JSON to JSONL Converter.

Convert JSON to JSONL online - free and private

A JSON to JSONL converter turns a normal JSON array into newline-delimited JSON. Paste an array above, and each item becomes one compact line. That output can be saved as .jsonl or .ndjson for logs, streaming tools, data imports, ML datasets, and ETL pipelines.

How to convert JSON to NDJSON in three steps

1. Paste a JSON array into the converter, or load a local file.
2. Convert to JSONL so every array item is serialized as one compact line.
3. Copy or download the newline-delimited output as data.jsonl.

Array to newline-delimited JSON

JSONL and NDJSON do not use surrounding brackets or commas between records. Instead, every line is a complete JSON value. This tool strips pretty-print indentation and writes one compact object per line, which makes the output streamable, appendable, and easy for command-line tools to process.

What the conversion does

A JSON array:

[
  { "id": 1, "name": "Ada" },
  { "id": 2, "name": "Alan" }
]

becomes one compact record per line:

{"id":1,"name":"Ada"}
{"id":2,"name":"Alan"}

The brackets and the separating commas disappear, and each record is minified onto a single line. That last part is not cosmetic — it is what the format requires.

Why each record must be one line

The entire point of JSONL is that a consumer can read it line by line without parsing the whole file. That only works if one line is exactly one record.

Pretty-printed JSON spanning multiple lines would break it completely — a line-by-line reader would see a lone opening brace as its first record and fail immediately. So the converter strips the formatting automatically. A JSONL file is not meant to be pretty; it is meant to be streamed.

Who wants JSONL

• OpenAI fine-tuning — training files must be JSONL, one example per line.
• BigQuery and Redshift — newline-delimited JSON is the expected bulk-load format.
• Elasticsearch — the bulk indexing API consumes exactly this.
• Log pipelines — Fluentd, Logstash, and most shippers emit and expect it.

The common thread: all of them want to start processing record one without waiting for the file to finish, and all of them want to append without rewriting a closing bracket.

The trailing newline

There is a toggle for it because some strict parsers require one, and it is good practice regardless. A final newline means every record is terminated consistently — so a process appending the next line does not accidentally weld it onto the last one and corrupt both records.

Need to go the other way? The JSONL to JSON converter wraps the lines back into an array, validating each one and reporting the exact line if any fail.

Why use this JSON to JSONL tool?

It is fast, private, and built for the exact JSONL format. Use it when an API gives you a JSON array but your logging, analytics, ML, or import workflow expects newline-delimited JSON. Need the reverse? Use JSONL to JSON, or read what JSONL and NDJSON are.