Internet Speed Test — Measure Download, Upload, Ping and Latency
Reading your result
One number isn't enough. It takes four to describe your connection.
Download
The direction you use when watching video, opening pages and downloading files. Most things at home depend on this number.
25 Mbps is comfortable for one person; a busy household wants more than 100 Mbps. What is Mbps? →
Upload
The direction used when you send: video calls, cloud backup, sharing photos.
5 Mbps is enough for a video meeting; above 20 Mbps is noticeable if you work from home. Why upload speed matters →
Latency (ping)
How long your request takes to travel there and back. It is independent of speed — a fast line can still respond slowly.
Under 30 ms is very good; above 60 ms is felt in games and video calls. What is ping? →
Fluency (RPM)
How many request-response round trips your connection completes per minute while the line is busy. If this is low, video calls freeze and games lag even when the speed looks high.
Above 1,200 is fluent and above 600 is good; below 200 the congestion is noticeable. Why internet slows under load →
Our own measurement data
How do providers in Türkiye look in these measurements?
The numbers below are not estimates or advertising: they are the typical value of measurements run against our own servers, the point where half of results are above and half below. We don't publish the average, because a minority of very fast lines pulls it above what most users actually see.
| Rank | Provider | Download | Upload | Latency |
|---|---|---|---|---|
| 01 | Turkcell Superonline | 84.4Mbps | 22.6Mbps | 22ms |
| 02 | TurkNet | 74.9Mbps | 22.3Mbps | 18ms |
| 03 | Türksat Kablonet | 52.3Mbps | 9.8Mbps | 23ms |
| 04 | Türk Telekom | 39.7Mbps | 13.8Mbps | 18ms |
| 05 | Millenicom | 37.0Mbps | 12.4Mbps | 17ms |
| 06 | Vodafone Net | 33.7Mbps | 14.1Mbps | 27ms |
| 07 | D-Smart | 22.1Mbps | 8.5Mbps | 31ms |
Fixed-line connections; mobile measurements and records arriving over a VPN are excluded. A provider needs at least 200 measurements in this period to appear. Results vary with where you are, your plan and the time of day — this is a snapshot from the field, not a plan comparison.
Open method
How is this measured?
The test runs against servers we operate ourselves rather than rent from someone else. We wrote the measurement method out in full and published it under a permanent identifier, so how a number was produced can be checked. This is the easiest part of a speed test to keep hidden; we leave it open.