Internet Stability Test
Leave this page open and it checks your connection every 3 seconds, logging each drop, slow spell and unreachable service. After an hour or a day you get uptime, latency and jitter, plus a report you can send to your internet provider.
Internet stability monitor
- Press Start monitoring. Keep this tab open, ideally in its own window.
- Watch the status: Online, Degraded or Down. Every drop and slow spell goes into the event log.
- After an hour or more, open Report and download it as HTML, CSV or JSON for your provider.
Monitoring uses about 13 MB of data per hour with the default settings (about 0.3 GB a day). Speed tests are off unless you start one; each uses up to 37 MB. How we measured this
Monitoring is stopped. Press Start monitoring to check your connection every few seconds. Browsers slow down background tabs, so keep this tab visible or in its own window.
Latency median, last 60 s How long a small HTTPS request takes to come back from Google and Cloudflare, the middle value of the last minute. Lower is better.
—
Failed requests last 5 min The share of heartbeat requests that timed out or got no answer. A browser cannot count lost packets, so this is not packet loss, but both rise when the line drops.
—
—
Jitter last 60 s How much the latency changes from one request to the next, on average. High jitter makes calls choppy even when the average ping looks fine.
—
—
Outages today An outage starts when two heartbeats in a row fail on every heartbeat target, or the browser reports it is offline. It ends at the first heartbeat that gets an answer.
0
no downtime
Degraded today Time when the line was up but poor: over the last 30 seconds, 20% or more of heartbeat requests failed or the average latency was above 800 ms.
0 s
—
Last speed test Download and upload in Mbps from the last Data saver speed test. A "≥" means the data cap ended the test first, so your line is at least that fast.
—
not run yet
Call quality MOS An estimate from 1 to 4.5 of how a voice call would sound, worked out from latency, jitter and failed requests in the last 5 minutes (a simplified E-model). The letter is our call-quality grade from A to F, not the bufferbloat grade; Help lists its limits.
— –
—
Core services up Seven large services on different networks (Google, Cloudflare, AWS, Microsoft, Facebook, Apple and Cloudflare DNS). If all of them fail at once, the problem is your line, not one website.
—
waiting for the first round
Live timeline
Services
Heartbeat latency (an HTTPS round trip, not ICMP ping). Red bands mark outages, amber bands degraded periods, and red marks at the bottom failed requests. The event log below lists the same periods as text.
Services
Every round (30 seconds by default) tests each service once. Select a name for its own chart, figures and incidents.
Speed test
Uses Data saver mode: up to 25 MB down and 10 MB up to Cloudflare's nearest server. On lines faster than about 60 Mbps download (or 25 Mbps upload) it stops at the cap and shows the speed as "at least". For a full-speed result use the internet speed test. Automatic speed tests are off.
No speed test yet in this session.
- Download How fast data reached this browser from the test server, in megabits per second. The line below gives the same speed in megabytes per second. Mbps
- Upload How fast this browser sent data to the test server. Video calls and backups need upload. Mbps
- Idle latency
- —
- Idle jitter
- —
- Latency while downloading
- —
- Latency while uploading
- —
- Bufferbloat How much the latency rises while the line is busy, with our A to F grade: A under 30 ms, B under 60, C under 150, D under 400, otherwise F.
- —
- Test server
- —
- Data used
- —
- Tested at
- —
Speed test history
| When | Down | Up | Idle | Loaded | Bufferbloat | Data | Server |
|---|---|---|---|---|---|---|---|
| No speed tests yet. | |||||||
Event log
No events yet. Outages, slow periods, service problems and speed tests will appear here. Clearing the list does not delete any data.
Report for your provider
Choose a period and press Generate report. The report opens here; download it as one HTML file, print it to PDF, or copy a short text summary for an e-mail or chat with your provider.
How to read your result
The status shows what your line is doing now; the event log keeps every change with its time and length.
- Down: two heartbeats in a row get no answer from any target, or the browser reports it is offline. It ends at the first answered heartbeat.
- Degraded: the line is up, but over the last 30 seconds 20% or more of heartbeats failed or average latency was above 800 ms.
- Service incident: one service fails three rounds in a row while your line is up. That points to the site or its route, not your connection.
- Uptime: the share of monitored time that was not an outage. Time with the page closed does not count.
| Figure | Healthy | Minor issues | Real problems |
|---|---|---|---|
| Uptime | 100% | 99.5% or more, with outages | under 99.5% |
| Failed requests | under 0.5% | 0.5–2% | 2% or more |
| Average latency | 150 ms or less | 150–300 ms | over 300 ms |
| Jitter | 30 ms or less | over 30 ms | – |
TRAI asks fixed-broadband providers to keep latency at or below 50 ms, jitter at or below 40 ms and packet loss at or below 1% in their own network tests (TRAI quality of service regulations). Our figures are measured differently, from your browser, so use them as a guide, not as a regulatory measurement.
What makes a connection drop or slow down
Wi-Fi causes most short drops at home. If the log shows many brief outages, run it again on a LAN cable; the YouTube Help Center also suggests a wired connection when streaming struggles. If the drops stop, check your Wi-Fi room by room.
A busy line comes next: a big download can push latency from tens to hundreds of milliseconds, which shows here as a degraded period. Measure your ping under load. If calls cut out or voices sound robotic, jitter or failed requests are the usual signs, so leave the monitor running while it happens.
Keep the tab open while it runs
The monitor measures only while this tab is open. Browsers slow down hidden tabs and a sleeping laptop stops everything, so give the page its own window and keep the device awake. Gaps are marked in the log and never count as outages. Monitoring resumes by itself when you reopen the page.
Send a report to your provider
A report turns hours of monitoring into evidence a support agent can read quickly: every outage and slow period with its times, uptime, hourly latency and failed requests, per-service figures and speed tests.
- Download it as one HTML file or print it to PDF, and attach it to your complaint. Add the CSV files if the agent wants raw data.
- When the log shows drops, run a traceroute to the same site. Loss that starts at one hop and continues to the end shows where the problem begins.
- For an official record you can show your provider, also run TRAI MySpeed.
How much data it uses
With the default settings, monitoring used about 13 MB per hour (about 0.3 GB a day), measured on a fibre connection in India, September 2026. Speed tests are off by default; each one uses Data saver mode and up to about 37 MB, so an hourly test adds up to about 0.9 GB a day. A longer round interval in Settings uses less data.
Where your results are stored
Everything stays in this browser's local database: raw samples for 7 days, hourly summaries, events and speed tests for a year. Nothing is sent to our server. The services checked see your IP address, as on any visit. Delete all monitor data in Settings at any time.
How this test works
Every 3 seconds a background worker sends a tiny HTTPS request to Google and to Cloudflare's 1.1.1.1. Every 30 seconds it also fetches a small file from each of 27 services. Each figure is an HTTPS round trip, not an ICMP ping, and a failed request is a timeout or no answer, not a counted lost packet.
The optional speed test uses Data saver mode against Cloudflare's nearest server (25 MB down, 10 MB up). On lines faster than about 60 Mbps download (or 25 Mbps upload) it stops at the cap and shows "at least"; for fibre plans use the internet speed test in Gigabit mode. The tab must stay open, and sleep or background throttling leaves gaps. Read the full method.
Questions about the internet stability test
Is this a good internet stability test for gaming?
The stability test shows whether your line drops or its latency jumps over hours. It does not contact game servers; to time cloud regions in Mumbai, Hyderabad and Singapore, use the gaming ping test.
Can this test packet loss?
A browser cannot count lost packets, so the monitor counts failed requests: requests that time out or get no answer. When a line loses many packets, failed requests rise too, so a high figure is still a warning sign.
Why is my internet slow at night, and can this show it?
Leave the monitor running through the evening on a LAN cable, with no big downloads in the house. The report's hourly pattern shows whether latency and failed requests rise at the same hours. A pattern that repeats every evening is worth sending to your provider.
Why does my internet drop suddenly for a minute?
Check the event log. If every service fails at the same moment, the problem is your Wi-Fi, router or line; if only one service fails, it is that site or its route.
How long should I run the stability test?
Run the stability test for at least an hour to see a pattern, and through a whole evening or day for drops that come and go. The heartbeat checks every 3 seconds, so even short drops are caught while the tab is open.