Project
ModelRegistry
ModelRegistry is a community-driven, machine-readable index tracking frontier AI models across 14 laboratories with a clear dual-tier structure: reigning foundation flagships developers use in production, plus newly dropped specialized checkpoints. Everything is syndicated through a public REST API, RSS feed, llms.txt, embeddable badges, and a zero-install curl dashboard.
Tech Stack
The Challenge
Frontier model releases ship weekly across 14+ labs, but developers rely on stale charts and buried changelogs, so outdated models clutter workflows while new flagships go unnoticed.
The Solution
Built an open registry with a dual-tier structure that separates heavyweight foundation flagships for production use from newly dropped specialized checkpoints, each tracked with context windows, pricing, and access terms.
Open Telemetry
Every model is machine-readable via a public JSON REST API with filters, an RSS 2.0 feed, llms.txt ground truth for AI crawlers, embeddable flagship badges, and a terminal dashboard with zero installation.
Contributing
A single-file contribution workflow: edit one data file, run validation that auto-syncs the README index, and open a pull request in under 60 seconds.
Metrics
laboratories
14
models
29
syndication
REST + RSS
license
MIT