With the significant growth of IoT devices and traffic, as well as the large-scale deployment of new applications, the demand for computing power is constantly increasing. Edge computing is an excellent solution. However, edge computing nodes are unable to meet these demands because of the hardware infrastructure’s restricted computing power. To address this issue, we design a cooperative offloading architecture to offload tasks to cooperative terminals through device-to-device (D2D) communication. And we propose a D2D cooperative offloading strategy based on the Kepler Optimization Algorithm (D2DCOBKOA). We formulate an objective function composed of latency and energy consumption under quality of service (QoS) constraints. We consider of- floading selection, computing power, and communication capability as decision variables, and employ KOA for multi-objective function optimization to reduce the user’s overall cost. Experimental results indicate that, compared to the existing strategy, the proposed strategy achieves a 40% reduction in user costs.