Internet Speed Test

How to Run a Speed Test

Features

The Number on the Bill vs. the Number You Actually Get

Every internet plan is sold on a headline number — "500 Mbps," "1 Gig" — that describes a ceiling, not a promise. What actually reaches your device depends on your wiring, your router, how many devices are sharing the connection right now, and how far your Wi-Fi signal has to travel before it even gets to you. This test measures what's really arriving at this exact moment, which is usually the more useful number when something feels slow.

The Mbps-vs-MB/s Mixup That Confuses Every Download Bar

This test reports Mbps — megabits per second — the standard unit every ISP advertises and every speed test measures in, since network throughput has always been counted in bits. Most download progress bars, on the other hand, show megabytes per second, and a byte is eight bits — so a 100 Mbps connection shows up as roughly 12.5 MB/s once a file actually starts downloading. Seeing a "100 Mbps" result here and then a "12 MB/s" reading in a browser download isn't a discrepancy or a sign anything's wrong; it's the exact same speed, just reported in two different units that happen to share a very similar-looking abbreviation.

Who Actually Reaches for This

• Anyone troubleshooting a slow connection — before calling an ISP, having an actual number to report speeds up the conversation considerably
• People shopping for a new plan — comparing what they currently get against what a provider is promising
• Remote workers — confirming their home connection can actually handle video calls before an important meeting
• Landlords and roommates — settling exactly whose devices are eating shared bandwidth
• Anyone who just moved — checking whether a new router placement or ISP switch actually improved anything

How the Measurement Actually Works

A single download request rarely reflects a connection's real capacity — one stream often can't fill a fast pipe on its own, the same reason download managers split large files into chunks. This test opens several connections at once and adds up everything transferred across all of them over a fixed window, which is much closer to how your connection behaves under real, everyday load. Ping is measured separately with small, quick round trips before the heavier transfers start, so latency isn't skewed by a connection already busy moving data.

When the Test Looks Fine But One Site Doesn't

A clean result here — good download, good ping — doesn't rule out a specific site or service still feeling slow, and that gap confuses more people than it should. This tool measures your path to Cloudflare's network; it says nothing about a slow DNS resolution step or a misconfigured cache header on the one server you're actually trying to reach. If a single site drags while everything else feels normal, our DNS Lookup Tool checks whether that domain is resolving cleanly in the first place, and the HTTP Header Checker shows whether its server is actually serving cached responses or rebuilding the page from scratch on every request.

Two Reasons People Actually Run This

Proving a problem to an ISP. "It feels slow" gets nowhere on a support call. A specific number, a time of day, and a repeat result across a few tests turns a vague complaint into something a technician can actually act on.

Deciding whether a router move or upgrade helped. Run the test before making a change — moving the router, switching to Ethernet, upgrading a plan — then run it again after. The comparison is usually a more honest answer than "it feels a bit faster." If you're setting up a home server after confirming your bandwidth, our My IP Address tool gives you the exact address to open ports against.

Getting a Result You Can Trust

• Test at different times of day. Evenings tend to run slower everywhere, since that's when most of a neighborhood is online at once — a morning test alone won't tell the full story.
• Use a wired connection when you can to isolate whether a slow result is your internet plan or your Wi-Fi.
• Run it two or three times, not once. Any single result can be skewed by a momentary blip — a consistent pattern across a few runs is far more meaningful.
• Turn on Extended Test on a fast connection. A short test can under-measure gigabit-class connections that take a moment to ramp up to full speed.
• Pause backups and streams elsewhere in the house first if you want a reading of your plan's real ceiling rather than what's left over after everyone else's usage.

The Test That Actually Settles an Argument

A laptop on Wi-Fi across the house shows 80 Mbps on a plan advertised at 500. Plugging the same laptop directly into the router with an Ethernet cable and running it again shows the full 480 Mbps the plan actually delivers — nothing about the ISP connection changed between the two tests, only the path the data took to reach the device. That single comparison usually resolves the argument faster than any amount of debate about whether "the internet is slow," because it separates a Wi-Fi placement problem from an actual plan or provider problem in about two minutes.

When the Numbers Don't Match What You Expect

A single low result usually isn't a verdict — congestion, a busy Wi-Fi channel, or a background update downloading on another device can all pull one test down without meaning anything is actually wrong. What matters more is the pattern across several tests at different times. And this measures your path to Cloudflare's network specifically, the same way every major speed test measures a path to its own infrastructure — it's an excellent proxy for your real-world connection quality, but it isn't a guaranteed measurement of the exact number your ISP will show on their own tools.