Calibration

CharXiv w/o tools → MathVision w/ Python

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

MathVision w/ Python is estimated from CharXiv w/o tools with a Michaelis–Menten curve fitted on 5 models measured on both: y = 1.6144·x / (0.57336 + x), R² = 0.43, cross-validated error 1.3 pp. It is used for 1 estimate.

Estimated model	CharXiv w/o tools	MathVision w/ Python	Source
Gemini 3.8 Flash	86.2%	97.0%	estimated ± 1.3 pp, low confidence