Calibration
GDPval-AA → WideResearch
WideResearch is estimated from GDPval-AA with a Michaelis–Menten + offset curve fitted on 5 models measured on both: y = 0.4159 + 2.0000·x / (1.22036 + x), R² = 0.38, cross-validated error 9.1 pp. It is used for 60 estimates.
| Estimated model | GDPval-AA | WideResearch | Source |
|---|---|---|---|
| A.X K2 | 23.2% | 73.5% | estimated ± 9.1 pp, low confidence |
| Apodex 1.1 Mini | 34.8% | 86.0% | estimated ± 9.1 pp, low confidence |
| Celeris-1 | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Claude Fable 5 | 55.6% | 100.0% | estimated ± 9.1 pp, low confidence |
| Command A+ | 0.9% | 43.1% | estimated ± 9.1 pp, low confidence |
| DeepSeek V3 | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| DeepSeek V3 0324 | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Gemini 3.5 Flash-Lite | 24.3% | 74.8% | estimated ± 9.1 pp, low confidence |
| Gemini 3.6 Flash | 39.3% | 90.3% | estimated ± 9.1 pp, low confidence |
| Gemini 3.7 Flash | 44.6% | 95.1% | estimated ± 9.1 pp, low confidence |
| Gemini 3.8 Flash | 46.8% | 97.0% | estimated ± 9.1 pp, low confidence |
| Gemini 4 Argon | 56.3% | 100.0% | estimated ± 9.1 pp, low confidence |
| Gemma 3 27B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Gemma 4 12B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Gemma 4 26B A4B | 3.4% | 47.0% | estimated ± 9.1 pp, low confidence |
| Gemma 4 E2B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Gemma 4 E4B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| GPT-4.1 mini | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| GPT-4.1 nano | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| GPT-4o | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| GPT-4o mini | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| GPT-6.1 Sol | 53.8% | 100.0% | estimated ± 9.1 pp, low confidence |
| GPT-6 Luna | 46.9% | 97.1% | estimated ± 9.1 pp, low confidence |
| GPT-6 Sol | 50.5% | 100.0% | estimated ± 9.1 pp, low confidence |
| GPT-OSS 20B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Granite 4.2 30B | 3.9% | 47.8% | estimated ± 9.1 pp, low confidence |
| Granite 4.2 3B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Granite 4.2 8B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Grok 4.5 | 44.5% | 95.0% | estimated ± 9.1 pp, low confidence |
| Grok 4.6 | 56.1% | 100.0% | estimated ± 9.1 pp, low confidence |
| Grok 4.7 | 60.8% | 100.0% | estimated ± 9.1 pp, low confidence |
| Hy3 | 28.3% | 79.2% | estimated ± 9.1 pp, low confidence |
| K-Exaone | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Kimi K2.7 Code | 27.0% | 77.8% | estimated ± 9.1 pp, low confidence |
| LFM2.5-2.6B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Ling 2.6 Flash | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Ling 3.0 Flash FP8 | 22.4% | 72.6% | estimated ± 9.1 pp, low confidence |
| Ling 3.0 Flash VL | 33.2% | 84.4% | estimated ± 9.1 pp, low confidence |
| Ling 3.0 Tiny | 3.2% | 46.7% | estimated ± 9.1 pp, low confidence |
| Llama 4 Maverick | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Llama 4 Scout | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Mercury 2.5 | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| MiMo-V2-Flash | 6.6% | 51.9% | estimated ± 9.1 pp, low confidence |
| MiniCPM5-2B | 10.5% | 57.4% | estimated ± 9.1 pp, low confidence |
| Mistral Large 3 | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Mistral Large 4 | 46.2% | 96.5% | estimated ± 9.1 pp, low confidence |
| Mistral Small 4 | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Mistral Small 4 (Reasoning) | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Muse Spark | 25.1% | 75.7% | estimated ± 9.1 pp, low confidence |
| Muse Spark 1.2 | 48.9% | 98.8% | estimated ± 9.1 pp, low confidence |
| Nemotron 3 Nano 30B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Nemotron 3 Nano Omni 30B A3B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Nemotron 3 Super 100B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| North Mini Code | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Quasar 438B | 34.2% | 85.4% | estimated ± 9.1 pp, low confidence |
| Qwen3.7 Plus | 13.5% | 61.5% | estimated ± 9.1 pp, low confidence |
| Qwen3.8 Max Preview | 58.6% | 100.0% | estimated ± 9.1 pp, low confidence |
| Solar Pro 3 | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Trinity-Large-Preview | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |
| Ultravox v0.6 Llama 3.3 70B | 0.0% | 41.6% | estimated ± 9.1 pp, low confidence |