Ookla’s Speedtest is the default bandwidth check for almost every household on the planet. The desktop app is fast to install, the web version is fast to load, and the result looks authoritative. It also runs ads, upsells an Ookla VPN, and shares anonymized test data with ISPs and regulators in ways some users do not want. Past that, the picked-server problem is real: Speedtest defaults to the nearest server, which is often hosted by your own ISP, which can flatter the number. People looking for Speedtest by Ookla alternatives usually want one of three things: a result less influenced by ISP-hosted servers, an ad-free interface, or a tool that measures latency and bufferbloat rather than just peak throughput. Here are seven that cover all three.
Quick comparison
| Tool | Best for | License | Platforms | Standout |
|---|---|---|---|---|
| Fast.com | One-click sanity check | Free | Web only | Hits Netflix CDN, hard to fake |
| Cloudflare Speed Test | Modern web test | Free | Web only | Latency under load, no ads |
| nperf | Multi-metric report | Free | Windows, web | Includes streaming and browsing scores |
| Speedsmart | HTML5, no plugin | Free | Web only | Works on any modern browser |
| Waveform Bufferbloat Test | Latency under load | Free | Web only | Measures the real lag problem |
| OpenSpeedTest | Self-hosted | Free, open-source | Windows, macOS, Linux | Run your own LAN test server |
| iPerf3 | LAN and WAN diagnostics | Free, open-source | Windows, macOS, Linux | The professional’s tool |
Why people leave Speedtest by Ookla
Ads in the free Windows app and on the website. Banners across the bottom, video ads after the test, and a Speedtest VPN upsell in the result screen.
Server selection skews results. Speedtest’s default server is often hosted directly by the local ISP, which produces best-case throughput numbers that may not reflect what real traffic to a normal destination sees.
Data sharing. Ookla shares aggregated and anonymized test data with carriers and regulators. The Speedtest privacy disclosures are clear about this, but some users prefer a tool that does not contribute to that pool.
It measures peak bandwidth, not what hurts. The thing that makes a video call stutter is latency under load (bufferbloat), and Speedtest does not surface that as the headline number.
The alternatives
Fast.com: one-click sanity check
Netflix’s Fast.com is a one-page test that runs against Netflix’s CDN nodes. There is no chooser, no ad, no signup, and the result is the speed you would get pulling video from Netflix specifically. Click the “Show more info” toggle to add latency, upload speed, and the loaded/unloaded latency split.
Where it falls short: Web only, no desktop app. Bound to Netflix’s CDN, so the result reflects Netflix specifically.
Pricing: Free.
vs Speedtest: Fewer ads. Less server-choice manipulation possible.
Migrating from Speedtest: Bookmark fast.com.
Download: fast.com
Bottom line: The simplest “what is my real download speed” test.
Cloudflare Speed Test: modern web test
Cloudflare’s tool runs from your browser against Cloudflare’s global edge. It reports download, upload, latency unloaded, latency under load (the bufferbloat figure), packet loss, and jitter. The result page explains what each number actually means.
Where it falls short: Web only. Results reflect connectivity to Cloudflare’s edge, which is close to almost everyone but not identical to every destination.
Pricing: Free.
vs Speedtest: No ads, more meaningful metrics, particularly latency under load.
Migrating from Speedtest: Bookmark speed.cloudflare.com.
Download: speed.cloudflare.com
Bottom line: The pick when you want to understand the result, not just see a number.
nperf: multi-metric report
nperf measures download, upload, latency, and browsing and streaming scores in one run. Results include a quality grade rather than a single throughput number, which makes nperf useful for diagnosing whether a connection is “fast enough” for a specific use case.
Where it falls short: Carries ads on the free web tier. Desktop app is Windows only.
Pricing: Free.
vs Speedtest: Different rubric. Reports use-case quality scores rather than raw peak.
Migrating from Speedtest: Use the website or install the Windows app.
Download: nperf.com
Bottom line: The pick when you care about whether streaming works, not just the megabits.
Speedsmart: HTML5, no plugin
Speedsmart is a long-running HTML5 speed test that picks from a small set of distributed servers and reports download, upload, ping, and jitter. The interface is minimal and the test completes quickly.
Where it falls short: Web only. Smaller server set than Ookla’s.
Pricing: Free.
vs Speedtest: Smaller, faster, fewer ads.
Migrating from Speedtest: Bookmark speedsmart.net.
Download: speedsmart.net
Bottom line: A clean HTML5 sanity test.
Waveform Bufferbloat Test: latency under load
Waveform’s tool is single-purpose: it measures latency before, during download, and during upload, and grades the bufferbloat. If your video calls go choppy when someone in the house starts a large download, this is the test that surfaces the cause.
Where it falls short: Web only. Reports a focused metric, not a general dashboard.
Pricing: Free.
vs Speedtest: Different question. Useful in addition to a throughput test, not instead.
Migrating from Speedtest: Use both. Speed for throughput, Waveform for the call-quality angle.
Download: waveform.com/tools/bufferbloat
Bottom line: The pick for diagnosing why calls glitch.
OpenSpeedTest: self-hosted
OpenSpeedTest is an HTML5 speed test you can run on your own LAN. Drop the server on a Raspberry Pi, a NAS, or a desktop, and measure intra-network throughput from any device with a browser. Useful for diagnosing wired-vs-wireless throughput, wifi access-point coverage, and switch saturation.
Where it falls short: Measures LAN by default; WAN testing requires public exposure or a cloud host.
Pricing: Free and open-source.
vs Speedtest: Different problem. Measures your network, not your ISP.
Migrating from Speedtest: Install OpenSpeedTest server on a home server; point any browser at it for LAN tests.
Download: openspeedtest.com
Bottom line: The pick when “is my wifi the problem” is the real question.
iPerf3: LAN and WAN diagnostics
iPerf3 is the command-line tool network engineers actually use. Run a server on one machine, a client on another, and get raw TCP or UDP throughput, jitter, and loss between them. Available on every desktop OS and most routers.
Where it falls short: Command line only. Requires you to control both ends of the test.
Pricing: Free and open-source.
vs Speedtest: Different category. iPerf3 measures the path between two endpoints you choose, with no ad layer and no server-pick bias.
Migrating from Speedtest: Install iperf3 with your package manager or download the binary; run iperf3 -s on one end and iperf3 -c <server> on the other.
Download: iperf.fr
Bottom line: The professional’s tool, and the right one for any controlled comparison.
How to choose
Pick Fast.com when you want the quickest possible sanity check.
Pick Cloudflare Speed Test when you want a result that explains itself, including latency under load.
Pick nperf when you want a use-case-shaped report rather than a raw throughput number.
Pick Waveform to diagnose call quality problems and bufferbloat specifically.
Pick OpenSpeedTest to test your own network instead of your ISP path.
Pick iPerf3 when you control both ends and want raw, repeatable numbers.
Stay on Speedtest by Ookla if you specifically want the historical record, the global comparison data, and the ISP-hosted servers, and you can tolerate the ads.
FAQ
Why do different speed tests give different numbers? Each test picks a different destination, runs a different number of parallel streams, and decides differently when to stop measuring. Two tests on the same connection can easily differ by 20–30%, especially when one runs against an ISP-hosted server and the other does not. For a fair number, use the same tool at the same time of day twice and compare those.
What is bufferbloat and why should I care? Bufferbloat is excess buffering inside a network device that causes latency to spike when the link is busy. A connection can have a fast peak throughput and still feel terrible on video calls because the latency jumps from 20ms to 500ms during a download. Waveform and Cloudflare’s tools surface this metric directly.
Can I run a speed test from the command line?
Yes. Ookla offers a speedtest CLI on its site. Cloudflare’s test has a speed-cloudflare-cli community wrapper. iPerf3 is command-line by design.
Are any of these self-hostable? OpenSpeedTest is the easiest to self-host: drop the server on a NAS or a Pi and any browser becomes a client. iPerf3 is also self-hosted by definition. LibreSpeed is another open-source option not covered above.
Which test best reflects what I actually do online? Cloudflare and nperf are the most use-case-aware: they include latency under load and per-use-case grades. Pure peak-throughput tests like Speedtest reflect the maximum possible, which is rarely what daily browsing or streaming experiences.
