If you bought 5G home internet as a rural stand-in for fiber, the newest national speed tests are not a single “rural is slower” story. Ookla’s Aug. 26 Speedtest Intelligence analysis found that urban median downloads beat rural ones on T-Mobile Home Internet, AT&T Internet Air, and Verizon 5G Home, and that rural latency ran higher on all three. Uploads do not follow. AT&T’s rural uploads beat its urban uploads. That split is the part of the report that actually changes a shopping decision this week.
The report covers the first half of 2026. U.S. 5G fixed wireless from those three brands topped about 17 million connections in the first quarter, or 13.74% of 123.75 million home broadband subscriptions in Census figures Ookla cites. Roughly 70% of Ookla’s second-quarter 5G home-internet test samples came from urban addresses and 30% from rural ones. Sample volume is a proxy for where testers live. It is not a subscriber count, and it is not a coverage map for your street.
What the Q2 medians actually say
T-Mobile still led in the second quarter. Ookla put its median download at 222.7 Mbps, 38.4% above AT&T Internet Air and 76% above Verizon 5G Home. T-Mobile’s median upload was 18.12 Mbps. AT&T’s second-quarter medians were 160.43 Mbps down and 10.06 Mbps up. Verizon’s were 126.53 Mbps down and 12.20 Mbps up. Those are national medians of Speedtest samples, not a promised speed at checkout.
All three brands lost speed from the first quarter to the second. Ookla ties that dip to spring and summer foliage and to heat, which can push radios toward slower, more stable connections. Verizon’s median download fell from 143.64 Mbps to 126.53 Mbps, the steepest drop of the three. Uploads fell more than downloads, including 21.5% on T-Mobile. That is seasonal, not a new monthly price.
Ookla’s Verizon 5G Home numbers exclude Starry, which Verizon still sells separately. The fastest Verizon tests also look capped near 300 Mbps even when the network is idle, while T-Mobile’s top 10% of tests often clear 600 Mbps off-peak. Treat those as product rules, not as a different plan hiding on the same bill.
AT&T’s climb into second place is older than this week’s headline. Ookla says the first-quarter median download of 167.34 Mbps was nearly 60% higher than 104.61 Mbps in the third quarter of 2025, and it ties that jump to 50 MHz of EchoStar low-band and mid-band spectrum AT&T is already using. The shopping question is whether Internet Air at your address now looks like that second-place median.
Rural vs. urban is three measurements, not one
Downloads are the clean rural gap. Urban medians beat rural medians on all three carriers in both the first and second quarters. T-Mobile had the smallest urban-rural download gap. Latency goes the same way: urban testers saw median multi-server latency 7 to 13 milliseconds lower than rural testers. If you work from home, video-call, or game, that delay is the part of “rural is worse” that the download headline can hide.
Uploads break the pattern. T-Mobile and Verizon still showed an urban upload advantage. AT&T did not. Rural AT&T Internet Air uploads beat urban ones in both quarters Ookla published. If your household sends a lot of video, backups, or camera feeds, an AT&T rural line is not automatically the weak upload of the three, even if its national median upload sits last.
State samples are harsher than the national medians. Only Hawaii, New York, and the District of Columbia had 40% or more of second-quarter 5G home-internet tests hit the FCC’s 100/20 broadband floor. Maine was the continental low at 23.3%. Alaska printed 4.8% on a sample Ookla called too small to trust. That is a share of tests, not a network-wide 100/20 failure.
What to do if you are shopping 5G home internet
Run the carrier’s address check, then a speed test at the hours you actually use the internet, before you port a home line or return a gateway. A national median of 222.7 Mbps does not travel with the box. Evening congestion shrinks the gap between brands. Verizon’s slowest tests held steadier through the day; that stability is not a higher median.
If you are rural, do not buy 5G home internet off an urban ad. Trees, hills, and tower distance show up in these medians. If you are urban, the same report is why 5G home internet is showing up as a cable alternative: short paths and denser sites, not a sudden rural buildout.
This story does not change advertised monthly prices. Compare the gateway in your house to the plan on your bill. If the Ookla split matches what you see at 8 p.m., try the other two 5G home brands at that same address, or keep fiber or cable if you already have it.
Sources
- Ookla: U.S. FWA Was Envisioned as a Rural Broadband Fix, but Urban Users Are Embracing It Too (Sue Marek, Aug. 26, 2026)