Performance Benchmarks
Benchmark your own data to see TOON's performance in your specific use case.
Format Comparison
Testing dataset: Complex nested object with arrays, numbers, and strings (typical API response)
JSON
1250 tokens4.2 KB
100%
YAML
1180 tokens3.8 KB
94%
XML
1450 tokens5.1 KB
116%
TOON
625 tokens2.1 KB
50%
Key Performance Metrics
50%
Avg Token Reduction
$3.50
Cost Savings/1M tokens
<1ms
Conversion Time
40%
Storage Savings
Real Dataset Benchmarks
| Dataset | JSON | TOON | Savings | GPT-4 Cost Saved |
|---|---|---|---|---|
| User Profiles (100 records) | 12,500 tokens | 6,250 tokens | 50% | $0.019 |
| E-commerce Products (500 items) | 45,000 tokens | 20,700 tokens | 54% | $0.073 |
| Chat Conversation (50 messages) | 8,200 tokens | 3,690 tokens | 55% | $0.013 |
| API Response (Nested Data) | 18,500 tokens | 9,065 tokens | 51% | $0.028 |
* Costs based on GPT-4 Turbo pricing ($0.01 per 1K input tokens, $0.03 per 1K output tokens)
Methodology
Tokenizer: OpenAI's cl100k_base (GPT-3.5 & GPT-4)
Test Environment: Chrome 120, M1 MacBook Pro
Data Sources: Real-world API responses, database exports, and configuration files
Measurements: Average of 1000 conversions per dataset