Token-Oriented Object Notation offers 30-60% token reduction and 4.8x faster parsing. Here's why that matters and how to use it.
The Problem with JSON (That You Probably Didn't Know Existed)
JSON is everywhere. It's the backbone of modern web development, API communication, and data storage. But here's the thing: JSON is verbose.
Every quote, comma, and brace adds up. In AI/LLM applications, this verbosity translates directly to higher costs and limited context windows. When you're paying per token, those extra characters matter.
Enter TOON (Token-Oriented Object Notation) — a format that promises to solve these problems while maintaining (or improving) human readability.
What Makes TOON Different?
1. Massive Token Reduction
TOON achieves 30-60% fewer tokens compared to JSON. Let's see why:
JSON:
{
"site": {
"id": "site_123",
"name": "My Awesome Site",
"pages": [
{
"id": "page_1",
"title": "Home",
"components": [
{
"type": "hero",
"props": {
"title": "Welcome",
"subtitle": "Get started today"
}
}
]
}
]
}
}
TOON:
site
id: site_123
name: My Awesome Site
pages
-
id: page_1
title: Home
components
-
type: hero
props
title: Welcome
subtitle: Get started today
Notice the difference? No quotes around keys. No commas. No braces. Just clean, indentation-based structure. The TOON version uses approximately 40% fewer tokens.
2. Faster Parsing
TOON isn't just more compact — it's also up to 4.8x faster to parse. This matters for real-time applications, high-throughput systems, and any scenario where performance counts.
3. Better Readability
The indentation-based syntax makes TOON easier to read and edit manually. No more hunting for missing commas or mismatched braces. The structure is immediately clear.
Real-World Impact
I've been testing TOON across several projects, and the results are impressive:
The Imaginatorium (narrative/event storage): 45% token reduction
VIBE CHAT (chat log storage): 38% token reduction
CML Quest (game content format): 42% token reduction
Mini-Cursy (telemetry logging): 50% token reduction
When you're processing thousands of records, these percentages translate to significant cost savings and more data per context window.
Getting Started: A Working Parser & Converter
The good news? You don't have to wait for ecosystem support. I've built a fully functional TOON parser and converter that you can use right now.
🚀 Live Demo
Try it yourself: https://futurevision-labs.github.io/toon-parser/
The demo shows real-time conversion between JSON and TOON, complete with token count comparisons and reduction percentages.
📦 Installation
npm install toon-parser
💻 Quick Start
const ToonConverter = require('toon-parser');
// Convert JSON to TOON
const json = {
user: {
name: "John Doe",
email: "[email protected]",
settings: {
theme: "dark",
notifications: true
}
}
};
const toon = ToonConverter.jsonToToon(json);
console.log(toon);
// Output:
// user
// name: John Doe
// email: [email protected]
// settings
// theme: dark
// notifications: true
// Convert TOON back to JSON
const backToJson = ToonConverter.toonToJson(toon);
// Returns the original JSON object
🎯 Key Features
- ✅ Bidirectional conversion (JSON ↔ TOON)
- ✅ Round-trip safe (converts back perfectly)
- ✅ Validation (syntax checking)
- ✅ Zero dependencies (lightweight)
- ✅ Browser & Node.js support
When Should You Use TOON?
✅ Perfect For:
AI/LLM Applications — Token efficiency = cost savings
Human-Readable Configs — Easier to edit manually
Large Datasets — Smaller file sizes
Real-Time Processing — Faster parsing
Telemetry Logs — Efficient event storage
Content Formats — Better for narrative/story data
❌ Stick with JSON If:
- Maximum compatibility is critical
- You need extensive tooling/ecosystem support
- Browser native APIs are required
- Your team isn't ready to adopt a new format
The Future of Data Formats
TOON isn't trying to replace JSON everywhere. But for specific use cases — especially AI/LLM applications, human-editable configs, and high-performance systems — it offers compelling advantages.
The format is still evolving, but with working parsers available now, you can start experimenting and seeing the benefits firsthand.
Try It Yourself
Check out the live demo: https://futurevision-labs.github.io/toon-parser/
View the source code: https://github.com/futurevision-labs/toon-parser
Read the original TOON article: https://medium.com/medialesson/json-vs-toon-a-new-era-of-structured-input-19cbb7fc552b
What Do You Think?
Have you tried TOON? What are your thoughts on token-efficient formats? Share your experiences in the comments below!
About the Author: Building tools for the future of web development and AI applications. Part of The Imaginatorium ecosystem.
Resources:
- TOON Parser GitHub: https://github.com/futurevision-labs/toon-parser
- Live Demo: https://futurevision-labs.github.io/toon-parser/
- Original TOON Article: https://medium.com/medialesson/json-vs-toon-a-new-era-of-structured-input-19cbb7fc552b