# Introduction
JSON is nice for APIs, storage, and utility logic. However inside giant language mannequin (LLM) pipelines, it usually carries numerous token overhead that doesn’t add a lot worth to the mannequin: braces, quotes, commas, and repeated subject names on each row. TOON, brief for Token-Oriented Object Notation, is a more moderen format designed particularly to maintain the identical JSON information mannequin whereas utilizing fewer tokens and giving fashions clearer structural cues. The official TOON docs describe it as a compact, lossless illustration of JSON for LLM enter, particularly sturdy on uniform arrays of objects.
On this article, you’ll be taught what TOON is, when it is smart to make use of it, and easy methods to begin utilizing it step-by-step in your individual LLM workflow. We will even maintain the tradeoffs trustworthy, as a result of TOON is beneficial in some instances, not all of them.
# Why JSON Wastes Tokens in LLM Pipelines
JSON turns into costly in prompts as a result of it repeats construction again and again. LLMs don’t care that JSON is a typical. They solely see tokens.
In case you ship 100 assist tickets, product rows, or person data to a mannequin, the identical subject names seem in each object. TOON reduces that repetition by declaring fields as soon as after which streaming row values in a compact tabular type. Right here is a straightforward instance.
JSON:
{
"customers": [
{ "id": 1, "name": "Alice", "role": "admin" },
{ "id": 2, "name": "Bob", "role": "user" },
{ "id": 3, "name": "Charlie", "role": "user" }
]
}
TOON:
customers[3]{id,title,position}:
1,Alice,admin
2,Bob,person
3,Charlie,person
Similar information, much less muddle.
The construction remains to be clear, however the repeated keys are gone. That’s the place TOON will get most of its worth.
# What TOON Truly Is and When It Is Price Utilizing
TOON is a serialization format for the JSON information mannequin. Which means it could possibly characterize objects, arrays, strings, numbers, booleans, and null values — however in a means that’s extra compact for mannequin enter. The TOON challenge presents it as lossless relative to JSON, which implies you’ll be able to convert JSON to TOON and again with out dropping data. The essential factor to grasp is that this:
You do not want to exchange JSON in your app.
A greater method is to maintain JSON in your backend, APIs, and storage, then convert it to TOON solely if you end up about to ship structured information into an LLM.
TOON is most helpful when your immediate accommodates repeated structured data with the identical fields. Good examples embody retrieved assist tickets, catalog rows, analytics data, software outputs, CRM entries, or reminiscence snapshots for agent techniques. Nonetheless, in case your construction is deeply nested, extremely irregular, purely flat, or very small, the advantages can shrink or disappear.
# Getting Began with TOON
// Step 1: Putting in the TOON Command-Line Interface
The best strategy to attempt TOON is with the official command-line interface (CLI) from the TOON challenge. The TOON web site hyperlinks on to its CLI, and the principle repository presents the format as a part of a broader SDK and tooling ecosystem.
Set up the bundle:
npm set up -g @toon-format/cli
// Step 2: Changing a JSON File into TOON
Let’s create a folder first:
mkdir toon-test
cd toon-test
Now, run the next command to create the JSON file:
Paste this:
[
{ "id": 1, "name": "Alice", "role": "admin" },
{ "id": 2, "name": "Bob", "role": "user" },
{ "id": 3, "name": "Charlie", "role": "user" }
]
Now convert it:
npx @toon-format/cli customers.json -o customers.toon
It is best to get a compact consequence much like this:
[3]{id,title,position}:
1,Alice,admin
2,Bob,person
3,Charlie,person
That is the core TOON sample: declare the form as soon as, then record the values row by row. That aligns with the official design purpose of tabular arrays for uniform objects.
// Step 3: Utilizing TOON as Mannequin Enter
The perfect place to make use of TOON is on the enter facet of your pipeline. As an alternative of pasting a big JSON blob right into a immediate, move the TOON model and maintain the instruction easy.
For instance:
The next information is in TOON format.
customers[3]{id,title,position}:
1,Alice,admin
2,Bob,person
3,Charlie,person
Summarize the person roles and level out something uncommon.
This works nicely as a result of TOON is designed to assist the mannequin learn repeated construction with much less overhead. That can be how the official challenge frames its benchmarks: as a take a look at of comprehension throughout completely different structured enter codecs.
// Step 4: Maintaining JSON for Outputs
This is likely one of the most essential sensible choices. TOON could be very helpful for enter, however JSON remains to be normally the higher selection for output when one other system must parse the mannequin response. That’s as a result of JSON has a lot stronger tooling assist, and fashionable APIs can implement structured JSON output with schemas.
In observe, the most secure sample is:
- JSON in your app.
- TOON for giant structured immediate context.
- JSON once more for machine-parseable mannequin responses.
This offers you effectivity on the enter facet and reliability on the output facet.
// Step 5: Benchmarking in Your Personal Pipeline
Don’t change codecs based mostly on hype alone.
Run a small benchmark in your individual workflow:
- Depend enter tokens for JSON.
- Depend enter tokens for TOON.
- Evaluate latency.
- Evaluate reply high quality.
- Evaluate whole value.
The official TOON challenge positions token financial savings as one of many fundamental advantages, and third-party protection repeats these claims, however neighborhood dialogue additionally reveals that outcomes rely closely on the form of the information. That’s the reason one of the best query just isn’t “Is TOON higher than JSON?”
The higher query is: “Is TOON higher for this particular LLM step?”
# Remaining Ideas
TOON just isn’t one thing it’s essential use in every single place.
It’s a focused optimization for one particular downside: losing tokens on repeated JSON construction inside LLM prompts. In case your pipeline passes plenty of repeated structured data right into a mannequin, TOON is value testing. In case your payloads are small, irregular, or closely nested, JSON should still be the higher selection.
The neatest strategy to undertake it’s easy: maintain JSON the place JSON already works nicely, use TOON the place you might be packing giant structured inputs into prompts, and benchmark the outcomes by yourself duties earlier than committing to it.
Kanwal Mehreen is a machine studying engineer and a technical author with a profound ardour for information science and the intersection of AI with drugs. She co-authored the e-book “Maximizing Productiveness with ChatGPT”. As a Google Era Scholar 2022 for APAC, she champions variety and educational excellence. She’s additionally acknowledged as a Teradata Range in Tech Scholar, Mitacs Globalink Analysis Scholar, and Harvard WeCode Scholar. Kanwal is an ardent advocate for change, having based FEMCodes to empower ladies in STEM fields.
