← Blog

TOON

TOON articles: what Token-Oriented Object Notation is, its tabular syntax, how it stacks up against JSON, CSV and YAML, and cutting LLM token costs with it.

Apr 17, 2026·8 min read

What is TOON Format? A Developer's Introduction

What is TOON? Token-Oriented Object Notation is a compact serialization format built for LLM apps. Why it was created, and when it beats JSON in a prompt.

Read article →
Apr 17, 2026·8 min read

TOON vs JSON — Which Should You Use?

TOON vs JSON: one is universal, the other is built to save tokens. A practical comparison putting the same records in both formats, and when each one fits.

Read article →
Apr 17, 2026·9 min read

TOON vs JSON vs CSV — Choosing the Right Format

TOON vs JSON vs CSV: JSON for nested data, CSV for flat tables, and TOON for both when every token in an LLM prompt counts. How to pick the right one.

Read article →
Apr 17, 2026·9 min read

TOON Syntax Guide — From Scalars to Tabular Data

TOON syntax reference: scalars, nested objects, plain arrays, and the tabular notation that keeps row-shaped datasets compact, with encode and decode samples.

Read article →
Apr 17, 2026·10 min read

Using TOON with LLMs — Cut Token Costs Without Losing Structure

TOON for LLM prompts: every token you send costs money, and TOON's tabular rows shrink structured data. How to use it with the OpenAI and Anthropic APIs.

Read article →
Apr 17, 2026·9 min read

TOON Format Examples — Real-World Data Patterns

TOON examples with realistic data, not toy objects: product catalogs, audit logs, API metrics, financial transactions and mixed documents, encoded in TOON.

Read article →
Apr 17, 2026·9 min read

TOON vs YAML — Two Compact Formats, Very Different Jobs

TOON vs YAML: both drop most of JSON's punctuation, but YAML suits people editing files and TOON suits models reading data. Why YAML can cost more tokens.

Read article →
Apr 17, 2026·10 min read

TOON in Node.js — From File I/O to LLM Pipelines

TOON in Node.js: reading and writing .toon files, parsing request bodies, validating at system boundaries, and feeding database rows into an LLM pipeline.

Read article →