Calibration

AA AutomationBench → JobBench

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

JobBench is estimated from AA AutomationBench with a offset logistic curve fitted on 5 models measured on both: y = 0.0071 + (0.5993 − 0.0071) / (1 + exp(−140.22·(x − 0.4805))), R² = 0.88, cross-validated error 9.1 pp. It is used for 1 estimate.

Estimated model	AA AutomationBench	JobBench	Source
Claude Haiku 5.5	35.4%	0.7%	estimated ± 9.1 pp, low confidence