Calibration
FrontierMath (legacy) → FrontierMath v2 (Tier 4)
FrontierMath v2 (Tier 4) is estimated from FrontierMath (legacy) with a Hill curve fitted on 6 models measured on both: y = 0.3398 + (1.2000 − 0.3398)·x^6.00 / (0.88796^6.00 + x^6.00), R² = 0.97, cross-validated error 5.0 pp. It is used for 1 estimate.
| Estimated model | FrontierMath (legacy) | FrontierMath v2 (Tier 4) | Source |
|---|---|---|---|
| Claude Opus 4.7 (Adaptive) | 43.8% | 35.2% | estimated ± 5.0 pp, low confidence |