Skip to main content

SDIF vs TOON

TOON is one of the formats included in the sdif-benchmarks comparison suite. It appears alongside JSON, YAML, XML, SDIF, and SDIF AI in the benchmark corpus.

What the benchmarks cover​

The sdif-benchmarks suite evaluates each format in the corpus across the same set of metrics:

  • Token efficiency — token count per document when encoded with cl100k_base and other tokenizers.
  • Semantic density — data content relative to structural overhead.
  • Round-trip fidelity — whether conversion to and from the format preserves the canonical SDIF representation.

TOON is tested against these same metrics. The benchmarks apply consistent methodology across all formats: the same document corpus, the same tokenizers, and the same fidelity test (hash equality on the round-trip result).

Where to find results​

This documentation does not describe TOON's syntax or claim measured comparisons between TOON and SDIF. TOON's format properties and the measured benchmark results are available in the sdif-benchmarks repository, where results are organized by format and document.

For methodology details, see Methodology. For the metrics being compared, see Token Efficiency, Semantic Density, and Round-trip Fidelity.