Just a few consumer tests but they look good. Gig wireless (shared) is behind schedule; Telstra in Australia and SKT in Korea thought they would have it for customers before the end of 2016 but they haven't made it by Xmas. Pieces of the necessary technology are deploying widely. T-Mobile in many places is using four antenna MIMO and 256 QAM advanced coding. They are not bonding 4 channels, which would take peaks to around a gigabit. Sprint is widely bonding three channels and will go to four channels (80 MHz.)40-50 megabits download (right), two or there times more that without those two improvements (left.) Larger picture below. That's much less than the 200-300 megabits peak, shared, from 20 MHz at the cell tower. Few users get the full 200+ most of the time. However, many will get much higher speeds than they are today.
Most users will not get close to the peak speed.
Speed drops with distance and walls. Most phones are not yet optimally tuned. Other users are also connecting. Many carriers artificially limit individual users, partly to ensure that if another user wants a higher speed, they get it right away.
If T-Mobile gets more spectrum at the current auction, they will be able to add 4 carrier aggregation for a peak speed of about a gigabit and real customer speeds in the hundreds of megabits. Both Sprint and AT&T have said they'll get to the gigabit in the next few years.
Some very senior telco execs think these advances will make "5G" unneeded most places for a decade or more.