Twelve daily feeds collect signals from public AI infrastructure —
pricing, token volumes, benchmark scores, GPU costs, agent commit
activity. The intelligence layer reconciles them into a canonical
registry and an IRT-fitted capability score. The output is the
research you read at synopticonresearch.com.
Below: today's status, the data scale, and a link into the
interactive pipeline map.
Full force-directed map: every feed, the intelligence layer, the derived datasets, the published output. Click any node for details.
→Per-feed page: source, auth, output shape, latest snapshot rendered as JSON. One scrolling page, twelve sections.
→How raw scores from 18 sources become an IRT-fitted capability ranking. Coverage heatmap of model × benchmark from real data.
→Articles + research, auto-built from the index files. Status pips, links to the rendered pieces.
→