OpenBMB/BMList

What it solves

BMList addresses the challenge of keeping track of the rapidly growing number of large AI models. Without a centralized resource, it's hard to see the big picture of which models exist, who built them, and how the field is evolving.

How it works

The project maintains a curated list of models that meet two criteria: at least 1 billion parameters and a public release announcement (via paper, article, or news). The data is presented in a multi-dimensional table that can be filtered by attributes like organization, language, and domain. A companion website displays statistics and hosts a gallery of figures, including model trends over time and parameter distributions by organization and release date. The repository accepts contributions through guidelines, and any addition automatically updates the table and website.

Who it’s for

Researchers, developers, and AI enthusiasts who want to monitor large model releases, compare models across different criteria, or study the evolution of model scale and adoption.

Highlights

• Community-driven with clear contribution guidelines
• Interactive multi-dimensional table for flexible filtering
• Visual gallery covering model trends, organization counts, and parameter distributions
• Automatic updates to the website when new models are added
• Open source and welcomes corrections via issues or pull requests