benchgap
Calibration

ScreenSpot Pro → MathVision

MathVision is estimated from ScreenSpot Pro with a Michaelis–Menten curve fitted on 6 models measured on both: y = 1.3359·x / (0.36038 + x), R² = 0.93, cross-validated error 2.4 pp. It is used for 7 estimates.

Estimated modelScreenSpot ProMathVisionSource
GPT-6 Astra92.7%96.2%estimated ± 2.4 pp, low confidence
Holo2-235B-A22B70.6%88.4%estimated ± 2.4 pp, medium confidence
Holo2-30B-A3B66.1%86.5%estimated ± 2.4 pp, medium confidence
Holo2-4B57.2%82.0%estimated ± 2.4 pp, medium confidence
Holo2-8B58.9%82.9%estimated ± 2.4 pp, medium confidence
Muse Glimmer 30B75.4%90.4%estimated ± 2.4 pp, medium confidence
Nemotron 3 Nano Omni 30B A3B57.8%82.3%estimated ± 2.4 pp, medium confidence