Calibration
AIME26 → HMMT Feb 2025
HMMT Feb 2025 is estimated from AIME26 with a Hill curve fitted on 7 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^6.00 / (0.75985^6.00 + x^6.00), R² = 0.54, cross-validated error 1.9 pp. It is used for 10 estimates.
| Estimated model | AIME26 | HMMT Feb 2025 | Source |
|---|---|---|---|
| A.X K2 | 97.1% | 97.6% | estimated ± 1.9 pp, low confidence |
| Gemma 4 12B | 77.5% | 63.5% | estimated ± 1.9 pp, low confidence |
| Inkling | 97.1% | 97.6% | estimated ± 1.9 pp, low confidence |
| LFM2.5-8B-A1B | 50.0% | 9.0% | estimated ± 1.9 pp, low confidence |
| LLaDA2.2-mini | 35.1% | 1.1% | estimated ± 1.9 pp, low confidence |
| Muse Glimmer 30B | 94.7% | 94.7% | estimated ± 1.9 pp, medium confidence |
| Beam | 97.8% | 98.4% | estimated ± 1.9 pp, low confidence |
| Solar Pro 4 | 95.3% | 95.5% | estimated ± 1.9 pp, medium confidence |
| Ternary Bonsai 2 27B | 95.8% | 96.1% | estimated ± 1.9 pp, low confidence |
| ZAYA1-74B-Preview | 76.4% | 61.0% | estimated ± 1.9 pp, low confidence |