Comment

Lahache Stéphane December 20, 2017

pretty good yes. It's as fast as ujson, for the same size.
Just make "pip install cbor", and you will have the high-speed implementation (with C implementation)

For another python implementations (or other languages), see more here: http://cbor.io/impls.html

i quickly performed a bench, and it gaves me these stats:
##########
*Encoding*
json -> op by sec: 86725.66920628605, (len:828)
rapidjson -> op by sec: 180454.7353457841, (len:768)
ujson -> op by sec: 190558.97336411255, (len:776)
msgpack -> op by sec: 153912.86095711155, (len:636)
cbor -> op by sec: 207190.78744576086, (len:635)
##########
*Decoding*
json -> op by sec: 91466.83280991958
rapidjson -> op by sec: 110461.44729824932
ujson -> op by sec: 129011.48500404017
msgpack -> op by sec: 243895.1547813709
cbor -> op by sec: 168211.05628806681