Recognises an equivalent solution not credited
Points withheld because the answer differs from the reference's wording or form.
Can AI do this?
Not settled yet The benchmark tested this, but the models' answers do not settle yet whether they master it. 0 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 83% | Not shown |
| gemini-3.5-flash-lite | Low cost | 83% | Not shown |
| gpt-6-luna | Low cost | 38% | Not shown |
| claude-haiku-5-5 | Low cost | 83% | Not shown |
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 criterion met by the substance
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: C1 = 0 for "recurring income" where the reference says "regular income" Acceptable: Full marks; the reason notes the equivalence
Also known as: Equivalent solution not credited
Related knobits
Builds on
All 16 knobits of Grade answers against a rubric.
Source: Review checklist 03-answer-grading, row 2 (Standards 6)