4G LTE with Massive MIMO as fast as 5G 3.5 GHz
Qualcomm's simulations are respected and the most accurate predictions available. They are consistent with other data. I've come to similar conclusions from expectations of the technology. One possible update to come: The results from Verizon's tests at 28 GHz are very good so far, and may raise performance estimates for millimeter wave.
Topologies vary. There's no way to predict perfectly the number of users, distances, window and tree interference, and trees, window glass, trucks passing by and other crucial factors. Simulating wireless networks is an art, not a science, and requires great skill.
Good models are crucial for planning actual deployments. I've used the actual Alcatel and Ericsson deployment models, which are very rich. Everyone in the business has learned models will never get everything right, but are an essential starting point.
Large wireless networks take two years to get started and more than five years to substantially complete.