Which Browsers Are Supported?
Speedtest.now works in all modern browsers. For the most accurate results, use a recent desktop Chrome or Firefox. Here's what each browser does well — and why mobile browsers often underreport speed.
Browser support matrix
| Browser | Supported | Recommended | Notes |
|---|---|---|---|
| Chrome (desktop) 90+ | ✓ | Best | Most accurate; best parallel fetch support |
| Firefox (desktop) 88+ | ✓ | Best | Equivalent accuracy to Chrome |
| Edge (desktop) 90+ | ✓ | Good | Chromium-based; near-identical to Chrome |
| Safari (desktop) 14+ | ✓ | Good | Minor fetch concurrency limits may lower peak results slightly |
| Chrome (Android) | ✓ | Limited | CPU-limited on many devices; may underreport by 20–40% |
| Safari (iOS / iPadOS) | ✓ | Limited | All iOS browsers use WebKit; CPU limits apply |
| Internet Explorer | ✗ | No | No Fetch API support; not supported |
Why Chrome and Firefox give the most accurate results
Speedtest.now uses the browser's Fetch API to make multiple parallel HTTP/HTTPS requests to the test server simultaneously. Accurate speed measurement requires saturating your connection — a single download stream rarely does so on fast connections. Chrome and Firefox support the highest concurrency for Fetch requests and have the most optimised JavaScript engines for processing the measurement loop quickly.
What the test measures inside the browser
- Download: Multiple parallel GET requests to the test server, measuring total throughput
- Upload: Multiple parallel POST requests with generated payload data
- Ping / latency: Repeated small HTTP requests timed from send to first-byte response
- Jitter: Standard deviation of those ping measurements
- Packet loss: Count of requests that time out or fail to complete
Why mobile browsers often show lower speeds
A mobile browser result that's lower than your Wi-Fi speed is usually a device limitation, not a network problem. There are two reasons:
1. CPU throttling
Processing incoming data at gigabit speeds requires significant CPU. A desktop processor handles this easily. Mid-range Android phones and even high-end iPhones can hit CPU saturation during a test, causing the measured throughput to fall below the actual network capacity. The bottleneck is the device's ability to process packets, not the network's ability to deliver them.
2. JavaScript engine differences
Mobile browsers run the same measurement JavaScript as desktop, but slower. V8 on Android Chrome is the same engine as desktop, but running on a device with less thermal headroom and typically slower memory bandwidth. The measurement loop itself takes longer, reducing the effective sampling rate and potentially underreporting peak speed.
Getting the most accurate result on any browser
- Close all other tabs — other pages may initiate network activity during the test
- Disable browser extensions, especially ad blockers and VPNs — they intercept and inspect requests, adding latency
- Use a wired Ethernet connection if testing on a desktop — eliminates Wi-Fi variability
- Run the test 2–3 times and compare — single-test variation is normal, especially on Wi-Fi
- On mobile, test when the device is plugged in — battery-saving modes can throttle CPU performance