Measure download, upload, idle responsiveness and latency under load using Cloudflare edge-network measurements.
—
—
Cloudflare edge infrastructure. Idle and loaded latency are HTTP edge timing; throughput is browser-to-edge transfer performance.
Repeat the test under similar conditions. Compare Ethernet/Wi-Fi and busy/quiet periods when diagnosing a persistent issue.
Loaded latency is measured by the Cloudflare engine while download or upload traffic is actively loading the path. 0bps classifies the largest positive increase above idle as MINIMAL ≤5 ms, LOW ≤15 ms, MODERATE ≤30 ms, HIGH ≤60 ms or SEVERE >60 ms. This is an advisory signal, not a router certification or ICMP measurement.
Measurement traffic runs against Cloudflare edge infrastructure. VPNs, Wi-Fi conditions, browser load, device limits and routing can change results. The connection score is a transparent 0bps heuristic, not an industry-standard certification.
The test ramps through progressively larger transfers. The live dial follows measured throughput callbacks on a standard shared test range that starts at 100 Mbps and expands only when real download throughput requires it. The same range is retained for upload, so a 90 Mbps download and 18 Mbps upload remain visually comparable. The SVG needle and large readout ease smoothly toward each real sample; the stored result is never replaced by an animated estimate.
Higher throughput is better. Low idle latency helps responsiveness; a small latency increase under load means heavy transfers are less likely to delay games, calls and other interactive traffic.
The result combines edge throughput, unloaded responsiveness and measured latency under load. Read the evidence together rather than treating download Mbps as the whole connection.
Download and upload are browser-to-Cloudflare edge throughput. Idle latency/jitter and simultaneous latency-under-load samples come from the same Cloudflare measurement engine.
High Mbps helps bulk transfers and streaming. Low idle latency helps interaction. A small loaded-latency increase means the path stays responsive while transfers are busy; a large increase is a bufferbloat warning signal.
This is not ICMP and it is not the latency to every game, call service, VPN endpoint or website. Wi-Fi, VPNs, device load and route selection can change a sample.
Repeat two or three tests, compare Wi-Fi with Ethernet if possible, pause large transfers and compare another time of day before blaming the ISP or device.
0BPS SHOWS MEASURED OR CALCULATED FACTS SEPARATELY FROM INTERPRETATION. WHERE A RESULT IS ADVISORY, ESTIMATED OR DESTINATION-SPECIFIC, THE LIMIT IS STATED IN THE TOOL.
The test uses adaptive edge-network transfers and multiple samples. Results can still vary with Wi-Fi quality, VPNs, browser load, routing and the test edge location.
Different services use different servers, routes, sample sizes and calculation methods. A modest difference is normal; repeated tests under similar conditions are more useful than one isolated result.
Latency is the round-trip response time. Jitter is how much that response time varies between samples. Lower and more stable values are generally better for gaming, voice and video calls.
It measures Cloudflare edge HTTP latency while download or upload traffic is actively loading the path. 0bps compares that loaded value with the idle baseline; a large increase is an advisory bufferbloat signal, not a router certification.
Adaptive edge transfers reduce the risk of estimating a fast line from one tiny request.
Compare repeated tests, not a single peak. Wi-Fi, VPNs, device load and routing can change throughput and latency.
Short examples using the same units and assumptions as this module.
Run two or three tests under similar conditions. A repeated range is more useful than one unusually high or low sample.
If Ethernet is much faster than Wi-Fi, the local wireless link may be the limiting factor rather than the ISP line.