Free · No sign-up · No app
Internet Speed Test
Check your download speed, upload speed, ping and jitter in about 15 seconds. The test runs against our own server on Cloudflare's global network, entirely in your browser: nothing to install, no account, no email address. Press GO below and watch the numbers update live.
- Ping
- –ms
- Jitter
- –ms
- Download
- –Mbps
- Upload
- –Mbps
Runs entirely in your browser against our own edge server: no app, no account, no installation. Each result is measured fresh; nothing is saved or sent anywhere beyond this one test.
How this speed test works
Most of what people call an "internet speed" is really four separate numbers, and this tool measures all four:
- Download speed (Mbps): how fast data can travel from our server to your device. This is what determines how quickly a video buffers, a webpage loads, or a large file finishes downloading.
- Upload speed (Mbps): how fast data travels the other way, from your device to our server. This matters for video calls, uploading photos or files to cloud storage, and live streaming.
- Ping (ms): the round-trip time for a tiny packet of data to reach our server and come back. Lower is better. Ping matters most for anything real-time: online gaming, video calls, and voice chat.
- Jitter (ms): how much your ping varies from one moment to the next. A connection with low ping but high jitter can still feel choppy in a video call or online game, because the timing keeps shifting.
How the test is measured
We built this tool from scratch on Cloudflare's edge network (the same infrastructure Cloudflare's own speed test runs on) rather than reusing anyone else's paid API, so there is no third-party service in the loop. The test runs in four stages:
- Ping: your browser sends six quick, empty requests to our server and times each round trip. The first is discarded (it usually includes some one-time connection setup overhead), and the remaining five are used to calculate your ping (the middle, or median, value) and jitter (the average change between consecutive pings).
- Download: your browser downloads a stream of data from our server in increasing chunks, starting at 1 MB. If that finishes in under two seconds, the test moves up to 5 MB, then 10 MB, then 25 MB, so a fast connection gets measured accurately without waiting around, and a slow connection isn't forced to sit through a 25 MB download it doesn't need.
- Upload: the same escalating-size approach, but in reverse: your browser generates a chunk of random data and sends it to our server, which confirms how many bytes it received.
- Result: your download and upload speeds are calculated from the total bytes transferred divided by the total time taken across every stage that actually ran, not just a single sample, which is a more representative number than one quick snapshot.
What is a good internet speed?
"Good" depends entirely on what you're doing and how many devices share the connection at once. This table shows roughly what each activity needs. Treat it as a planning guide, not a hard cutoff; most things still work below these numbers, just less comfortably:
- Email, browsing, social media: 5-10 Mbps download is comfortable for one person.
- Standard-definition video calls (Zoom, Teams, FaceTime): around 3-4 Mbps down and up.
- HD video streaming (1080p, one stream): roughly 5-10 Mbps download.
- 4K video streaming (one stream): commonly cited as 25 Mbps or more download, since 4K carries far more data per second than 1080p.
- Competitive online gaming: the download/upload numbers matter less than a low, stable ping (under about 50 ms is generally considered good, and consistency (low jitter) matters as much as the raw ping figure).
- A household with several people streaming, gaming and video-calling at the same time: speeds need to be shared across everyone, so add up each activity's rough requirement rather than assuming one person's "good enough" number covers the whole home.
Three examples: matching speed to how you actually use the internet
Raw Mbps numbers are hard to picture on their own, so here are three concrete scenarios showing how the same test result plays out differently depending on the situation.
- Example 1: a single remote worker on video calls all day. A test result of 40 Mbps down / 10 Mbps up is comfortably more than enough for back-to-back video meetings, screen sharing, and normal browsing in the background. The upload number matters more here than the download number, since a video call sends your camera feed continuously. If upload tests consistently under 3 Mbps, calls are more likely to freeze or drop to audio-only.
- Example 2: a family of four, two people streaming 4K, one on a video call, one gaming online. Each 4K stream alone can use 25 Mbps, so two of them plus a video call and some gaming background traffic can add up to well over 60-70 Mbps of simultaneous download demand. A test result of 50 Mbps down might feel fine for any one of those activities tested alone, but the household could still see buffering or dropped call quality once all four things happen at once. The fix here is testing at the busiest moment of the day, not first thing in the morning when nobody else is on the network.
- Example 3: someone testing on wifi from another room and getting a surprisingly low result. A test that shows 30 Mbps on wifi but the ISP's plan is rated for 300 Mbps is a classic wifi-distance or wifi-interference problem, not necessarily a problem with the internet connection itself. Re-running the test with a device plugged directly into the router via an ethernet cable, or standing right next to the router, is the fastest way to tell whether the bottleneck is the wifi signal or the connection coming into the building. See the section below for more on why results vary.
Why your result can vary from one test to the next
It's completely normal for two speed tests run minutes apart on the same connection to show somewhat different numbers. The most common reasons:
- Wifi vs. a wired connection. Wifi signal strength drops with distance and walls, and can be crowded out by neighbouring wifi networks, microwaves, and other 2.4 GHz devices. A device connected directly to the router by ethernet cable will almost always test faster and more consistently than the same device over wifi.
- Other devices and people sharing the connection. A speed test measures what's available at that exact moment. If someone else in the house is mid-download, streaming 4K, or a device is quietly running a large backup in the background, your test result reflects the leftover capacity, not the full plan speed.
- Time of day. In neighbourhoods where many households share the same local infrastructure, speeds can dip during peak evening hours (commonly referred to as "peak congestion") when everyone is online at once, then recover late at night.
- A VPN or proxy. Routing your connection through a VPN adds an extra hop (and often encryption overhead), which can meaningfully reduce the speed a test reports compared to testing with the VPN switched off.
- Server distance. Every speed test measures against a specific server somewhere on the internet. This tool runs on Cloudflare's global network, which has edge locations in a very large number of cities worldwide specifically to keep that distance short, but it's still one factor among several, alongside your ISP's network and your local wifi.
- ISP-specific quirks. Some providers handle certain kinds of traffic differently at busy times. If your results consistently look low specifically on your provider's own network, the dedicated pages below cover what a normal result looks like for a few major ISPs.
Why download and upload speeds are usually different
On most home internet plans (cable, DSL, and even a lot of fiber and 5G Home Internet plans), the advertised download speed is much higher than the advertised upload speed. A common example is a plan sold as "300 Mbps" that actually means roughly 300 Mbps download and only 10-20 Mbps upload. This is called an asymmetric connection, and it's a deliberate design choice, not a fault: most home internet use is download-heavy (loading webpages, streaming video, downloading files), so providers allocate more of the available capacity to download.
- Symmetric connections (equal download and upload speed) are more common on business-grade plans and some newer fiber-to-the-home services, where the provider's network technology doesn't need to split capacity unevenly.
- Why upload speed still matters even on an asymmetric plan: video calls, livestreaming, uploading large files to cloud storage, and backing up photos all depend on upload speed specifically: a connection with a fast download number can still feel slow for these tasks if the upload number is low.
- What to check: compare both numbers from this test against what your plan actually promises for each direction (not just the headline download figure), since that's the number your ISP is contractually offering.
Speed tests for specific internet providers
If you know which provider serves your home, one of these pages runs the exact same test above but with guidance written for that provider's typical setup, useful for telling a normal result from an actual problem worth calling support about.
- Xfinity speed test: for Comcast Xfinity cable and fiber customers.
- Spectrum speed test: for Charter Spectrum cable customers.
- Verizon speed test: for Verizon Fios fiber and 5G Home Internet customers.
- AT&T speed test: for AT&T Fiber and AT&T Internet customers.
- TM speed test: for Telekom Malaysia Unifi fiber customers.