Calibration
ARC-AGI-2 → AA-GPQA Diamond
AA-GPQA Diamond is estimated from ARC-AGI-2 with a linear curve fitted on 30 models measured on both: y = 0.1139·x + 0.8395, R² = 0.50, cross-validated error 2.7 pp. It is used for 6 estimates.
| Estimated model | ARC-AGI-2 | AA-GPQA Diamond | Source |
|---|---|---|---|
| Claude Sonnet 4.5 | 13.6% | 85.5% | estimated ± 2.7 pp, high confidence |
| Claude Sonnet 4.5 Thinking | 13.6% | 85.5% | estimated ± 2.7 pp, high confidence |
| dots3-note Preview | 81.4% | 93.2% | estimated ± 2.7 pp, high confidence |
| Gemini 3 Pro Deep Think | 45.1% | 89.1% | estimated ± 2.7 pp, high confidence |
| GPT-5.4 Pro | 83.3% | 93.4% | estimated ± 2.7 pp, high confidence |
| GPT-5.5 Pro | 84.2% | 93.5% | estimated ± 2.7 pp, high confidence |