In response to the application demand for multi-objective reactive power optimization in power systems containing wind farms and the issues of local easy convergence and poor global search ability of optimization algorithms, a vector fireworks algorithm is proposed for multi-objective reactive power optimization. Firstly, the explosion method of the fireworks algorithm is modified, and the dimensions of the fireworks explosion are correlated to avoid falling into a local optimal solution during the iteration process and to enhance the global search capability of the algorithm. Secondly, a novel congestion selection strategy is introduced to accelerate the convergence speed of the algorithm. Lastly, a multi-objective reactive power optimization model for power systems containing wind farms is established, with the objectives of minimizing active network loss and voltage deviation. The model is solved as an example of an IEEE 14-node power network with a wind farm. The results indicate that the proposed algorithm mitigates the issue of local convergence, enhances the global search capability of the algorithm, and can obtain effective reactive power optimization solutions under different optimization requirements.