Back to Blog
November 20, 2025
104 views
By Admin

What is TOON?

What is TOON? How Token-Oriented Object Notation works under the hood ?

Web Development
Toon
Json
Data-format
Performance

What is Toon?

TOON is a new data serialization format designed with a code objective:

Reduce the number of tokens when exchanging structured data with language models.

While JSON uses verbose syntax with braces, quotes, and commas, TOON relies on a token-efficient tabular style, which is much closer to how LLMs naturally understand structured data.


The name “Toon” sometimes refers to “Typed Object Optimized Notation,” a modern approach where every field has a known type and is stored in an optimized binary layout, similar to formats like MessagePack, Protocol Buffers, Avro, or CBOR — but typically simpler and even lighter.


In simple terms:


Toon = JSON with strict types + binary encoding + zero waste.


Let’s make a quick comparison between JSON and TOON:

Here is some JSON with a users array that contains information about two users (two objects):

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

If you wanted to represent the same data in TOON, it would look like this:

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

JSON vs Toon: Why JSON Becomes a Bottleneck

## 1. JSON is Text-Based (Slow to Parse)


JSON parsing requires:

reading characters

scanning strings

converting "123" into an integer

converting "true" into a boolean

sanitizing special characters

These steps cost CPU cycles and block event loops in Node.js, Python, or Go if you parse large payloads.

Toon, being binary, doesn't need string parsing at all.


## 2. JSON is Verbose (Large Size)


Size:

  • JSON: ~52 bytes

  • Toon (binary): ~16 bytes


With Toon:

  • field names are not repeated

  • numbers are stored as binary integers (4 bytes)

  • booleans are stored as 1 byte

  • strings are length-prefixed

On large datasets, Toon commonly reduces size by 40–70%.



## 3. JSON Has No Native Types

Everything in JSON is text:

  • "123" → must be parsed into a number

  • "true" → parsed into boolean

  • "2024-01-01" → must be parsed into date

  • "3.14" → parsed to float

Toon stores typed values directly:

  • int32

  • uint64

  • float

  • bool

  • string

  • map

  • array

  • binary blob

  • datetime

No conversion step → faster processing.


## 4. JSON Requires More Memory

JSON parsing creates intermediate objects, especially in JavaScript, Python, or Java.

Toon decodes directly into typed memory structures (like Protobuf or MessagePack).

This reduces:

  • GC pressure

  • CPU usage

  • memory overhead

  • latency spikes



Performance Comparison

OperationJSONToon
Size100%30–60% smaller
Serialization speedSlow (string building)Very fast (binary write)
Deserialization speedSlow (character parsing)Extremely fast (binary read)
CPU usageHighLow
Memory usageHighLow
Streaming supportLimitedExcellent
Suitable for real-time❌ No✅ Yes


Where Toon Is Used (Real Use-Cases)

1. Real-Time Messaging Systems

  • Toon-like binary formats are used in:
  • WhatsApp

  • Telegram

  • Zoom / WebRTC metadata

  • Online games (Unity, Unreal)

  • IoT devices

Because binary is much faster and smaller than JSON.




2. Microservices with High Throughput

Companies like:

  • Netflix

  • Uber

  • Google

  • Cloudflare

use binary formats between internal services to reduce CPU and bandwidth usage.




3. Edge Computing & CDNs

On edge workers (like Cloudflare Workers), binary formats reduce:

  • cold start time

  • memory limits

  • CPU billing



4. AI, ML & Vector Stores

JSON is too slow for numeric-heavy workloads (embeddings, tensors, statistics).


Why Toon Is Efficient: The Core Principles


1. Binary, Not Text

No unnecessary quotes or parsing.


2. Typed Fields

No need to convert "123" → number.


3. Compact Field Schema

Field names are stored once or not needed at all.


4. Predictable Layout

Much easier and faster for machines to process.


5. Works Well Over Network

Small payload → lower latency.



Conclusion: JSON Is Good — but Toon Is Better

JSON is not going away.
It's:

  • easy

  • human-readable

  • great for APIs

  • simple for frontend apps

But JSON is also:

  • slow to parse

  • heavy to transfer

  • not type-safe

  • inefficient at scale

Toon is the next step: a lightweight, typed, binary format designed for speed and efficiency.

If you're building:

  • real-time apps

  • microservices

  • high-performance backends

  • low-latency communication

  • data-heavy pipelines

You should absolutely consider switching from JSON to Toon)))


In our next blog we will try building a simple chatapp using toon


Share this article

Related Posts

Oct 21, 2025158 views

Microservices vs Monolithic Architecture

Microservices vs Monolithic Architecture: A Comprehensive Guide to Choosing the Right Approach

Oct 21, 2025144 views

What is REST API vs GraphQL?

REST API vs GraphQL: Choosing the Right API Architecture for Your Project

Modern web development in 2025
Oct 17, 2025150 views

Modern web development in 2025

Modern web development in 2025