The average speed of over 10,000 Open Signal tests of U.S. 5G was under 60 Mbps.* The average speed of 4G at Bell Canada last fall was 70 Mbps. At Canadian #2 Telus, it was 75 Mbps. The methodology was essentially the same. The results are consistent with testimony by T-Mobile's CTO Neville Ray; Verizon's Glenn Wellbrock; and just about every engineer building networks. (A CTO has just turned on 5G at a major carrier and tells me off the record their 4G is currently better.)
Ray testified to the FCC that in theory 5G could be 20% faster in low-band and 50% faster in mid-band. For now, 4G is far ahead until 5G adds LAA and other tools already working in 4G. Wellbrock pointed out that in these frequencies, neither 5G NR nor any other software can add much to capacity. 4G is close to the Shannon Limit already. Higher performance without more spectrum or antennas would break the laws of physics.
Next time a pundit or politician talks of the very high speeds of 5G, ask them what is the actual speed of the 5G being deployed? Be ready with those numbers.
Open Signal "found no significant improvement in 5G latency over 4G." It will be years before 5G low latency is widely deployed - and 4G will be almost as low by then.