Daily scan
Multimodal Model Trend
Scores rank discovery confidence, not research quality. Open the Rubric panel for the arithmetic.
Matching records
Release watch
Model families and new weights
A family tag records that the text names that lineage. It is a routing decision, not a claim about who built what.
Tracked families
Newest model, omni, and world-model records
Corpus rhythm
Volume over time
Counts describe discovery volume. A jump can mean a busy week upstream or a source that came back online.
By topic
Records per scan
Daily ledger
Source mix counts published records. Fetch health counts raw rows returned before scoring, so a source can be healthy and still contribute nothing.
| Scan | Coverage (UTC) | Records | Pinned | Source mix | Top topics | Funnel | Health |
|---|
Cumulative corpus
Artifacts and their context
Artifacts merge only on exact identifiers: DOI, arXiv id, Hugging Face or GitHub repository. Never on similar titles.
Curated catalogue
What already exists
Entries
Dashboard unavailable
The data file could not be loaded.
Build it locally with vlm-radar seed (offline) or
vlm-radar run (live), then reload this page.