VPS Speed Test — Server Latency & Throughput
Measure how fast a server responds from your location. Enter a URL pointing at a small static file on your VPS to get TTFB, latency spread (p50/p95/p99) and download throughput.
⚡ What TTFB tells you
- <100ms — excellent, server close & fast
- 100–300ms — good
- 300–800ms — average, check backend/region
- >800ms — slow, wrong region or overloaded
🧪 How it measures
Your browser fetches the URL repeatedly with cache disabled and a unique query string, timing each round trip with the high-resolution Performance API. p95/p99 expose tail latency that averages hide.
🛠️ Get accurate numbers
- Point at a static file, not a heavy app route
- Use a 50–200KB file for throughput
- Run from the region your users are in
- Cross-origin servers must send
Access-Control-Allow-Originfor byte counts
Frequently Asked Questions
Can a browser benchmark a VPS's CPU, RAM or disk?
No. A web browser is sandboxed and cannot run code on a remote server, so CPU/RAM/disk-IO of a VPS can only be measured by running a benchmark on the server (for example sysbench or fio over SSH). This tool measures what a browser truly can: the network latency, jitter and download throughput between your location and the server — the numbers that decide how fast the VPS feels to real visitors.
Why is my throughput showing 'n/a'?
Download byte counting needs the server to send an Access-Control-Allow-Origin header for cross-origin requests. Latency, TTFB and jitter are still measured accurately even without it.
Which providers does this work with?
Any reachable HTTP/HTTPS server — Hetzner, Contabo, OVHcloud, DigitalOcean, Vultr, Linode, AWS, your own box, etc. Point the URL at a small static file for the cleanest results.
Is anything sent to Speedelly?
No. The fetches go directly from your browser to the server you enter. We never see the URL or the results.