Calibration

ARC-AGI-1 → Vals GPQA Diamond

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

Vals GPQA Diamond is estimated from ARC-AGI-1 with a linear curve fitted on 21 models measured on both: y = 0.3479·x + 0.5998, R² = 0.77, cross-validated error 2.3 pp. It is used for 1 estimate.

Estimated model	ARC-AGI-1	Vals GPQA Diamond	Source
GPT-5.1	72.8%	85.3%	estimated ± 2.3 pp, high confidence