JSON to TOON:
Converter

What TOON looks like

A JSON list such as {"users":[{"id":1,"name":"Ali"},{"id":2,"name":"Sara"}]} becomes a header line, users[2]{id,name}:, followed by one row per object: 1,Ali and 2,Sara. Nested objects are written with indentation, like YAML, and simple lists stay on one line.

The [2] tells a model how many rows to expect, and {id,name} names the columns once instead of on every row. That repetition is what TOON saves.

When it saves the most

• Arrays of objects that share the same fields, such as rows from a database, a spreadsheet or an API list.
• Data you paste into ChatGPT, Claude or another model many times, where every token counts.
• Deeply nested or very irregular data saves less, and can occasionally be longer than minified JSON. Compare both if size matters.

Settings

Separator chooses how values in a row are split: comma (the default), tab or pipe. Tabs are often a single token in common tokenizers, so they can save a little more; pipes are handy when your values contain many commas. Values that contain the separator are quoted automatically.

Is the data the same?

Yes. Strings, numbers, booleans, null, objects and arrays all keep their values, and TOON to JSON turns the file back into the same data. Key order is kept. Only formatting changes.