In order to solve voltage fluctuation and off-limit problems caused by the large-scale grid connection of new energy stations such as wind power and photovoltaic, as well as the problem of traditional reactive power and voltage calculation methods easily falling into local optima, a new optimization model based on improved firefly algorithm (IFA) for the regional power grid is proposed. In this model, the sum of standardized active power losses, node voltage deviations, and reactive output deviations is used as the optimization objective. Meanwhile, regional power flow calculation, terminal voltage and reactive capacity of thermal power units, transformer ratios, active power of new energy stations and reactive power of static var compensators are used as constraints that needs to be met. Then, by using an IFA to solve the optimization model, the algorithm’s computational accuracy and global optimization ability are increased by introducing new step size and position update formulas. Finally, the effectiveness of the presented model is validated using an IEEE-30 bus system. The results indicate that the proposed optimization model can reduce the active power loss of the system, improve the reactive power distribution of the regional power grid, resist the voltage fluctuation of the system, and improve the power quality of the regional power grid. IFA can effectively improve the calculation accuracy and speed, and improve the global optimization ability.