New Token-Oriented Object Notation (TOON) Hopes to Cut LLM Costs by Reducing Token Consumption

The Token-Oriented Object Notation (TOON) format was released in November 2025, claiming a 40% reduction in token usage for certain data structures compared to JSON. Benchmarks show a 55% token reduction in one example when compared to pretty-printed JSON.

Why This Matters

LLM inference costs are heavily tied to token consumption, with JSON’s verbose syntax often leading to inefficiencies. While TOON introduces a 5% overhead for headers and array declarations, it achieves 99.4% accuracy on GPT-5 Nano with 46% fewer tokens. This trade-off could significantly lower costs for applications relying on frequent LLM interactions, though non-uniform data may still favor JSON.

Key Insights

“55% token reduction in benchmarks vs. pretty-printed JSON, 2025”: Demonstrated in the InfoQ example.
“Hybrid YAML/CSV layout for nested and uniform data”: TOON combines YAML’s nesting with CSV’s efficiency for arrays.
“Reference implementation at github.com/toon-format/toon”: MIT-licensed tools include encoder/decoder and benchmarks.

Working Example

{
  "context": {
    "task": "Our favorite hikes together",
    "location": "Boulder",
    "season": "spring_2025"
  },
  "friends": ["ana", "luis", "sam"],
  "hikes": [
    {
      "id": 1,
      "name": "Blue Lake Trail",
      "distanceKm": 7.5,
      "elevationGain": 320,
      "companion": "ana",
      "wasSunny": true
    },
    {
      "id": 2,
      "name": "Ridge Overlook",
      "distanceKm": 9.2,
      "elevationGain": 540,
      "companion": "luis",
      "wasSunny": false
    },
    {
      "id": 3,
      "name": "Wildflower Loop",
      "distanceKm": 5.1,
      "elevationGain": 180,
      "companion": "sam",
      "wasSunny": true
    }
  ]
}

context:
task: Our favorite hikes together
location: Boulder
season: spring_2025
friends[3]: ana,luis,sam
hikes[3]{id,name,distanceKm,elevationGain,companion,wasSunny}:
1,Blue Lake Trail,7.5,320,ana,true
2,Ridge Overlook,9.2,540,luis,false
3,Wildflower Loop,5.1,180,sam,true

Practical Applications

Use Case: High-volume LLM prompt processing in cost-sensitive systems (e.g., chatbots, analytics pipelines).
Pitfall: Applying TOON to non-uniform datasets may increase token usage compared to JSON.

References:

On This Page

New Token-Oriented Object Notation (TOON) Hopes to Cut LLM Costs by Reducing Token Consumption