Abstract: As a vital component of traditional Chinese culture, Sichuan liquor culture (SLC) carries rich historical and cultural connotations. This study examines the introductory texts from the official websites of Sichuan liquor enterprises. Using the transitivity system of systemic functional linguistics, it compares how DeepSeek and ChatGPT, the large language models, construct experiential meaning in their English translations. Based on a self-built corpus of 20 English translations from 10 enterprise websites, this study conducts a quantitative analysis of six transitivity processes (material, mental, relational, behavioral, verbal, and existential). The functional differences among these processes are further explored through qualitative methods. The findings reveal that translations from both models are predominantly characterized by material and relational processes. These processes highlight the enterprises’ "action-result" orientation and identity-defining function, thereby forming the core framework of corporate introductions. Other processes appear at low frequencies, serving as rhetorical supplements. DeepSeek prefers dynamic expression and conceptual restructuring, often omitting the subject. In contrast, ChatGPT tends to use static statements, preserving the subject and the original informational structure. These contrasting patterns fundamentally reflect two distinct cognitive modeling tendencies. The former adopts an institutional discourse model, constructing institutionalized reality through systematic transformation of process types, while the latter’s translations align more closely with everyday discourse patterns, preserving the transitivity diversity of the experiential world. This study offers linguistically informed optimization strategies for the international communication of SLC and reveals the strengths and limitations of large language models in translating specialized texts. Furthermore, it emphasizes the necessity of human-machine collaboration for the precise communication of culture.