Recognises a cap on a peripheral criterion
A cap on a criterion that is not the point of the test.
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 | 94% | Mastered |
| gemini-3.5-flash-lite | Low cost | 54% | Not shown |
| gpt-6-luna | Low cost | 94% | Mastered |
| claude-haiku-5-5 | Low cost | 94% | Mastered |
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: The cap on the central criterion
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: when_zero: Q1 (the word limit), max_total: 4 Acceptable: The cap on the criterion without which the task is missed
Also known as: Cap that guards no construct
Related knobits
Builds on
All 15 knobits of Write and review test items.
Source: Review checklist 02-test-items, row 6 (Practice)