This dataset contains the calibrated GOSAT XCO2 retrievals against the OCO-2 XCO2 benchmark based on the machine learning models, which is a harmonized dataset that aligns more congruently with OCO-2 XCO2 retrievals. Such integration is pivotal for enhancing observational constraints in carbon flux inversions. The source of Level 2 products for both satellites is NASA's Goddard Earth Science Data and Information Services Center (GES-DISC), accessible at https://disc.gsfc.nasa.gov/. The dataset is shared in the format of the ACOS_L2_Lite_FP product (GOSAT Level 2 XCO2 product) with retrievals of poor quality having been filtered out. The latest version of the dataset features modifications in the model adjustment process and also includes recalibrated uncertainties for each observation resulting from model corrections. The data time range is from April 2009 to June 2020, and the folders are divided by year.