Latency Test
Measure round-trip time from your browser to the access points that matter.
Sign in to run a test
Measurements are attributed to the person who submitted them, so running a test needs an identity.
Or use the shared access password:
No sign-in method is configured yet. An administrator needs to set up GitHub or LinkedIn sign-in, or set a shared client password in the admin console.
Locating you…
Loading targets…
Results
Rows are grouped by measurement method. A tcp_connect of 12 ms and an
https_fetch_approx of 12 ms are not the same quantity, so they are never
ranked against each other.
| Server | Method | Platform | Median | P95 | Best | Jitter | Samples |
|---|
What each method actually measures
https_ttfb- Browser. From request sent to first response byte. Needs the target to send
Timing-Allow-Origin; without it no value is recorded. https_fetch_approx- Browser. Wall clock around a
no-corsfetch — DNS, TCP, TLS, HTTP, error handling and browser scheduling all included. Useful for comparing targets within one run; not an absolute round-trip time. tcp_connect- Probe agent. SYN to SYN-ACK. This is the only honest measurement of MT4/MT5 access points, which speak a proprietary binary protocol on :443 and are not HTTPS servers at all — a browser can neither time them nor open a raw socket.
icmp_ping- Probe agent. ICMP echo round trip. Widely rate-limited or dropped; absence of a reply is not evidence that a host is down.