<p>Mass production of lithium-ion batteries for electric vehicles will result in a large volume of retired batteries requiring recycling or reuse for less demanding applications. In the sorting process of decommissioned batteries for reuse, methods based on time consuming tests become unsuitable when a large number of batteries are processed. In this study, a new method for rapid sorting of retired batteries, based on parameters obtained from Electrochemical Impedance Spectroscopy (EIS), is proposed. An EIS capacity prediction model, based on Extreme Gradient Boosting optimized by Sparrow Search Algorithm (SSA-XGBoost), has been trained to rapidly forecast battery capacity. Distribution of Relaxation Times (DRT) and Distribution of Capacitive Times (DCT) were used to extract the relevant features from the medium-high and low frequencies of EIS, respectively, to construct a multi-dimensional features set. Deep Embedded Clustering (DEC), using the obtained features set, was employed to rapidly cluster a batch of retired ternary lithium batteries, stored at low state of charge. Battery performance consistency has been evaluated through comprehensive evaluation metrics, obtained from experiments employing differential capacity (ΔQ(V)) and incremental capacity (IC) curves. The overall consistency of the batteries improved by 51.07% in the comprehensive consistency indicator, compared to unsorted batteries. Moreover, the proposed method, compared to a method using only features obtained by Electrochemical Model (ECM) fitting, and another using only DRT based features, showed improvements of 36.87% and 37.06%, respectively, representing an interesting and promising way to pursue and promote the reuse of lithium-ion batteries.</p>