TOON is a lossless encoding of the JSON data model. Every value — strings, numbers, booleans, nulls, objects, and arrays — round-trips perfectly back to JSON without any data loss or type coercion.
TOON's compact tabular format (key[N]{fields}: rows) is fully decoded back into standard JSON arrays of objects, restoring the full structure from minimal token representation.
If you send TOON to an LLM and receive TOON back in the response, use this tool to decode it to standard JSON for further processing by APIs, databases, or application code.
Decoding runs entirely in your browser — no data is transmitted to any server. Safe for TOON payloads containing sensitive or proprietary structured data.
TOON (Token-Oriented Object Notation) is a token-efficient format for sending structured data to LLMs. After an LLM processes or returns TOON, you decode it back to JSON to feed it into standard APIs, databases, or application code that expects JSON.
Yes. TOON encodes the same data model as JSON — objects, arrays, strings, numbers, booleans, and nulls — with no information loss. A round-trip JSON → TOON → JSON produces bit-identical output for all representable values.
A TOON tabular header like users[3]{id,name,role}: followed by three comma-separated rows expands to a standard JSON array of three objects, each with id, name, and role fields.
The TOON specification and a reference parser are available at toonformat.dev. The format is simple enough that a custom parser can be written in under 200 lines in most languages. This online tool is the fastest option for manual inspection and one-off decoding.