Calibration
Claw-Eval → Gert Labs
Gert Labs is estimated from Claw-Eval with a Michaelis–Menten + offset curve fitted on 22 models measured on both: y = 0.0000 + 2.0000·x / (1.70099 + x), R² = 0.33, cross-validated error 9.1 pp. It is used for 7 estimates.
| Estimated model | Claw-Eval | Gert Labs | Source |
|---|---|---|---|
| GLM-5-Turbo | 55.8% | 49.4% | estimated ± 9.1 pp, low confidence |
| K-EXAONE 2.0 | 77.7% | 62.7% | estimated ± 9.1 pp, low confidence |
| LLaDA2.2-mini | 57.2% | 50.3% | estimated ± 9.1 pp, low confidence |
| MiMo-V2-Omni | 45.2% | 42.0% | estimated ± 9.1 pp, low confidence |
| Ornith-1.0-35B | 69.8% | 58.2% | estimated ± 9.1 pp, low confidence |
| Ornith-1.0-397B | 77.1% | 62.4% | estimated ± 9.1 pp, low confidence |
| Ornith-1.0-9B | 63.1% | 54.1% | estimated ± 9.1 pp, low confidence |