Based on balanced panel data of 284 prefecture-level cities in China from 2007 to 2021, this paper employs the SBM-DEA model, which is based on a two-stage value chain perspective, to measure green technology efficiency (GTE) and green result efficiency (GRE). The study utilizes fixed effect panel models (FE), dynamic threshold models, and spatial Durbin models (SDM) to explore the mechanisms and effects of green finance on GTE and GRE. The results indicate that green finance significantly improves both GTE and GRE: for every 1% increase in green finance, GTE and GRE increase by 0.203% and 0.258%, respectively. These findings remain robust under various tests for endogeneity and sensitivity. Green finance primarily enhances GTE by alleviating financing constraints and stimulating social consumption. Additionally, formal and informal environmental regulations significantly strengthen the impact of green finance on GTE and GRE. When using green finance as a threshold variable, the analysis reveals a nonlinear increasing effect on both GTE and GRE. Furthermore, green finance exhibits spatial spillover effects on GTE and GRE and demonstrates significant spatio-temporal heterogeneity. This study contributes to understanding and promoting the application of green finance in supporting rural revitalization and smart agricultural energy solutions.