Calibration

ARC-AGI-3 → AA-Omniscience Accuracy

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

AA-Omniscience Accuracy is estimated from ARC-AGI-3 with a Hill curve fitted on 16 models measured on both: y = 0.4385 + (0.9797 − 0.4385)·x^0.30 / (5.00000^0.30 + x^0.30), R² = 0.64, cross-validated error 4.6 pp. It is used for 1 estimate.

Estimated model	ARC-AGI-3	AA-Omniscience Accuracy	Source
Grok 4.20	0.1%	47.6%	estimated ± 4.6 pp, medium confidence