Explains severity, threshold and failure action
What a failing check does to rows and to people, and how severity, threshold and action must agree.
Can AI do this?
AI does this well Most of the tested AI models master this. 3 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 100% | Mastered |
| gemini-3.1-pro-preview* | Flagship | 100% | Mastered |
| gpt-6-astra* | Flagship | 100% | Mastered |
| claude-fable-5-1* | Flagship | 100% | Mastered |
| gemini-3.8-flash* | Mid tier | 100% | Mastered |
| gpt-6.1-sol* | Mid tier | 100% | Mastered |
| claude-sonnet-5-5* | Mid tier | 99% | Mastered |
| gemini-3.5-flash-lite | Low cost | 90% | Not shown |
| gpt-6-luna | Low cost | 100% | Mastered |
| claude-haiku-5-5 | Low cost | 98% | Mastered |
* From the earlier benchmark of 2026-10: 10 data-modelling test tasks, 3 runs per task, graded by claude-opus-5-5. These models have not taken the current 123 tasks yet; the two methods differ, so compare their figures with care. They are left out of the “most models” counts and the best-model picks.
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: What a failing check does to rows and to people, and how severity, threshold and action must agree
How it is assessed: Makes the settings of sample checks consistent
Related knobits
Builds on
Needed for
- Q-321 Recognises disagreeing severity, threshold and failure action
- Q-322 Recognises a rejection destroying more than it protects
All 30 knobits of Derive data quality rules.
Source: Data-management practice used by the stage 3 reference and review checklist; formerly data-modelling K-755