Benchmarks

Local JSON server comparisons against scrapy (10,000 requests per run). The scripts used for these tests are available in this Gist; results below are frozen from those runs. Both frameworks ran with uvloop instead of the default event loop.

Overall, aioscraper throughput stays flat across CPython 3.11–3.14, while scrapy speeds up on newer interpreters but remains far slower once payload size grows.

Endpoint	3.11	3.12	3.13	3.14
/json?size=10	13.7x	12.4x	10.6x	10.6x
/json?size=100	73.2x	71.7x	50.7x	40.1x
/json?size=10&t=0.1	5.2x	4.7x	2.2x	2.2x

Endpoint breakdown

Legend: columns show elapsed (seconds), rps (requests/sec), and ips (items/sec) for each library.

/json?size=10

10,000 requests / 100,000 items

Python	aioscraper	scrapy	Gap
3.11	2.1 s / 4,761.9 rps / 47,619.0 ips	28.8 s / 347.2 rps / 3,472.2 ips	13.7x
3.12	2.1 s / 4,761.9 rps / 47,619.0 ips	26.1 s / 383.1 rps / 3,831.4 ips	12.4x
3.13	2.1 s / 4,761.9 rps / 47,619.0 ips	22.2 s / 450.5 rps / 4,504.5 ips	10.6x
3.14	2.1 s / 4,761.9 rps / 47,619.0 ips	22.2 s / 450.5 rps / 4,504.5 ips	10.6x

/json?size=100

10,000 requests / 1,000,000 items

Python	aioscraper	scrapy	Gap
3.11	2.9 s / 3,448.3 rps / 344,827.6 ips	212.2 s / 47.1 rps / 4,712.5 ips	73.2x
3.12	2.8 s / 3,571.4 rps / 357,142.9 ips	200.7 s / 49.8 rps / 4,982.6 ips	71.7x
3.13	2.8 s / 3,571.4 rps / 357,142.9 ips	142.0 s / 70.4 rps / 7,042.3 ips	50.7x
3.14	2.8 s / 3,571.4 rps / 357,142.9 ips	112.4 s / 89.0 rps / 8,896.8 ips	40.1x

/json?size=10&t=0.1

10,000 requests / 100,000 items, +100 ms delay

Python	aioscraper	scrapy	Gap
3.11	10.4 s / 961.5 rps / 9,615.4 ips	54.1 s / 184.8 rps / 1,848.4 ips	5.2x
3.12	10.4 s / 961.5 rps / 9,615.4 ips	49.0 s / 204.1 rps / 2,040.8 ips	4.7x
3.13	10.4 s / 961.5 rps / 9,615.4 ips	22.8 s / 438.6 rps / 4,386.0 ips	2.2x
3.14	10.3 s / 970.9 rps / 9,708.7 ips	22.7 s / 440.5 rps / 4,405.3 ips	2.2x