AMPA receptors (AMPARs) mediate fast excitatory neurotransmission and participate in memory formation. Synaptic strength is proportional to the number of AMPARs in the postsynaptic membrane. Previous models of AMPA receptor trafficking can explain their fast, brief incorporation in dendritic spines after LTP induction but offer limited insights into long-lasting changes in AMPAR content at potentiated spines[1,2]. Hence, a unified modeling framework is necessary to understand persistent synaptic changes in AMPAR numbers. Molecular trafficking, such as diffusion, active transport, endo/exocytosis, and protein synthesis/degradation, underlie the copy number of AMPARs in spines at different dendritic locations [3]. We blended computational modeling and experimental work to explain the kinetics of trafficking steps necessary to explain the experimentally observed localizations of AMPARs and the response of different AMPAR subtypes to LTP induction. Notably, we estimate the global transport (diffusion and drift constants) rates by fitting our model to the fluorescent intensity profile of endogenous GluA2, measured along 100 µm long dendritic stretches. Our results show that active transport is critical to explain the observed protein distribution. Then, we conducted fluorescent labeling of endogenous GluA2 and, using our model, estimated that the GluA2 endocytosis rate exceeded the exocytosis rate, with their ratio being constant along the dendrites. We also found the synaptic enrichment of GluA2, as the fluorescence intensity ratio between the spine and shaft surface, was ~ 1 and increased along the dendrites. A recent study using metabolic labeling, ribosomal profiling, and micro-dissection found that auxiliary subunits of AMPARs show distinct patterns of translation localization [4]. Using similar methods, we found that the auxiliary subunit CNIH2 showed significant local translation, further enhanced by LTP induction. Disrupting CNIH-2 trafficking reduced surface trafficking of nascent GluA2 subunits, but not GluA1 subunits, of AMPARs. Next, we used CNIH2’s selectivity toward trafficking GluA2 subunit in our model as a potential mechanism to explain the difference in temporal response of two main sub-types of AMPARs, namely the calcium-impermeable (GluA2-containing) and calcium-permeable (GluA2-lacking) AMPARs, upon plasticity induction. Our work brings together the two timescales of AMPAR plasticity: one, fast and transient, and second, slow and prolonged.