To address the issues of low delivery efficiency, high carbon emissions, and elevated distribution costs in fresh cold chain logistics under the O2O mode, a comprehensive optimization model is constructed. This model considers business profitability, environmental benefits, fresh product characteristics, and customer timeliness requirements to optimize cold chain logistics distribution paths in the O2O mode. A three-stage algorithm comprising region partitioning, region adjustment, and path optimization is designed to solve this NP-hard problem. In the first stage, delivery regions are initially partitioned using the K-Means clustering algorithm. In the second stage, load balance indicators are introduced to adjust the delivery regions. In the third stage, a genetic algorithm is designed to solve the optimal delivery paths for each region. Empirical results show that after region partitioning and adjustment, delivery distance decreases by 15.39%, and early or delayed delivery rates decrease by 59.08%, significantly improving delivery efficiency. Furthermore, considering carbon emissions, a single delivery center can reduce daily emissions by 23.13%, resulting in a 22.98% cost saving for businesses. These results validate the scientific and practical effectiveness of the model and provide valuable references for similar logistics enterprises.