Food systems are responsible for a third of global anthropogenic greenhouse gas emissions central to global warming and climate change. Increasing awareness of the environmental impact of food-centric emissions has led to the carbon footprint quantification of food products. However, food consumption is influenced by traditional dishes, the cultural capsules that encode traditional protocols for culinary preparations. Carbon footprint estimation of recipes will provide actionable insights into the environmental sustainability of culturally influenced patterns in recipe compositions. We implemented language models to integrate the carbon footprint data of food products with a gold standard repository of recipes to investigate the carbon load of recipes. Our results highlight the role of culture in shaping the carbon load of recipes. South American dishes have high carbon footprints owing to the high culinary utility of dairy and meat products. On the contrary, dishes from the Indian Subcontinent are the most environmentally friendly, with low carbon footprints due to the higher use of vegetables. While emphasizing the widely understood harms of animal-sourced ingredients, this article presents a nuanced perspective on the environmental impact of culturally influenced dietary practices. The study points to the need for efforts to develop techniques for plant-based and lab-grown meat to meet the protein requirements. We present an extensive repository of the estimated carbon footprints of recipes from across the world and enable their exploration through a user-friendly web server, SustainableFoodDB. Going forward, our study points to the need for high spatial resolution data on carbon footprints of ingredients.