Explains the traceability of target placements
Each subject area's target placement cites its current decision and keeps the agreed system of record.
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 | 97% | Not shown |
| gemini-3.5-flash-lite | Low cost | 93% | Not shown |
| gpt-6-luna | Low cost | 96% | Not shown |
| claude-haiku-5-5 | Low cost | 95% | 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: Each subject area's target placement cites its current decision and keeps the agreed system of record.
How it is assessed: Answers a written question on the target placement correctly
Related knobits
Builds on
Needed for
- A-405 Recognises a placement contradicting the system of record
- A-408 Recognises a placement citing a superseded decision
- A-409 Recognises a placement with no decision
- A-410 Recognises a subject area with no target placement
All 16 knobits of Record the target architecture and its decisions.
Source: Practice the stage 4 reference and review checklist rest on