Conflicts arise when different enterprises have different value systems for a given issue, typically accompanied by complexity and uncertainty. How to find equilibrium and track evolution paths in conflicts has become a critical issue. However, classical game theory proves unreliable and difficult to apply in many conflicts. In this paper, the graph model for conflict resolution (GMCR) and Markov chain are utilized to explore the inner mechanism and possible evolution of conflict with a specific case of automobile enterprises. Firstly, the modeling and analysis of GMCR is briefly introduced. Secondly, the conflict of automobile enterprises is modeled using large language model (LLM) and expert knowledge to acquire decision-makers (DMs), options, feasible states, state transitions and preference information. Subsequently, equilibrium states are determined by stability analysis of the conflict. Finally, the evolution paths of the conflict is analyzed based on Markov process. The result demonstrates that the model is highly consistent with reality, offering valuable insights into conflict resolution.