Explore distinguished foundation models, iconic architectures, and meticulously curated metrics across the globe's most advanced artificial intelligence research.
We ignore marketing hype. Every parameter count, context length, and license type is extracted directly from official research papers and primary source code.
We strictly audit open weights vs. open-source terms, custom commercial licenses, and restricted API boundaries so you build with legal confidence.
Scores are stamped as verified only when published alongside reproducible code or standardized datasets, keeping our records fully scientific.
How model specifications go from raw research papers to a verified entry in our catalog.
We monitor research preprints, Hugging Face repositories, and developers' release calendars to catch new foundation models the moment they land.
Curators cross-reference parameter claims, licensing terms, and evaluation criteria with primary technical documentations to root out unverified claims.
Specifications are structured into a Zod-verified JSON model scheme, ensuring exact parameters, license, and deployment values match global standards.
Once validated, the entry is compiled into a static webpage route, providing fast, secure, and permanent tracking for developer reference.