Internet Speed Test

Measure download, upload and browser HTTP latency on Wi-Fi or Ethernet. Compare repeat trials and the conditions behind them.

0
Mbps
Ready
Your broadband plan (optional)

Enter the speed in your contract to see what percentage you're getting. Many contracts also state a minimum or typical speed — check yours.

Download
–Mbps
Upload
–Mbps
Ping (idle)
–ms
Jitter
–ms
Ping during download
–ms
Ping during upload
–ms

What this connection handles

Run a test to see

Connection

IP address
–
Provider
–
Location
–
Test server
–
Network type
–
Browser estimate
–
Data used
–
  • Testing WiFi? Run once next to your router and once where you usually sit, and label each with a location.
  • For a quiet baseline, pause nonessential transfers on devices you control; compare your normal workload separately.
  • Browsers can't read WiFi signal strength, so compare results across rooms instead.

History

DownloadUpload

Broadband help guides

Why is my WiFi slow?Practical checks for slow WiFi, from router placement to mesh. Broadband slower than advertised?Gather evidence and check the applicable complaint options. What is bufferbloat?Why calls and games lag when the line is busy. Slow internet in the evening?Find out if it's congestion and what to do. What internet speed do I need?Realistic numbers for streaming, gaming and working from home. Ping, jitter and latency explainedWhat the numbers mean for gaming and video calls. Mesh vs extender vs powerlineCompare backhaul, placement and installation trade-offs. How to test your speed properlyRecord repeat trials and understand what they can show.

Use your result to answer a practical question

Content reviewed 16 October 2026. A result is a snapshot of one device and endpoint, not a certification of the broadband line or every application. Use download and upload to assess capacity, then compare idle and loaded delay with the activity you want to improve.

  1. For a quiet baseline, use a capable wired device where practical and pause nonessential transfers. Record the server, duration, device and local time.
  2. Run several trials under comparable conditions and retain the range as well as a typical reading. Avoid running benchmarks together.
  3. Compare Wi-Fi near the router and in the affected room. Change one condition at a time so differences remain interpretable.
  4. Repeat during the actual problem. Use the result and application symptoms to choose the next investigation.

Understand the method and data cost

Download uses four concurrent requests and upload uses three. The throughput calculation excludes an initial warm-up period; it is not a prediction of every single-file transfer. Loaded latency is measured with small requests during each transfer direction. This test does not report a packet-loss percentage.

A constant 500 Mbps transfer for ten seconds would move about 625 MB in one direction alone. Actual usage varies, and startup or other requests add traffic. Check your allowance before testing; shorter duration does not impose a fixed byte limit. Do not disable required VPN or security settings simply to improve a score.

Choose the next check from the evidence

  • Good wired performance but a poor room result: investigate the wireless path, placement and client capabilities.
  • Delay rises while a backup runs: compare scheduling or rate limits, then assess supported queue management if needed.
  • Repeated quiet wired shortfalls: check device and link limits, compare endpoints and record evidence for the provider.
  • Only one application struggles: check that service’s diagnostics and status as well as the local connection.

For a repeatable log, use the testing procedure. For capacity planning, use the internet speed calculator with your actual simultaneous activities. The tool’s capability labels and diagnostic text are estimates, not guarantees.

Internet speed test FAQs

How does this internet speed test work?

The browser times small HTTP requests, then downloads and uploads over several concurrent requests. Quick, Standard and Extended choose 6, 10 or 15 seconds per direction; startup and latency sampling add time. Initial throughput warm-up is excluded. The result describes this device’s path to the selected endpoint during the trial.

Is the displayed ping the same as a game ping?

No. This tool measures browser HTTP request timings. Idle latency is the median of 20 samples after a warm-up request, and jitter is the mean absolute change between consecutive samples. A game or an ICMP utility may use a different protocol, destination and method.

What does the bufferbloat grade mean?

It grades the larger loaded-latency increase above idle using this tool’s thresholds. Queueing can contribute, but Wi-Fi contention, browser processing and endpoint conditions also matter. Treat the grade and automated insights as investigation prompts rather than a certain diagnosis.

What is a good internet speed for my home?

Budget download and upload for activities that overlap, using the requirements of the chosen services. Check file deadlines and responsiveness separately. A household size or one Mbps threshold cannot guarantee a good experience.

Why do servers and repeated tests give different results?

Routes, endpoint load, Wi-Fi and household traffic can change. Auto uses Cloudflare routing rather than guaranteeing the geographically closest server. Compare repeated trials under recorded conditions; keep trials to another endpoint labelled separately.

How much data does a test use?

The test transfers real data in both directions and can consume a large allowance on fast connections. Quick is shorter, not a data cap. The connection panel reports tracked transfer bytes, which are not an exact carrier billing measurement.

What do Mbps and MB/s mean?

Mbps is megabits per second; MB/s is megabytes per second. Eight bits make one byte, so 100 Mbps corresponds to 12.5 MB/s in decimal units. Actual file completion also depends on the application, remote service and storage.

Where are settings and results stored?

This tool saves history and settings in this browser’s local storage. Loading the page can also contact Cloudflare for approximate connection information and public LibreSpeed services for server discovery or probes. Test and service operators receive ordinary network-request information, including your IP address; local history does not mean no third-party processing.

What is included in a shared result link?

The link contains an encoded, unencrypted result with timestamp, speeds, idle and loaded latency, jitter and server name. It omits the tool’s IP field, custom room label and entered plan speeds. Anyone with the full link can read the result, including its server description; inspect it before sharing.

Can a low test prove my provider breached the contract?

No. Compare the appropriate contract estimate or minimum, use repeatable tests and follow the provider’s diagnostics. A percentage of an entered plan speed does not establish entitlement to a refund or exit.

Auto uses Cloudflare’s speed endpoints; other choices use public LibreSpeed servers. Connection location is an approximate network lookup, not GPS. Read the storage and sharing FAQs before sending a result.