Under the current operation mode dominated by on-demand mobility services: there is a tendency for an imbalance in the supply and demand matching between drivers and passengers during peak hours, which results in a high workload for drivers; during non peak hours, there is a tendency for mobility services to exceed demand, which increases the non productive operating hours of drivers. In this background, some drivers choose to sacrifice rest time and remain in a state of fatigue driving in order to increase operating income. How to innovate the sustainable operation mode of the mobility platform is a management problem. This innovated sustainable operation mode should achieve the platform's social responsibility as the leading factor, regulate the workload of drivers, and ensure the economic revenue of mobility services. Hence, this research proposed a sustainable service operation model for digital mobility platforms that considers the sustainable development of driver workload and mobility service revenue. It is led by a reservation mobility service and supplemented by an instant mobility service. A two-stage 0-1 mixed integer programming model was constructed: in stage-I, the driver's mobility service route was customized for scheduled passengers; in stage-II, the phenomenon of passenger absenteeism was considered, achieving on-demand mobility demand matching and optimizing the service hours. Given the different complexities of the two models, the stage-I model was solved by a column generation-based heuristic algorithm (CG-BH) and a hybrid meta-heuristic algorithm based on adaptive large neighborhood search and simulated annealing (ALNS-SA). The stage-II model was solved using Gurobi. The rationality of the two-stage model was verified by numerical experiments. A comparison was adopted among the Gurobi, CG-BH and ALNS-SA algorithms, which highlighted the superiority of the ALNS-SA algorithm in balancing solution quality and computational efficiency. In the sensitivity analysis, the influence of drivers’ number on the economy revenue, cumulative working hours and scheduling fairness was quantified. The practicality of the proposed method was also verified. Under the proposed method: the driver's continuous driving, rest, and cumulative working hours were controlled within a reasonable range; the routes with different workload can be matched with heterogeneity drivers; according to the travel attributes of defaulting passengers, an economic revenue function was fitted for real-time mobility service, and it revealed that the economic revenue generated by a high-quality instant mobility service can be higher than the economic losses incurred by the missed reservation mobility service. This research is beneficial to the physical and mental health of drivers, and provides theoretical reference and technical support for the social responsibility of the mobility platform and the sustainable development of economic revenue.