benchgap
Calibration

GDPval-AA → Toolathlon-Verified

Toolathlon-Verified is estimated from GDPval-AA with a Michaelis–Menten curve fitted on 13 models measured on both: y = 1.2437·x / (0.37828 + x), R² = 0.93, cross-validated error 2.4 pp. It is used for 32 estimates.

Estimated modelGDPval-AAToolathlon-VerifiedSource
A.X K223.2%47.3%estimated ± 2.4 pp, medium confidence
Apodex 1.134.8%59.6%estimated ± 2.4 pp, high confidence
Apodex 1.1 Mini34.8%59.6%estimated ± 2.4 pp, high confidence
DeepSeek V3 03240.0%0.0%estimated ± 2.4 pp, medium confidence
Gemma 4 12B0.0%0.0%estimated ± 2.4 pp, medium confidence
Gemma 4 26B A4B3.4%10.3%estimated ± 2.4 pp, medium confidence
Gemma 4 E2B0.0%0.0%estimated ± 2.4 pp, medium confidence
Gemma 4 E4B0.0%0.0%estimated ± 2.4 pp, medium confidence
GLM-4.725.8%50.4%estimated ± 2.4 pp, medium confidence
GPT-4.1 mini0.0%0.0%estimated ± 2.4 pp, medium confidence
GPT-4.1 nano0.0%0.0%estimated ± 2.4 pp, medium confidence
GPT-4o0.0%0.0%estimated ± 2.4 pp, medium confidence
GPT-4o mini0.0%0.0%estimated ± 2.4 pp, medium confidence
GPT-5.116.5%37.8%estimated ± 2.4 pp, medium confidence
Granite 4.2 30B3.9%11.6%estimated ± 2.4 pp, medium confidence
Granite 4.2 3B0.0%0.0%estimated ± 2.4 pp, medium confidence
K-Exaone0.0%0.0%estimated ± 2.4 pp, medium confidence
Kimi K2.5 (Reasoning)17.2%38.9%estimated ± 2.4 pp, medium confidence
LFM2.5-2.6B0.0%0.0%estimated ± 2.4 pp, medium confidence
Ling 2.6 Flash0.0%0.0%estimated ± 2.4 pp, medium confidence
Ling 3.0 Flash VL33.2%58.1%estimated ± 2.4 pp, high confidence
Ling 3.0 Tiny3.2%9.7%estimated ± 2.4 pp, medium confidence
Mercury 2.50.0%0.0%estimated ± 2.4 pp, medium confidence
MiMo-V2-Flash6.6%18.5%estimated ± 2.4 pp, medium confidence
MiniCPM5-2B10.5%27.0%estimated ± 2.4 pp, medium confidence
Nemotron 3 Nano Omni 30B A3B0.0%0.0%estimated ± 2.4 pp, medium confidence
North Mini Code0.0%0.0%estimated ± 2.4 pp, medium confidence
Qwen3.6 Plus24.7%49.1%estimated ± 2.4 pp, medium confidence
Solar Pro 30.0%0.0%estimated ± 2.4 pp, medium confidence
Solar Pro 430.5%55.5%estimated ± 2.4 pp, medium confidence
Step 3.7 Flash25.8%50.4%estimated ± 2.4 pp, medium confidence
Ultravox v0.6 Llama 3.3 70B0.0%0.0%estimated ± 2.4 pp, medium confidence