ABOUT
Search the open-weight ecosystem — and remember how it changed.
Open Model Weights combines an open-weight AI model search engine with a source-linked historical evidence layer for developers, researchers, organizations and AI agents.
An open-weight AI model search engine
Search and discovery remain a core product. The registry lets users find field-verified open-weight model records by developer, license, context, parameters, format, hardware and other source-backed fields. The goal is not to replace discovery with an audit product; it is to make discovery more trustworthy.
A historical evidence layer
Current values are only one part of the record. Open Model Weights retains observed evidence snapshots, repository revisions, evidence hashes and material field diffs so a model can be inspected over time. History starts when Open Model Weights first observes a state; earlier states are not reconstructed retroactively.
One data layer for humans and AI agents
The website, versioned JSON endpoints, OpenAPI document and read-only MCP server expose the same underlying evidence. This allows a person to search a model visually while an AI agent can retrieve current records, history and change data programmatically.
Source-first and explicit about uncertainty
Important claims are tied to listed source repositories or publisher evidence. Derived values remain labeled as derived, and missing evidence stays unknown or not disclosed instead of being converted into a negative claim.
Independent
Open Model Weights is not affiliated with the model developers listed in the registry. Model names and trademarks belong to their respective owners.