Calibration
GDPval-AA → Gert Labs
Gert Labs is estimated from GDPval-AA with a Hill curve fitted on 25 models measured on both: y = 0.3597 + (1.2000 − 0.3597)·x^2.11 / (0.56360^2.11 + x^2.11), R² = 0.60, cross-validated error 8.9 pp. It is used for 31 estimates.
| Estimated model | GDPval-AA | Gert Labs | Source |
|---|---|---|---|
| A.X K2 | 23.2% | 47.2% | estimated ± 8.9 pp, medium confidence |
| Apodex 1.1 | 34.8% | 58.3% | estimated ± 8.9 pp, medium confidence |
| Apodex 1.1 Mini | 34.8% | 58.3% | estimated ± 8.9 pp, medium confidence |
| Claude Sonnet 5.5 | 67.0% | 85.6% | estimated ± 8.9 pp, low confidence |
| DeepSeek V3 0324 | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Gemma 4 12B | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Gemma 4 26B A4B | 3.4% | 36.2% | estimated ± 8.9 pp, medium confidence |
| Gemma 4 E2B | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Gemma 4 E4B | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| GPT-4.1 mini | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| GPT-4.1 nano | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| GPT-4o | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| GPT-4o mini | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| GPT-6.1 Sol | 53.8% | 75.9% | estimated ± 8.9 pp, low confidence |
| GPT-6 Luna | 46.9% | 70.0% | estimated ± 8.9 pp, medium confidence |
| Granite 4.2 30B | 3.9% | 36.3% | estimated ± 8.9 pp, medium confidence |
| Granite 4.2 3B | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Grok 4.7 | 60.8% | 81.3% | estimated ± 8.9 pp, low confidence |
| K-Exaone | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| LFM2.5-2.6B | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Ling 2.6 Flash | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Ling 3.0 Flash VL | 33.2% | 56.7% | estimated ± 8.9 pp, medium confidence |
| Ling 3.0 Tiny | 3.2% | 36.2% | estimated ± 8.9 pp, medium confidence |
| Mercury 2.5 | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| MiMo-V2-Flash | 6.6% | 36.9% | estimated ± 8.9 pp, medium confidence |
| MiniCPM5-2B | 10.5% | 38.3% | estimated ± 8.9 pp, medium confidence |
| Mistral Large 4 | 46.2% | 69.3% | estimated ± 8.9 pp, medium confidence |
| Nemotron 3 Nano Omni 30B A3B | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| North Mini Code | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Solar Pro 3 | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |
| Ultravox v0.6 Llama 3.3 70B | 0.0% | 36.0% | estimated ± 8.9 pp, medium confidence |