With the deepening of economic globalisation and the rapid development of the Internet, open innovation has become the main innovation paradigm for enterprises to win a competition. There are three modes of open innovation: inbound, outbound, and coupling. Among them, coupling open innovation can better meet the innovation needs of enterprises. This paper takes the survey data of 421 enterprises in 10 cities in East China as samples, and uses a neural network model to explore the relationship between coupling open innovation and innovation performance, as well as the moderating effect of open innovation platforms. The study found that for enterprises in the start-up, growing and transition stages, the proportion of inbound open innovation positively affects innovation performance; for enterprises in the mature stage, the proportion of outbound open innovation positively affects innovation performance. At the same time, there are peak points for both. The breadth of platform openness does not significantly moderate coupling open innovation and innovation performance, while the depth of platform openness has a significant moderating effect on coupling open innovation and innovation performance.