Run a browser connection snapshot
Choose how many samples to run and how long each request can wait before it times out. More samples usually smooth out jitter, but they also take longer to finish.
Assumptions and rounding: latency is shown as the median time to receive response headers, download speed is estimated from response body bytes divided by body download time, and displayed values are rounded to two decimals where it helps readability.
Results
Performance disclaimer: this test measures the path between your browser and this site. VPNs, Wi-Fi congestion, browser throttling, caching behavior, and server load can change the result. For a full ISP benchmark, compare it with a dedicated multi-server speed test.
How it works
1. Browser hints
The page reads values like effectiveType, downlink, and rtt when the browser exposes them. Those values are hints from the device and network stack, not direct measurements from this page.
2. Active sampling
Each sample sends an uncached fetch request back to the current origin with a unique query string. The test records when response headers arrive and how long the response body takes to download.
3. Safer summaries
The headline latency uses the median successful sample because a single outlier can mislead. Consistency is derived from spread across successful runs, and failed samples are reported instead of silently ignored.
4. Practical limits
Because this is a single-page browser tool, it does not spin up multiple remote endpoints or saturate the line like a dedicated benchmarking service. Its strength is quick, repeatable snapshots from the same device and site.