The lenses
Cross-Model lens
The Cross-Model lens runs the IADTX Structural Comparison Test against a second candidate model on the same outcome. It returns a side-by-side reading of which structural shape the data favours.
What it answers
The lens answers whether a candidate alternative model resolves the same outcome through the same structural shape, or a different one. The answer reads as two structural verdicts side by side, with the IADTX Structural Comparison Test winner shown as a separate row.
When to run it
Run the Cross-Model lens when the buyer has two competing model artefacts and wants to know which one the data favours structurally. The lens is also used during model selection, where the buyer is choosing between a baseline and a candidate, and the structural reading is the deciding evidence.
Public-shaped output
The result page renders a Cross-Model panel with two columns. The left column carries the verdict for the primary model. The right column carries the verdict for the candidate. A third row below the columns names the winner under the IADTX Structural Comparison Test, with the same closed vocabulary used elsewhere on the result page.
Input requirements
The lens consumes a CSV shaped like cross_model_sample.csv. The CSV carries the outcome column, the primary model's predictions in one column, the candidate model's predictions in a second column, and the shared feature set in the remaining columns. The lens runs both structural tests against the same outcome and reports the result.
Tier gate
Pro tier and above unlock the lens. Free and Growth do not. Pro and Business render the same lens panel; Enterprise adds a third optional column for a third candidate model on the same shared feature set.
| Tier | Unlocked | Candidate models |
|---|---|---|
| Free | ||
| Growth | ||
| Pro | ✓ | One candidate |
| Business | ✓ | One candidate |
| Enterprise | ✓ | Up to two candidates |
What buyers do with it
Buyers pick the model whose structural shape matches the outcome, or commission a third candidate when the two on file fail to resolve. Reviewers reading the report under SR 11-7 treat the Cross-Model panel as documented model-selection evidence; the panel reports the structural reading on each candidate, the IADTX Structural Comparison Test winner, and nothing more.