Why orjson Beats ujson for Fast Python JSON Serialization

When using json.dumps(data) in Python, the built‑in json library can be surprisingly slow even for modest data structures, prompting a search for faster JSON handling libraries. The author compared three libraries: ujson (UltraJSON, written in C), python‑rapidjson (C++), and the standard json module, later adding the Rust‑based orjson after discovering its impressive performance.

Installation commands are straightforward:

pip install python-rapidjson pip install simplejson

A benchmark script ( test.py) generates a large list of dictionaries and measures the time taken by each library's dumps() function. The script imports the chosen library based on a command‑line argument and prints the elapsed time.

from time import time
import sys, string
num = int(sys.argv[1])
lib = sys.argv[2]
items = []
for i in range(num):
    items.append({c: c for c in string.ascii_letters})
start = time()
if lib == 'ujson':
    import ujson
    ujson.dumps(items)
elif lib == 'rapidjson':
    import rapidjson
    rapidjson.dumps(items)
elif lib == 'orjson':
    import orjson
    orjson.dumps(items)
else:
    import json
    json.dumps(items)
print(time() - start)

Running the script with various data sizes (1 000 to 1 000 000 items) and libraries produced timing results that match benchmarks from other sources: orjson consistently outperformed ujson, which in turn was faster than the standard json module, while rapidjson showed comparable speed to ujson.

Additional benchmark images from the original source illustrate these performance differences.

The author also notes that orjson uses Rust rather than C/C++, yet achieves comparable speed, making it a compelling choice. However, orjson 's dumps() returns bytes and does not accept the indent parameter; formatting requires decoding the bytes, e.g.:

import orjson
json_str = orjson.dumps(record, option=orjson.OPT_INDENT_2).decode()

A known bug in ujson versions 3.0.0 and 3.1.0 causes the indent argument to be ignored for values greater than 1, leading the author to recommend downgrading to ujson==2.0.3 for correct indentation support, though that version cannot serialize datetime objects.

Overall, the article concludes that while ujson is a solid fast alternative, orjson offers superior performance and should be preferred when speed is critical.