How dumb is your model?

Paste your text and get measured diagnostics in seconds: effective rank, space utilization, redundancy, all under a production embedding model. Free, no signup. Benchmarks measured on Blue Hen RE eval gates, not marketing claims.

Run the diagnostics

Run the free health check Compare models Hall of Cone Museum of Collapse Headquarters →

What you get

Collapse score

Effective rank and retrieval on a rotating slice, summarized for sharing. Higher collapse = lower effective rank. Org-trained models typically score better.

Side-by-side RAG

Same query, same corpus. Compare a commercial baseline against an org-trained model on multi-hop retrieval tasks.

Hall of Cone

Reference panel of baseline embedders ranked by effective rank. Validate improvements on the Validation Lab (slasso.com).

Museum of Collapse

The failure modes we measure against: anisotropy, dimension starvation, MRL truncation cliffs. Each one has a diagnostic or gate that catches it before a model ships.

Blue Hen RE

Relay Engine for governed embedding operations · RAG Embeddings in production. Enterprise lifecycle: measure, validate, deploy, improve.