<p>Inter-firm patent transfers dominate overall patent transactions, yet relevant research remains limited, particularly regarding the essence of patent transfers, i.e., knowledge flow. This study provides a pilot exploration of the impact of firms’ multidimensional knowledge characteristics on inter-firm patent transfers, distinguishing between different industry stages and firm roles as suppliers or recipients. Knowledge characteristics are classified into the structural features of firms’ knowledge elements (degree and betweenness centrality) and the attributes of firms’ knowledge stock (scale and diversity). Using patent data from China’s new energy vehicle industry (2012–2021), we constructed patent transfer and knowledge combination networks to quantify these characteristics and applied exponential random graph models (ERGM) to analyze their stage-specific effects on firms’ tendencies to transfer or acquire patents. The results reveal differential impacts across industry stages. In the initial stage, firms with intermediary knowledge elements tend to engage in patent transfer. The degree centrality of firm’s knowledge elements negatively impacts firms’ patent transfer propensity, while knowledge diversity inhibits firms from acquiring patents. In the development stage, the scale of knowledge stock increase patent transfer likelihood whereas betweenness centrality of firm’s knowledge elements demonstrates inhibitory effects. Firms possessing high-degree centrality knowledge elements prefer to provide patents, while firms with diverse knowledge elements are inclined to acquire patents. This study offers insights for firms to optimize resource allocation and for policymakers to enhance the vitality of patent transfer markets.</p>