The INFOGEST semi-dynamic digestion model more closely aligns the kinetics of nutrient digestion with structural changes in the food matrix during gastric digestion, which can significantly influence polyphenol bioaccessibility. In this study, the static and semi-dynamic INFOGEST models were compared to assess polyphenol bioaccessibility across various matrix scenarios, using different apple fractions. Each digesta, regardless of the model used, underwent re-solubilization, centrifugal filtration, and UHPLC-ESI-QTOF-MS/MS analysis to approximate transepithelial absorption and facilitate untargeted polyphenol screening and semi-quantification. The semi-dynamic model was initially optimized using whole apple. Overhead stirring with a paddle led to greater browning and degradation of phenolic acids and dihydrochalcones than magnetic stirring, the latter showing bolus stratification and closer physiological conditions for oxygenation and intragastric chyme homogenization. The suitability of a 2 kcal/min gastric emptying rate was tested with pomace, resulting in 8.25 min total gastric emptying time due to low caloric content. Compared to the gastric emptying time of whole apple (139.5 min), the caloric-driven emptying of pomace produced similar polyphenol bioaccessibility but a three-fold higher coefficient of variation (19.5 % vs. 69.4 %). Finally, using several apple fractions, the semi-dynamic setup with magnetic stirring and a fixed gastric emptying rate of 139.5 min showed greater extraction of hydroxybenzoic acids and dihydrochalcones from apple and of hydroxybenzoic and hydroxycinnamic acids from pomace than the static model. However, flavanols in juice degraded more extensively under semi-dynamic conditions. Minimal differences were observed between models for an apple polyphenol extract, indicating that in the absence of matrix, the static setup might be preferred.