Calibration

C-Eval → SuperGPQA

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

SuperGPQA is estimated from C-Eval with a Hill curve fitted on 5 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^6.00 / (0.87618^6.00 + x^6.00), R² = 0.87, cross-validated error 1.2 pp. It is used for 1 estimate.

Estimated model	C-Eval	SuperGPQA	Source
LongCat-Flash-Lite-Sparse	85.8%	56.1%	estimated ± 1.2 pp, low confidence