Calibration
GDPval-AA → AA EnterpriseOps-Gym
AA EnterpriseOps-Gym is estimated from GDPval-AA with a Hill curve fitted on 12 models measured on both: y = 0.3252 + (0.4910 − 0.3252)·x^6.00 / (0.41139^6.00 + x^6.00), R² = 0.50, cross-validated error 6.7 pp. It is used for 25 estimates.
| Estimated model | GDPval-AA | AA EnterpriseOps-Gym | Source |
|---|---|---|---|
| A.X K2 | 23.2% | 33.0% | estimated ± 6.7 pp, low confidence |
| Apodex 1.1 | 34.8% | 37.0% | estimated ± 6.7 pp, low confidence |
| Apodex 1.1 Mini | 34.8% | 37.0% | estimated ± 6.7 pp, low confidence |
| Celeris-1 | 0.0% | 32.5% | estimated ± 6.7 pp, low confidence |
| GPT-4o mini | 0.0% | 32.5% | estimated ± 6.7 pp, low confidence |
| GPT-5.6 Luna | 48.2% | 44.5% | estimated ± 6.7 pp, low confidence |
| Granite 4.2 30B | 3.9% | 32.5% | estimated ± 6.7 pp, low confidence |
| Granite 4.2 3B | 0.0% | 32.5% | estimated ± 6.7 pp, low confidence |
| Granite 4.2 8B | 0.0% | 32.5% | estimated ± 6.7 pp, low confidence |
| Grok 4.5 | 44.5% | 42.7% | estimated ± 6.7 pp, low confidence |
| Hy3 | 28.3% | 34.1% | estimated ± 6.7 pp, low confidence |
| Hy3 Preview | 35.8% | 37.5% | estimated ± 6.7 pp, low confidence |
| Inkling-Small | 31.2% | 35.2% | estimated ± 6.7 pp, low confidence |
| LFM2.5-2.6B | 0.0% | 32.5% | estimated ± 6.7 pp, low confidence |
| Ling 3.0 Flash FP8 | 22.4% | 32.9% | estimated ± 6.7 pp, low confidence |
| Ling 3.0 Flash VL | 33.2% | 36.1% | estimated ± 6.7 pp, low confidence |
| Ling 3.0 Tiny | 3.2% | 32.5% | estimated ± 6.7 pp, low confidence |
| Mercury 2.5 | 0.0% | 32.5% | estimated ± 6.7 pp, low confidence |
| MiniCPM5-2B | 10.5% | 32.5% | estimated ± 6.7 pp, low confidence |
| Muse Spark 1.2 | 48.9% | 44.8% | estimated ± 6.7 pp, low confidence |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 7.0% | 32.5% | estimated ± 6.7 pp, low confidence |
| Quasar 438B | 34.2% | 36.6% | estimated ± 6.7 pp, low confidence |
| Qwen3.8-Flash-Next | 56.6% | 47.0% | estimated ± 6.7 pp, low confidence |
| Qwen3.8 Max Preview | 58.6% | 47.3% | estimated ± 6.7 pp, low confidence |
| Solar Pro 4 | 30.5% | 34.9% | estimated ± 6.7 pp, low confidence |