Calibration
Claw-Eval → GDPval-AA
GDPval-AA is estimated from Claw-Eval with a inverse Michaelis–Menten curve fitted on 17 models measured on both: y = 0.52105·(x − 0.0000) / (0.0000 + 2.0000 − x), R² = 0.32, cross-validated error 9.4 pp. It is used for 6 estimates.
| Estimated model | Claw-Eval | GDPval-AA | Source |
|---|---|---|---|
| GLM-5-Turbo | 55.8% | 20.2% | estimated ± 9.4 pp, low confidence |
| K-EXAONE 2.0 | 77.7% | 33.1% | estimated ± 9.4 pp, low confidence |
| MiMo-V2-Omni | 45.2% | 15.2% | estimated ± 9.4 pp, low confidence |
| Ornith-1.0-35B | 69.8% | 27.9% | estimated ± 9.4 pp, low confidence |
| Ornith-1.0-397B | 77.1% | 32.7% | estimated ± 9.4 pp, low confidence |
| Ornith-1.0-9B | 63.1% | 24.0% | estimated ± 9.4 pp, low confidence |