Calibration
AIME26 → AIME 2025
AIME 2025 is estimated from AIME26 with a Hill curve fitted on 6 models measured on both: y = 0.3858 + (1.1490 − 0.3858)·x^6.00 / (0.79247^6.00 + x^6.00), R² = 1.00, cross-validated error 1.5 pp. It is used for 23 estimates.
| Estimated model | AIME26 | AIME 2025 | Source |
|---|---|---|---|
| A.X K2 | 97.1% | 97.5% | estimated ± 1.5 pp, low confidence |
| Claude Opus 4.5 | 95.1% | 95.8% | estimated ± 1.5 pp, medium confidence |
| Gemma 4 12B | 77.5% | 74.2% | estimated ± 1.5 pp, medium confidence |
| GLM-5 | 95.8% | 96.4% | estimated ± 1.5 pp, medium confidence |
| GLM-5.1 | 95.3% | 95.9% | estimated ± 1.5 pp, medium confidence |
| GLM-5.2 | 99.2% | 99.2% | estimated ± 1.5 pp, low confidence |
| Inkling | 97.1% | 97.5% | estimated ± 1.5 pp, low confidence |
| Inkling-Small | 95.5% | 96.1% | estimated ± 1.5 pp, medium confidence |
| K-EXAONE 2.0 | 92.3% | 93.1% | estimated ± 1.5 pp, medium confidence |
| Kimi K2.6 | 96.4% | 96.9% | estimated ± 1.5 pp, low confidence |
| Ling 3.0 Flash | 93.2% | 94.0% | estimated ± 1.5 pp, medium confidence |
| LLaDA2.2-mini | 35.1% | 39.1% | estimated ± 1.5 pp, low confidence |
| LongCat-Flash-Lite-Sparse | 65.7% | 57.3% | estimated ± 1.5 pp, medium confidence |
| Muse Glimmer 30B | 94.7% | 95.4% | estimated ± 1.5 pp, medium confidence |
| Qwen3.5 397B | 93.3% | 94.1% | estimated ± 1.5 pp, medium confidence |
| Qwen3.6-27B | 94.1% | 94.8% | estimated ± 1.5 pp, medium confidence |
| Qwen3.6-35B-A3B | 92.7% | 93.5% | estimated ± 1.5 pp, medium confidence |
| Qwen3.6 Plus | 95.3% | 95.9% | estimated ± 1.5 pp, medium confidence |
| Beam | 97.8% | 98.1% | estimated ± 1.5 pp, low confidence |
| Solar Open 2 | 95.7% | 96.3% | estimated ± 1.5 pp, medium confidence |
| Solar Pro 4 | 95.3% | 95.9% | estimated ± 1.5 pp, medium confidence |
| ZAYA1-74B-Preview | 76.4% | 72.6% | estimated ± 1.5 pp, medium confidence |
| ZAYA1-8B | 89.1% | 89.6% | estimated ± 1.5 pp, medium confidence |