Building 3D printing objects are usually divided manually in blocks and printed parallel. The model division directly affects the printing efficiency and overall appearance. In order to improve the dividing quality, a task al-location algorithm for dual devices cooperation is proposed. Firstly, the original model is roughly sliced according to its support relationship. And block features are identified to avoid dividing of support columns, short walls or decorations which are not fit for parting. Then, to improve printing efficiency, geometric aggregation index and task balance index are set, and a multi-objective optimization model of printing task allocation is established. Finally, based on the solution of optimization problem, the original model is cut into two STLs to output. The method focuses on dual devices pattern, but can extend for multi-devices from parameters adjustment. This paper compared the results of the proposed method with that of common geometric equalization and partition by wall connection. It showed that this method is superior in task balance, which can promote the efficiency of building 3D printing, and the key features of original model are retained from dividing.