HomeAll ToolsSpeed Test

Network Latency (Ping) & Stability

Monitor Ping, Jitter, and Packet Loss in real-time.

state:onlinetest:Not running/completed
Stability Score
Poor
NaN
A heuristic score (0–100) that combines packet loss, p95 latency, jitter, and offline events.
Recent RTT (lower is better)
No data yet
avg-
p95:-
Jitter0ms
Packet Loss0%

Diagnosis

No major issues detected.
Quick troubleshooting suggestions
  • Prioritize using network cables or 5GHz Wi‑Fi, stay close to the router, and avoid interference sources such as microwave ovens/Bluetooth.
  • Turn off/switch VPN, proxy, accelerator; check whether company network policy (DPI/firewall) causes reset or packet loss.
  • Pause bandwidth-consuming tasks (cloud disk synchronization, downloads, system updates), and observe whether p95/packet loss is significantly improved.
  • If only one site is "stuck", it is more likely to be the other party's server/cross-border link/DNS; you can copy the report and send it to the other party for troubleshooting.

Network information and events

Network Information API
Not supported/unavailable
Some browsers/environments do not expose this API. Doesn't affect ping tests, but is missing some "network type/estimated RTT" supplementary information.
Event (recent)
Keep up to 200 entries
No data yet
Original sample (most recent)
No data yet

How to use this page to quickly locate problems

The key to "stuck" is p95 and jitter: avg is very low but p95 is very high, which is often caused by physical stuttering caused by occasional spikes (more obvious in meetings/games).
The key to "disconnection" depends on the failure rate and offline events: If the failure rate increases or offline events occur, it is usually due to Wi‑Fi interference, router reconnection, VPN/proxy instability, or operator jitter.
First use the default endpoint /api/ping for same-domain testing; only when you clearly want to test a service, change it to the health check/static resource direct link of the service.
When copying the report to the other party for troubleshooting, focus on: sampling times, packet loss rate, p95/p99, offline events, and the timeline of the last 10 events.

Network Guide

Run continuous pings to detect intermittent lag spikes.

Step 1

Start testing (same domain ping)

About 10–20 seconds

Keep the default /api/ping and click "Start Test".

Keep the default endpoint /api/ping (same domain, best reflects the network quality of "you come to this site")
Set the number of sampling times (it is recommended to start with 20; for more stability, 50+ can be used)
Set the interval and timeout (too small an interval may amplify instantaneous jitter; too small a timeout may cause misjudgment failure)
Click "Start Test" and wait for the sampling to complete
Tip: The same-domain test is more like a comprehensive sense of "network link + DNS + local server"; to test a certain business, please change the endpoint to the health check or static resource direct link of the business.
Step 2

Read the results (stuck vs broken)

about 20 seconds

Focus on packet loss rate, p95, jitter and offline events.

"Stuck": Prioritize p95 delay and jitter (MAD/Std). The more spikes, the worse the somatosensory experience.
"Disconnect": Prioritize the failure rate and offline events. The occurrence of offline is often caused by link interruption/reconnection.
Observe the "event" timeline: whether there are prompts such as network switching, VPN switching, etc.
Step 3

Copy the report for troubleshooting

about 5 seconds

Copy text reports containing key statistics and events with one click.

Click "Copy Report"
Send the report to the network administrator/colleagues/customer service to quickly locate the problem based on time points
If it is "only a certain site is stuck", it is recommended to change the endpoint to a direct link to the site resource and run the comparison again.

FAQ

A summary of frequently asked questions about network diagnosis and "stuck/outage" troubleshooting.

1.

What exactly is the "network" tested on this page?

It estimates RTT (round trip delay), jitter, and failure rate by making multiple requests to the specified endpoint (similar to ping sampling), and combines browser online/offline and Network Information API events to help you quickly determine whether it is "stuck" or "disconnected."

2.

Why is the default endpoint /api/ping?

By default, using the same domain /api/ping can reduce the differences between cross-domain and third-party servers, and is closer to the real link quality of "you to this site", and is more stable and reproducible. To diagnose a certain business/site, change the endpoint to its health check or static resource direct link for comparison.

3.

Where does the "stability score" come from? Is it reliable?

The scoring is heuristic on a 0–100 scale: combined packet loss/failure rate, p95 latency, jitter (adjacent RTT variation), and offline events give an intuitive result. It is suitable for quick comparison (change network/change VPN/before and after approaching the router) and is not equivalent to a strict network measurement instrument.

4.

avg is low but p95 is high, what does it mean?

Usually represents "sporadic spikes": fast most of the time, but a few very slow requests often occur. From a somatosensory perspective, it is easy to experience occasional lags in conference voice, occasional circling of web pages, and momentary frame drops in games, etc.

5.

Does an increase in packet loss/failure rate necessarily indicate a network problem?

uncertain. The failure may come from the network (Wi‑Fi interference, router reconnection, VPN instability), it may come from the endpoint server (overload/throttle/failure) or the browser timeout setting is too small. Suggestion: Test the endpoint in the same domain and the target endpoint once each, and increase the timeout appropriately for comparison.

6.

Will the test upload my private data?

Won't. It will only initiate a request to the endpoint you fill in and record the time-consuming/successful failure and other indicators; "Copy Report" just writes the statistical text to the clipboard.

7.

Why does it sometimes show "Network Information API is not supported"?

Some browsers/environments do not expose information such as navigator.connection due to privacy or implementation restrictions. This does not affect ping sampling and result judgment, but only lacks auxiliary information such as network type/estimated RTT.

8.

What are the most common troubleshooting suggestions?

Prioritize three steps of comparison: 1) Turn off/change VPN; 2) Get close to the router and switch to 5GHz or switch to a network cable; 3) Pause bandwidth-consuming tasks such as downloading/synchronization. Running a test every step of the way is the easiest way to identify the main bottlenecks.