DeepViscosity is an ensemble deep-learning artificial neural network model that predicts the high-concentration (150 mg/mL) viscosity class (Low <= 20 cP vs. High > 20 cP) of monoclonal antibodies (mAbs) from heavy- and light-chain variable-region amino-acid sequences. It first numbers the input sequences with ANARCI/ANARCII (IMGT scheme), runs the DeepSP CNN surrogate to produce 30 spatial property descriptors (SAP_pos, SCM_neg, SCM_pos for CDRH1/2/3, CDRL1/2/3, CDR, Hv, Lv, Fv), then averages predictions across an ensemble of 102 ANN classifiers (LOGO cross-validation seeds) to output a binary class label, the mean class probability, and the across-ensemble standard deviation.