New·See the latest water task rankings

Water Benchmarks

Use the Framework

Guides, the evaluation toolkit and resolved cases — structured scoring resources for the water sector, no vendor lock-in.

Guides

How the rubric works

The 0–12 scale, thresholds and the critical safety gate, explained.

Reading the leaderboard

What Q, Brier and ECE mean, and why calibration is orthogonal to quality.

The five failure families

FOUL, SCAL, OXID, MECH and NOWE — the taxonomy behind every case.

Downloads & toolkit

Report: AI diagnostic reasoning in reverse-osmosis desalination operations

The full v1.0 report: benchmark design, rubric and findings.

Download PDF (EN)Download PDF (ES)

Evaluation scorecard (XLSX)

The structured 0–12 scorecard your team can apply to your own cases.

Vendor questionnaire

The metadata we collect before evaluating a submitted model.

Resolved cases

RO-FOUL-001, end to end

A single fouling case shown the way the benchmark sees it: plant symptoms and sensor trends → the model's structured answer → the 0–12 scoring against the expert gold, with the safety gate applied.

See the example case

Want new guides and the toolkit as they ship?

Join the community