Calibration
Gert Labs → GDPval-AA
GDPval-AA is estimated from Gert Labs with a inverse Michaelis–Menten curve fitted on 25 models measured on both: y = 0.41519·(x − 0.0000) / (0.0000 + 1.4071 − x), R² = 0.59, cross-validated error 8.8 pp. It is used for 9 estimates.
| Estimated model | Gert Labs | GDPval-AA | Source |
|---|---|---|---|
| DeepSeek V3.2 | 29.6% | 11.0% | estimated ± 8.8 pp, low confidence |
| Gemini 3.1 Flash-Lite | 38.5% | 15.6% | estimated ± 8.8 pp, medium confidence |
| GLM-5V-Turbo | 30.8% | 11.6% | estimated ± 8.8 pp, medium confidence |
| GPT-4.1 | 25.7% | 9.3% | estimated ± 8.8 pp, low confidence |
| Grok 4 | 42.3% | 17.9% | estimated ± 8.8 pp, medium confidence |
| Grok 4.1 Fast | 47.3% | 21.0% | estimated ± 8.8 pp, medium confidence |
| Grok 4.20 | 38.4% | 15.6% | estimated ± 8.8 pp, medium confidence |
| Grok Build 0.1 | 49.2% | 22.3% | estimated ± 8.8 pp, medium confidence |
| Qwen3 Max | 43.7% | 18.7% | estimated ± 8.8 pp, medium confidence |