Many metal alloys are strengthened by controlling precipitation to achieve an optimal peak-aged condition where the strength-limiting processes of precipitate shearing and Orowan looping are thought to be comparable. Qualitative models have long captured the basic mechanisms but realistic predictions have been challenging due to both the lack of accurate material parameters and an inability to quantitatively validate the models. Here, dislocation/precipitate interaction mechanisms are studied in Al-6xxx Al–Mg–Si alloys using atomistic simulations in tandem with a near-chemically-accurate Al–Mg–Si neural network interatomic potentials. Results show that a given precipitate can exhibit shearing or looping depending on the relative orientation of the precipitate and dislocation, as influenced by the matrix and precipitate coherency stresses, direction-dependence of precipitate shearing energies, and dislocation line tension. Analytic models for shearing and calibrated discrete dislocation models of looping accurately capture the trends and magnitudes of strengthening in most cases. Good agreement with experiments is then approached by using the theories together with the more-accurate first-principles material properties. The combination of theories and simulations demonstrated here constitutes a path for understanding and predicting the role of chemistry and microstructure on alloy strength that can be applied in many different alloys.