Narrowing the urban-rural income gap and promoting the sustained growth of urban and rural residents’ income are core issues in advancing common prosperity. Digital inclusive finance, relying on mobile internet and big data credit scoring, provides a new path to overcome the geographical constraints and entry barriers of traditional finance. Based on panel data of prefecture-level cities in China from 2011 to 2024, this paper uses a dual machine learning (DML) partially linear model to identify its causal effect on the per capita disposable income of urban and rural residents. The study found that for every 1-unit increase in the digital inclusive finance index, urban and rural residents’ income increased by approximately RMB 0.274 per person after controlling for the double fixed effects (using the 2011 average as the base period). This conclusion remained robust even after changing the machine learning algorithm, adjusting the cross-fitting split ratio, introducing the time × city and province high-dimensional fixed effects, shortening the sample window, using the city clustering standard error, and adding control variables. Mechanism testing showed that entrepreneurial activity, industrial structure upgrading, and education expenditure levels constituted three significant mediating paths. In terms of heterogeneity, the income-increasing effect was stronger in the eastern, western, and southern regions, as well as in non-old industrial cities, large cities, central cities, and non-resource-based cities, while it was insignificant or weak in the northeast and old industrial cities. Further analysis showed that digital inclusive finance significantly suppressed the Theil index and the urban-rural income ratio. This study provides reusable methods and empirical support for differentiated digital finance policies.