API latency, measured live.
This page shows our monitor's timings of a real API request for an address already in our result cache, failures included. You can also time the API from your browser.
POST /v1/validatecached result · every 60 s
API response time: cached result
Measuring…Measuring…ms
Loading the latest measurements…
p50Median- – ms
- Last 60 minutes. Half of the requests were faster.
p9595th percentile- – ms
- Last 60 minutes. 19 in 20 were faster.
p9999th percentile- – ms
- Last 24 hours. 99 in 100 were faster.
- Probes, last 24 h
- –
- Failures are shown, never dropped. failed, not from the cache.
Response time
Milliseconds, UTC. One probe a minute; each point covers the 60 minutes before it, every 10 minutes.
Show the data as a table
No data loaded yet.
Test from your browser
Times 5 requests from your browser to GET /v1/latency, a cache lookup on our API. Server time (the shorter bar inside each bar) is what the API reports; the rest is your network. No credits are used.
| Run | Your round trip | Server time | Entry point |
|---|---|---|---|
| 1 | – | – | – |
| 2 | – | – | – |
| 3 | – | – | – |
| 4 | – | – | – |
| 5 | – | – | – |
The API runs in one US East region, so your round trip includes the distance between you and us. The first run also opens the connection.
How we measure
- What is timed
- One real, authenticated
POST /v1/validateevery 60 seconds, over HTTPS through the public API host. - The address
- A fixed test address on a reserved .invalid domain that belongs to nobody. Its result is in our 12-month cache, so this times our API, not a mail server.
- Cache check
- Each answer is checked to be a cache hit. Percentiles (nearest rank) count hits only; misses are shown apart.
- From where
- Cloudflare's network to our API, which runs in one US East region. Your own requests add your distance to us.
- What it leaves out
- A first-time check waits for the mail server, up to 15 seconds; if it runs out of time, it is unknown and free.
- Cost
- A cached answer costs 1 credit, like any completed check. The monitor pays for its probes.
- Every probe counts
- Failed probes stay in the chart and in the counts.
What this page shows, including when it can't.
The readout is always in one of these states. It never shows a figure it did not measure.
Live
A measurement arrived in the last few minutes. The big number is the median of the last 60 minutes.
Measuring
Loading, or nothing published yet. We never show a placeholder number.
Stale
No new measurement for a few minutes, so no live figure. Check the system status.
Failed
The last measurement failed. We show failures too, in the chart and the counts.
Measure it yourself
Time it from your own machine. Use your own key.
Run it twice with the same address: when the first answer is a definite one (not unknown, catch-all or mailbox full), the second comes from our result cache. time_total is in seconds and includes the distance between you and our API in US East.
curl -s -o /dev/null -w "%{time_total}\n" \
-H "Authorization: Bearer $EV_API_KEY" \
-H "Content-Type: application/json" \
-d '{"email":"name@example.com"}' \
https://api.email-validation.org/v1/validateEach request uses 1 credit, cached or not; unknown results are free. Sandbox test addresses are answered on the spot, so they do not time the cache.
Built for speed
Fast on purpose. Measured, not promised.
A first-round-trip budget
Every build checks our home page, the free checker and this page against 14,600 bytes with headers: about what arrives in the first network round trip.
No third-party requests
Our pages load nothing from other companies' servers. The bot check loads only when a form needs it.
Nothing blocks the first paint
No external stylesheet, font download or blocking script before you see the page.
Next page, already loaded
Pages you are likely to open next are fetched before you click, in browsers that support it.
Repeat checks come back at once
Checked recently? The answer comes from our result cache, with no mail server to wait on.
A 15-second ceiling
No check runs past 15 seconds. Slower ones are unknown and free.
Up to 500 requests per second
Per account after your first purchase; 20 per second before that.
One pass through your list
Up to 1,000,000 rows. Each unique address is checked once, with no second pass.
No uptime percentage is shown until there are enough measurements to back it.
Updated · Maintained by the email-validation.org team