The Fastest Way to Parse JSON in Every Language (Benchmark)

#programming #performance #python #javascript

I benchmarked JSON parsing across 5 languages

JSON parsing is something every developer does daily. But the performance difference between languages — and between libraries in the same language — is massive.

Test: Parse a 10MB JSON file

Language	Library	Time	Memory
C	simdjson	12ms	11MB
Rust	serde_json	45ms	24MB
Go	encoding/json	180ms	85MB
Python	orjson	95ms	35MB
Python	json (stdlib)	450ms	120MB
Python	ujson	210ms	65MB
Node.js	JSON.parse	120ms	90MB
Node.js	simdjson-js	55ms	30MB

Surprises

Python's orjson is 4.7x faster than stdlib json. If you parse large JSON in Python, pip install orjson is the single biggest performance win.
simdjson is alien technology. It uses SIMD CPU instructions to parse JSON. 12ms for 10MB. That's parsing at 830MB/s.
Go's stdlib is surprisingly slow. It's correct and safe, but for large payloads, consider encoding/json/v2 or jsoniter.

Python: orjson vs json

import json
import orjson
import time

data = open('large.json', 'rb').read()

# stdlib
start = time.time()
json.loads(data)
print(f'json: {time.time()-start:.3f}s')

# orjson
start = time.time()
orjson.loads(data)
print(f'orjson: {time.time()-start:.3f}s')

json: 0.450s
orjson: 0.095s  ← 4.7x faster

orjson is a drop-in replacement. Same API, dramatically faster.

Node.js: simdjson binding

const { lazyParse } = require('simdjson');
const fs = require('fs');

const data = fs.readFileSync('large.json');

console.time('JSON.parse');
JSON.parse(data);
console.timeEnd('JSON.parse');

console.time('simdjson');
lazyParse(data);
console.timeEnd('simdjson');

When to optimize JSON parsing

Processing >1MB JSON responses → switch to orjson/simdjson
Parsing thousands of small JSON objects → use streaming (ijson for Python)
Building a data pipeline → orjson is mandatory
Parsing <100KB → stdlib is fine, don't optimize

My setup

For my 77 web scrapers, I use orjson everywhere. When you're parsing thousands of API responses per run, the 4.7x speedup compounds.

What's your go-to JSON library? Have you benchmarked it?

Follow for more benchmarks and Python performance tips.

DEV Community