: Measuring the Open Language Model Ecosystem” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Nathan Lambert, Florian Brand

An analysis of the open language model ecosystem quantifies adoption trends, revealing a shift in leadership towards Chinese models, with Qwen emerging as the most-used family. It introduces the Relative Adoption Metric (RAM) to provide a contextualized assessment of model popularity across different sizes and over time.

Key insights

  • The open language model ecosystem has experienced a rapid and significant shift in adoption leadership from the U.S. to China, reflected across downloads, derivative creation, and inference market share.
  • Specific model families and organizations, notably Alibaba's Qwen and DeepSeek, have become dominant forces in adoption and performance, surpassing earlier leaders.
  • Raw download counts are insufficient for a nuanced understanding of adoption, requiring metrics like RAM that account for model size and temporal dynamics to reveal true momentum.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.