Calibration
MMLU-Redux → Artificial Analysis Intelligence Index
Artificial Analysis Intelligence Index is estimated from MMLU-Redux with a Hill curve fitted on 7 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^6.00 / (1.18956^6.00 + x^6.00), R² = 0.69, cross-validated error 4.2 pp. It is used for 3 estimates.
| Estimated model | MMLU-Redux | Artificial Analysis Intelligence Index | Source |
|---|---|---|---|
| Mellum2-12B-A2.5B-Instruct | 78.1% | 8.9% | estimated ± 4.2 pp, low confidence |
| Mellum2-12B-A2.5B-Thinking | 86.2% | 15.2% | estimated ± 4.2 pp, medium confidence |
| Ternary Bonsai 2 27B | 89.1% | 18.0% | estimated ± 4.2 pp, medium confidence |