The green improvement of manufacturing high-temperature martensitic alloy steel using minimum quantity lubrication (MQL) involves an MQL device and a scientific method to assess the efficacy of the device application. This paper proposed a weight-variable machinability evaluation model based on multivariate heterogeneous data to compare the MQL process with the conventional machining lubrication processes. The proposed model comprises experimental-based intuitive evaluation and numerical machinability index (MI). The assessment model considers a multi-indicator and time-quality-cost-resource-environment (TQCRE) system. The MI is based on static-dynamic proximity, which is calculated according to indicator weights for subjective–objective combinations. The model was applied to a novel MQL system developed for manufacturing high-temperature martensitic alloys by performing milling experiments under four lubrication conditions. Experimental intuitive data indicated the superior feasibility of the MQL device, that is, the developed MQL method enhanced machining efficiency, ensured good machining quality, reduced tool wear by 17%, and cut forces by 7–14%. Moreover, the MQL process achieved the maximum machinability index regardless of the priority allocated to any indicator of machining quality, time, cost, resource loss, and environmental pollution. There was no degradation in machining via MQL under different environments, which validates the feasibility of the field application of the MQL method. The result is consistent with the experimental intuitive assessment, confirming the reasonableness and practical ability of the mathematical assessment model. This study may be considered as further validation of the multi-indicator machinability assessment for the green lubrication process. Future research on more cases might extend the explanations of the stability and hidden factors of the developed model.