benchgap
Calibration

NOVA-63 → SWE Multilingual

SWE Multilingual is estimated from NOVA-63 with a Michaelis–Menten + offset curve fitted on 6 models measured on both: y = 0.0000 + 2.0000·x / (0.94818 + x), R² = 0.38, cross-validated error 2.1 pp. It is used for 1 estimate.

Estimated modelNOVA-63SWE MultilingualSource
Qwen3.5 397B59.1%76.8%estimated ± 2.1 pp, low confidence