JSON to TOON Converter

What Is TOON?

TOON, or Token-Oriented Object Notation, is a compact representation of JSON-compatible structured data aimed at token-sensitive workflows such as LLM prompts, AI agents, and model tool calls. It uses a tabular representation for repeated object arrays while keeping a readable form for nested data.

TOON is not a universal replacement for JSON. JSON remains the standard format for APIs, browsers, databases, and general application interoperability. TOON is an optional representation to test when reducing LLM input size matters.

How to Compare JSON and TOON

1. Enter JSON or TOON, depending on the direction you need.
2. Convert and read the generated output.
3. Compare TOON against minified JSON by characters, UTF-8 bytes, and estimated tokens.
4. Read the notices: whether TOON is actually smaller, and whether a uniform object array was found.
5. Round-trip with Swap Direction to confirm the data survives.
6. Test comprehension: confirm that the target model and prompt instructions interpret the TOON output correctly.

Why Uniform Arrays Can Benefit

A repeated array of objects spells out the same field names many times in JSON:

{"users":[{"id":1,"name":"Alice","role":"admin"},{"id":2,"name":"Bob","role":"editor"}]}

The tabular form declares the fields once and writes the values as rows:

users[2]{id,name,role}:
  1,Alice,admin
  2,Bob,editor

This is the main structural situation in which TOON removes repeated syntax.

TOON Is Not Always Smaller

Size depends on the shape of the data. Uniform arrays may benefit, while deeply nested objects, sparse records, mixed arrays, or configuration objects can produce little difference or increase the output. That is why this page reports the measured result and warns when TOON is larger, instead of repeating a fixed savings range.

Characters, Bytes, and Tokens Are Different

Characters are what the editor shows, UTF-8 bytes are what is transmitted or stored, and tokens are what a model bills. A representation can be smaller in characters but not in tokens, because tokenizers split text in their own way. Compare the metric that matches your cost.

About the Token Estimate

This page estimates tokens by dividing the character count by four and rounding up, using the same formula for both sides. That is a rough comparative heuristic, not a count from a specific model tokenizer. Real counts differ between models and tokenizers, so verify with your own tokenizer before making cost decisions.

What This Page Supports

The converter handles objects and nested objects, primitive arrays in inline form, uniform arrays of objects in tabular form, non-uniform and mixed arrays in list form, empty arrays and empty objects, root primitives, root objects and root arrays, and comma, tab, or pipe delimiters. It does not implement the keyed tabular root form, nested field groups, key folding, or path expansion.

Round-Trip and Validation

The TOON reader checks that declared array lengths match the number of rows or items, so a document with a wrong count is reported rather than silently decoded. Use Swap Direction to feed the output back and check that your data survives a round trip, paying attention to strings that contain delimiters, colons, quotes, or leading and trailing spaces.

Privacy and Browser-Based Processing

Conversion and comparison happen in the browser, so pasted payloads stay on your device. That suits prompts and development data, but it is not a secret-management system. Avoid entering credentials, private prompts, or confidential production payloads into any online tool.

When to Use Local Code

For repeated pipelines, very large payloads, or confidential data, run the conversion in your own environment where you control logging, token counting, and data retention. The browser tool is best for inspection, one-off comparison, and deciding whether TOON is worth adopting for a given payload shape.