Accurate prediction of drivetrain performance and energy efficiency in electric vehicles (EVs) remains a challenge due to the limitations of conventional chassis dynamometer testing, which often overlooks dynamic vehicle-roller interactions, suspension responses, and regenerative braking effects. This paper presents a novel digital twin-enabled testing framework using a compact, two-roller chassis dynamometer integrated with real-time multibody dynamics and electromechanical co-simulation. The physical test bench is coupled with a virtual model of the EV, including suspension geometry, rear-to-front axle load distribution, tire–roller contact mechanics, and motor-inverter dynamics. By enabling bi-directional data exchange between the physical and virtual domains, the system emulates real-world load shifts, torque demand variations, and energy recovery under standardized drive cycles. This integrated approach enhances simulation fidelity, supports certification-compliant testing, and