The newsletter Interconnects is launching two free projects dedicated to the open model ecosystem: an Artifacts Hub and a daily-updated adoption dashboard. These tools extend the analyses the publication has previously focused on specific models and monthly summaries.
The first project, Artifacts Hub, is a curated view of trending models on Hugging Face, the artificial intelligence model-sharing platform. It highlights inference tokens—the units of text processed when using a model—tracked via Open Router, model intelligence according to Artificial Analysis, as well as adoption metrics designed by the team using Hugging Face data. The site currently lists 792 models released over the last two years, whether they are large language models focused on text or multimodal generative models.
To compile this inventory, Interconnects says it tracks data for each model published on Hugging Face and analyzes the few thousand most important large language models, the list of which is public on GitHub and regularly updated. A selection of a few hundred of them then receives more detailed explanations.
For the most popular models, the Hub allows users to see at a glance the gap between them and the intelligence frontier, measured by Artificial Analysis’s Intelligence Index. It also allows comparison of adoption on Hugging Face and Open Router for similar models, consultation of scores for the metric called RAM (relative adoption metric), which relates downloads to availability duration and model size, or the VAIL similarity index, which links models of neighboring generations. This work was done in collaboration with Project VAIL, a startup specializing in AI verification, which the newsletter describes as one of its most loyal supporters.
A Dashboard on the Gap Between the US and China
The second project, the Adoption Dashboard, presents download counts and derived models, categorized by geographic region and organization. According to the publication, it highlights the gap between the United States and China, as well as the rise of new players in the open ecosystem.
Interconnects indicates that the comparison between the two countries, already presented in its The ATOM Project report, regularly came up in industry discussions without being sufficiently updated. The dashboard now aggregates data from that report and the ecosystem, with daily updates.
The publication specifies that these tools are part of its mission, which aims to support the open model ecosystem as users seek to employ them, particularly in a cost-competitive manner against cutting-edge models. It thanks Hugging Face, Open Router, and Artificial Analysis for their contributions, as well as Project VAIL, and invites its readers to share their feedback and ideas for using the data at [email protected].
