In order to solve the complexity problem caused by the conflict between the information capacity and the user's cognitive capacity in multimodal information interfaces, the work aims to introduce visual clue design at the cognitive level and explore the effect of integrating visual clues on optimizing the complexity of multimodal information interfaces. The eye-tracking technology was used to collect eye movement data and behavioral data from 35 subjects under different visual cue forms and three interface complexity levels of high, medium and low. Eye movement indicators such as Total Visit Duration, Saccade Count, and Scanning Paths in the AOI were extracted for comprehensive analysis. Both visual cues and interface complexity in multimodal information interfaces had a significant impact on subjects' behavioral response indicators and eye movement indicators, and the visual search performance of dynamic cues and surface cues was better. The integration of visual cues can effectively guide the subjects' visual movement at the same level of complexity, helping them to find the target information more quickly and effectively, which can not only improve the cognitive performance of multimodal information interfaces, but also help reduce the cognitive difficulty in high-complexity interfaces. The results of this study provide a basis for optimizing the complexity of multimodal information interfaces.