Reduce LLM token usage by 30-60% with Token-Oriented Object Notation
Create with 💛 From JSONTOON.net
TOON reduces token usage by 30-60% compared to JSON, cutting your LLM API costs significantly for large structured data.
Specifically designed for AI prompts and LLM contexts, TOON preserves data structure while minimizing token count.
Uniform arrays are converted to compact table format with headers, dramatically reducing repetitive field names.
Despite its compactness, TOON remains easy to read with YAML-like indentation and clear structure markers.
TOON preserves all your data semantics - it's a faithful representation that can be decoded back to JSON.
Ideal for Retrieval-Augmented Generation pipelines where you need to pass large datasets to your AI model.