Calibration

JobBench → ApprenticeBench

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

ApprenticeBench is estimated from JobBench with a Hill curve fitted on 6 models measured on both: y = 0.0000 + (0.2185 − 0.0000)·x^6.00 / (0.40491^6.00 + x^6.00), R² = 0.66, cross-validated error 5.6 pp. It is used for 5 estimates.

Estimated model	JobBench	ApprenticeBench	Source
Claude 4.1 Opus	21.9%	0.5%	estimated ± 5.6 pp, low confidence
Claude Haiku 4.5	16.0%	0.1%	estimated ± 5.6 pp, low confidence
GPT-5 (high)	8.5%	0.0%	estimated ± 5.6 pp, low confidence
Qwen3.5 Plus	18.5%	0.2%	estimated ± 5.6 pp, low confidence
Qwen3.8-Flash-Next	55.7%	19.0%	estimated ± 5.6 pp, low confidence