Measure download speed, upload speed, browser HTTP round-trip time (often called ping), jitter, p95 latency and short-sample request stability from your browser to GearCheck Lab's nearest available edge test node. The test is designed for practical troubleshooting: finding slow Wi‑Fi, unstable gaming latency, poor video-call performance or a connection that does not appear to match expectations.
No app installBrowser basedTemporary test dataEdge-node measurement
TEST NODE READY
0.0Mbps
READY TO TEST
Download—Mbps
Upload—Mbps
Ping / HTTP RTT—ms · browser request round trip
Jitter—ms · successful adjacent probes
P95 latency—ms · 95% of successful probes at/below
Request failures—% · HTTPS probes, not raw packet loss
—Browsing & everyday use
—Video calls & streaming
—Gaming latency
—Browser request stability
How the GearCheck Lab speed test works
A useful speed test needs more than one quick file transfer. Short tests can be dominated by connection setup, server start-up, congestion control or a single slow request. GearCheck Lab therefore warms the test path first, then measures the connection over a longer window using adaptive payload sizes and multiple parallel transfers. The goal is to estimate the usable browser-to-edge throughput rather than report a momentary burst.
HTTP round-trip latency, jitter and stability
The test sends a warm-up followed by repeated small HTTPS requests to the GearCheck Lab edge endpoint. The displayed “ping” is a trimmed median of successful browser HTTP round trips, not ICMP ping. Jitter is the median absolute change between adjacent successful probes. P95 shows the slower tail of successful latency samples, while Request failures counts timed-out or failed HTTPS probes in the short measurement window.
Request failures are not raw packet loss. TCP/QUIC retransmission can hide underlying packet loss, while an endpoint, browser or connection error can also make a request fail.
Download and upload
A short probe helps choose an appropriate transfer size and number of parallel streams. The measured stage then runs for several seconds. Download is calculated from bytes received over elapsed time; upload is calculated from temporary generated bytes successfully received by the test endpoint.
Because this is a browser test, the result includes the performance of your device, browser, local network and route to the selected edge location. It is not a laboratory certification of your ISP plan. Read our broader testing methodology for how GearCheck Lab treats browser-visible measurements and limitations.
How to understand your internet speed test result
No single number decides whether a connection is "good." Download and upload describe throughput, while ping and jitter describe responsiveness and consistency. A connection can have high Mbps but still feel poor in games or calls when latency is unstable. The ranges below are practical reference points, not guarantees for every service.
Metric
Strong result
Usable / typical
When to investigate
Download
25+ Mbps for one or a few users
10–25 Mbps for general use
Below about 5–10 Mbps if normal browsing or HD streaming struggles
Upload
8+ Mbps for calls and cloud use
3–8 Mbps for many everyday tasks
Below about 3 Mbps if calls, uploads or backups are unreliable
Ping
Under 40 ms
40–80 ms
Above 80–100 ms when interactive apps feel delayed
Jitter
Under 10 ms
10–25 ms
Above 25–30 ms when calls or games feel inconsistent
P95 latency
Close to the normal RTT
Some slower tail latency
Large repeatable gap above normal RTT can indicate spikes or congestion
Request failures
0% observed in the short sample
An isolated failed request should be retested
Repeated failures across runs deserve Wi-Fi/router/route investigation
For household internet, also consider how many devices are active at once. A 25 Mbps line may be comfortable for one person but constrained when several televisions, laptops, phones, game consoles and cloud backups share the same connection.
What internet speed do you actually need?
Browsing and work
Web pages, email and documents use relatively little sustained bandwidth. Around 10 Mbps can feel responsive for one user when latency is reasonable, but large software downloads and cloud syncing will take longer.
Streaming and video calls
HD streaming commonly needs several Mbps per stream, while 4K uses considerably more. Video calls depend on upload as well as download. Stable latency and low jitter are often more important than having extremely high headline speed.
Online gaming
Most games do not require huge Mbps once connected. Ping, jitter and packet stability matter more. A fast download result cannot compensate for a long or unstable route to the actual game server.
Do not compare Mbps alone
A 500 Mbps plan on congested Wi‑Fi can feel worse than a slower wired connection for a call or game. When diagnosing an experience problem, match the symptom to the metric: slow file transfer points toward throughput; delayed actions point toward latency; stuttering or uneven response can point toward jitter, interference or congestion.
How to troubleshoot a slow internet connection
Repeat the test
Run two or three tests rather than trusting one result. A repeated pattern is more useful than an isolated low score caused by temporary congestion or a background task.
Pause other traffic
Stop large downloads, cloud backup, game updates and high-resolution streaming. These compete for the same bandwidth and can increase latency under load.
Test close to the router
If the result improves dramatically near the router, the internet service may be fine and the problem is more likely Wi‑Fi coverage, interference or router placement.
Compare Ethernet if possible
A wired test removes much of the uncertainty introduced by wireless signal strength and congestion. A good Ethernet result with poor Wi‑Fi is strong evidence that the local wireless network needs attention.
Check another device
If only one laptop or phone is slow, investigate that device, its wireless adapter, browser extensions, power-saving settings or background software before blaming the connection.
Test at another time
If speed drops mainly in the evening, repeated peak-time congestion may be involved. Keep a few results with date and time if you need to discuss a recurring problem with your provider.
Is the problem your Wi-Fi or your ISP?
Signs the ISP connection may be okay
Ethernet speed is close to expectations.
Wi-Fi improves substantially near the router.
Only one room, floor or device is slow.
Switching between 2.4 GHz and 5/6 GHz changes the result significantly.
Signs the wider connection needs investigation
Ethernet and Wi-Fi are both consistently slow.
Several devices show the same problem.
Latency rises sharply even when local Wi-Fi signal is strong.
The issue repeats at the same time of day or after modem/router checks.
Router placement matters. Keep it away from thick walls, metal cabinets and other electronics where possible. Higher-frequency Wi-Fi bands can deliver more speed at short range but generally lose strength faster through walls; 2.4 GHz often reaches farther but may be more congested.
Why different speed tests can show different numbers
It is normal for two reputable speed tests to disagree. They may be measuring different routes rather than making the same measurement badly. The selected server can be in another city or network, one service may use many parallel connections while another uses fewer, and test duration can affect whether the connection has time to reach full throughput.
Server and route
Your packets do not travel through the internet in a straight line. Peering arrangements and routing can make one nearby-looking server faster than another. GearCheck Lab reports the browser-to-GearCheck Lab edge path; another test may use a different network entirely.
Device and browser conditions
CPU load, battery-saving modes, browser tabs, security software, VPNs and extensions can affect browser-based tests. Wi-Fi conditions can also change in seconds as other devices transmit.
For comparison, run tests close together under the same conditions and look for the broad pattern rather than expecting identical decimals. Large repeatable differences deserve investigation; small differences are expected.
Accuracy, privacy and limitations
Privacy-first measurement
The upload phase sends temporary generated test bytes, not documents, photos or other personal files from your device. Download test payloads are temporary measurement data. GearCheck Lab does not need access to your local files to run the test.
As with normal web requests, hosting and network infrastructure may process standard request information needed to deliver the service. See the privacy policy for site-level information.
What this test cannot prove
This result does not certify the maximum capacity of your ISP line, measure raw ICMP ping, diagnose packet loss at every hop, measure the latency to a specific game server, or guarantee performance in another app. Browser scheduling, device performance, Wi-Fi, VPNs, network congestion and the route to the edge node can all influence the result.
True packet-loss measurement requires a transport/test endpoint designed to expose missing packets without normal reliable-transport retransmission hiding them. GearCheck Lab therefore labels this browser metric Request failures, not packet loss.
A fair test is best run with unnecessary traffic paused, from a device that is not heavily loaded, and repeated more than once. If you are checking whether your ISP delivers an expected service tier, compare a wired result as well as Wi-Fi and follow the provider's own testing requirements.
Internet speed test FAQ
Why is my result different from Google, Ookla or another speed test?
Different services can use different servers, networks, routes, transfer sizes, durations and parallel connection counts. Your local conditions may also change between runs. Compare repeated tests under similar conditions rather than expecting identical results.
What is a good ping for gaming?
Lower is generally better. Under roughly 40 ms is often comfortable, 40–80 ms can still be playable, and higher values are more likely to feel delayed. The actual game server may have a different ping from this test, and low jitter matters too.
What does jitter mean?
Jitter describes how much latency changes from sample to sample. GearCheck Lab uses the median absolute change between adjacent successful HTTPS round-trip probes. Consistent 30 ms latency can feel smoother than latency that rapidly swings between 15 ms and 100 ms.
Is “Request failures” the same as packet loss?
No. It is the percentage of this page’s short HTTPS latency probes that timed out or failed. Reliable web transports can retransmit lost packets before JavaScript sees the response, so underlying packet loss can exist even when every request succeeds. Conversely, a browser, edge endpoint or connection error can fail a request without proving packet loss at a specific network hop.
Is this ping the same as command-line ICMP ping?
No. A normal webpage cannot send raw ICMP echo requests. This test measures browser HTTP round-trip time to GearCheck Lab’s edge endpoint. Use it for repeatable browser-path comparisons, not as a claim about ICMP latency to every destination.
Should I test using Wi-Fi or Ethernet?
Use both when possible. Ethernet is useful for checking the connection available from the router or modem without normal Wi-Fi interference. Wi-Fi then shows the real experience on the wireless network where you actually use the device.
Does the upload test send my files?
No. It creates temporary test bytes for the measurement. The browser does not select a personal file from your device.
Can a VPN make the result slower?
Yes. VPN traffic travels through the VPN provider and is encrypted. Server distance, congestion and the extra route can reduce throughput or raise latency. Compare with and without the VPN if you are trying to isolate the underlying connection.
Why is Wi-Fi slower than the speed on my internet plan?
Your plan describes the service connection, while Wi-Fi is another link between your device and router. Distance, walls, interference, router capability and the device's own radio can all reduce the speed that reaches one device.
How many tests should I run?
Two or three runs are a sensible minimum for troubleshooting. If the problem is intermittent, repeat tests at different times and keep notes about location, connection type and what else was using the network.
Related GearCheck Lab diagnostics
If your connection looks healthy but a particular activity still feels wrong, test the hardware involved in that activity as well.
