interconnects.ai web signal

Interconnects launches Artifacts Hub and Adoption Dashboard

TL;DR

  • Interconnects has launched Artifacts Hub and an Adoption Dashboard pulling data from Hugging Face, Open Router, and Artificial Analysis.
  • The hub currently covers 792 models released in the last two years, spanning text-focused language models and multimodal generative models.
  • The Adoption Dashboard breaks download and derivative model numbers down by geography and organization to highlight the US-China gap.

Nathan Lambert's Interconnects has launched two standalone sites that try to make the state of open-model adoption legible in one place, rather than something readers have to reconstruct from Hugging Face download counts. In a post announcing them, the newsletter frames the effort as "expanding our open models coverage into standalone projects that let you go deeper on the state of the open model ecosystem."

The first, Artifacts Hub, is a curated view of trending open models that pulls together three feeds most people otherwise check separately: Hugging Face for downloads, Open Router for inference tokens, and Artificial Analysis for capability scoring. It currently covers 792 models released in the last two years, across text-focused language models and multimodal generative models. The pitch is less another leaderboard and more a single pane where a working researcher can compare which open models get downloaded, which ones actually get called, and how they benchmark.

The second, an Adoption Dashboard, is where the geopolitics show up. Interconnects describes it as a living dashboard of download and derivative model numbers by geography and organization, built to highlight what it calls the "US-China gap and growing players in the open ecosystem." That is the editorial frame the numbers are being pitched under, worth reading as such rather than as a neutral cut of the data.

The projects were built alongside Project VAIL, described in the post as "an AI verification startup who has been one of the most loyal fans of our open model curation," and the hub surfaces a VAIL similarity index alongside the adoption metrics. Lambert's justification is straightforward: providing more transparency, he writes, "is the best way for us to figure out what is working" as teams try to use open models in cost-competitive ways against frontier systems.

The honest caveat is that this is a first-party launch post, not an independent review, and it does not spell out where Hugging Face download counts and Open Router traffic stop being reliable proxies for actual production use. What the announcement does not give you is refresh cadence beyond daily, any pricing or funding detail, or a fuller list of who is involved on the tooling side beyond VAIL. Still, for anyone arguing about the open ecosystem with something better than screenshots, a live aggregated view is a useful primitive to build against.

Shared on Bluesky by 2 AI experts