benchgap
Calibration

ERQA → MedXpertQA (MM)

MedXpertQA (MM) is estimated from ERQA with a Hill curve fitted on 7 models measured on both: y = 0.1360 + (0.8154 − 0.1360)·x^6.00 / (0.43298^6.00 + x^6.00), R² = 0.75, cross-validated error 4.2 pp. It is used for 3 estimates.

Estimated modelERQAMedXpertQA (MM)Source
Qwen3.8-27B65.5%76.3%estimated ± 4.2 pp, medium confidence
Qwen3.8-Flash-Next72.3%78.5%estimated ± 4.2 pp, medium confidence
Qwen3.8-Omni-Flash71.0%78.2%estimated ± 4.2 pp, medium confidence