benchgap
Calibration

ARC-AGI-2 → AA-HLE

AA-HLE is estimated from ARC-AGI-2 with a offset logistic curve fitted on 34 models measured on both: y = 0.2877 + (1.2000 − 0.2877) / (1 + exp(−3.46·(x − 1.2080))), R² = 0.62, cross-validated error 6.3 pp. It is used for 2 estimates.

Estimated modelARC-AGI-2AA-HLESource
Claude Sonnet 4.513.6%30.9%estimated ± 6.3 pp, medium confidence
dots3-note Preview81.4%47.3%estimated ± 6.3 pp, medium confidence