benchgap
Calibration

DAYJOB Healthcare → StealthMark

StealthMark is estimated from DAYJOB Healthcare with a Michaelis–Menten + offset curve fitted on 12 models measured on both: y = 0.3545 + 0.4742·x / (0.05068 + x), R² = 0.90, cross-validated error 5.4 pp. It is used for 0 estimates.

Fitted on

ModelDAYJOB HealthcareStealthMark
Claude Opus 5.525.2%72.6%
Claude Sonnet 5.516.0%70.2%
Gemini 3.8 Flash1.2%47.9%
Gemini 4 Argon11.6%67.4%
GPT-5.50.0%36.5%
GPT-5.6 Luna1.2%37.8%
GPT-5.6 Sol1.2%53.7%
GPT-5.6 Terra2.0%45.8%
GPT-6.1 Sol10.8%67.8%
GPT-6 Astra11.2%74.6%
GPT-6 Luna1.2%39.1%
GPT-6 Sol2.8%52.0%