A compact notation for structured data in prompts. It drops quotes around keys, braces, and commas and uses indentation for nesting, so a pretty-printed JSON payload usually gets shorter when converted.

How are the token savings estimated?

From the characters-per-token profile you select: generic at 4 characters per token, dense code at 3, and prose-heavy at 5. It is a planning estimate, not a tokenizer run, so actual counts vary by model.

Will an LLM understand TOON?

Models generally read indented key: value text well, but test your own prompts and check the answers before switching a production workflow. If the model must return JSON, ask for JSON output explicitly.

Can I convert TOON back to JSON?

Yes. Choose TOON to JSON as the conversion mode to rebuild the JSON with the indent size you set.

Is my JSON sent anywhere?

No. The conversion runs entirely in your browser and nothing is uploaded.

JSON to TOON converter

TOON (token-oriented object notation) represents JSON-like data with indentation instead of braces, quotes, and commas, so the same data takes fewer tokens in an LLM prompt.

Convert JSON to TOON, a compact notation for LLM prompts, convert TOON back to JSON, and estimate the token savings. Free and runs in your browser.

Browser-sideNo accountCopy, download, or print

Last reviewed 30 Oct 2026

Your inputs are saved as a draft in this browser only. Nothing is uploaded.

Jump to result

Result

Input tokens46

Output tokens45

Estimated savings1 tokens

Savings rate2.17%

{
  user: {
    id: 123
    name: "Asha Rao"
    roles: [
      "admin"
      "developer"
    ]
    active: true
  }
  settings: {
    theme: "dark"
    notifications: true
  }
}

{
  user: {
    id: 123
    name: "Asha Rao"
    roles: [
      "admin"
      "developer"
    ]
    active: true
  }
  settings: {
    theme: "dark"
    notifications: true
  }
}

How to use it

Choose the conversion mode: JSON to TOON, or TOON to JSON to convert back.
Paste your JSON or TOON payload into the input.
Pick the token estimate profile that matches your content: generic, dense code, or prose-heavy.
Generate the output and check the estimated token savings, then copy or download it.

Questions

What is TOON format?: A compact notation for structured data in prompts. It drops quotes around keys, braces, and commas and uses indentation for nesting, so a pretty-printed JSON payload usually gets shorter when converted.
How are the token savings estimated?: From the characters-per-token profile you select: generic at 4 characters per token, dense code at 3, and prose-heavy at 5. It is a planning estimate, not a tokenizer run, so actual counts vary by model.
Will an LLM understand TOON?: Models generally read indented key: value text well, but test your own prompts and check the answers before switching a production workflow. If the model must return JSON, ask for JSON output explicitly.
Can I convert TOON back to JSON?: Yes. Choose TOON to JSON as the conversion mode to rebuild the JSON with the indent size you set.
Is my JSON sent anywhere?: No. The conversion runs entirely in your browser and nothing is uploaded.