Performance
Traffic climbs. Response time doesn’t.
99 of every 100 requests come back in under 15 milliseconds, whether ten people are online or a launch-day rush just landed.
Latency under loadp99 under 15 mstrafficresponse timeload up to 2× target
What you can count on
Fast, steady and always there.
Four promises, each one measured on the live service.
[ 01 / 04 ]·Latency
15ms
Answers in milliseconds. Every time.
99 of every 100 API requests come back in under 15 ms, even at peak load.
[ 02 / 04 ]·Reliability
0dropped
Nothing gets dropped.
Not a single failed request in nearly 600,000 at peak, and not one failed scrape in more than 4,500.
[ 03 / 04 ]·Surges
10ms
A surge doesn’t slow you down.
Hundreds of users can arrive in seconds and responses stay just as fast, with no slowdown after the rush.
[ 04 / 04 ]·Scrapes
<1s
Pages back in under a second.
Most pages are fetched, converted to clean markdown and stored in under a second.
Built for load
Room to grow, built in.
Ready for launch-day traffic
The edge keeps warm connections to every service behind it, so a rush of visitors never waits on a handshake.
Scrapes run side by side
Jobs fan out across parallel workers and real browsers. They run next to each other, never in a line.
Headroom at every step
We pushed load to twice our target and response time stayed flat. The ceiling is still above us.
Bring your traffic. We have the headroom.
Crawl, parse and search at full speed from the first request.