Stop Wasting Tokens.
Start Using Tauq.
The most token-efficient data notation for LLMs. Schema-driven, streaming-ready, and state-of-the-art.
[{"id": 1, "name": "Alice"}, {"id": 2, "name": "Bob"}]!def User id name
1 Alice
2 Bob* 1,000 records: JSON 24,005 → TOON 14,015 → Tauq 11,012
Built for the AI Era
Two formats for different needs: TQN for LLM token efficiency, TBF for binary size and speed.
Token-Optimal (TQN)
49% fewer tokens than JSON for LLM inputs. Space delimiters tokenize better than commas. Verified with tiktoken.
Binary Format (TBF)
83% smaller than JSON. Schema-aware columnar encoding with adaptive compression. Perfect for storage and network.
True Streaming
StreamingParser iterator API processes records one at a time. No count prefix required unlike other formats.
Schema-Driven
Define data shapes with !def. Switch schemas with !use. Type hints for binary encoding optimization.
Iceberg Integration
TBF integrates with Apache Iceberg for data lake workflows. Arrow RecordBatch conversion included.
Programmable
Tauq Query for data transformations. Unix pipe model with polyglot support (Python, Rhai, JavaScript).
Production CLI
Full-featured CLI with build, format, minify, exec, and validate commands. Ready for production use.
Secure & Robust
408 passing tests. Continuous Fuzzing (Google Grade) for parser stability. Safe-by-default TauqQ execution, Rhai resource limits, allocation amplification protection, and import cycle prevention.
Verified Performance
Two formats for different needs: TQN for LLM token efficiency, TBF for binary size optimization.
Token Efficiency
Best for LLM inputs • 1,000 records
Binary Size
Best for storage & network • 1,000 records
Token counts verified with tiktoken cl100k_base (GPT-4o/Claude tokenizer)
Cost based on GPT-4o/Claude Sonnet pricing ($3/1M input tokens, March 2026)
Get Started in Minutes
Install Tauq and start saving tokens immediately. Available for all major languages.
1. Install
# Install via cargo
cargo install tauq
# Or via package managers
pip install tauq # Python
npm install tauq # JavaScript
go get github.com/epistates/tauq # Go2. Use
# Create a config file
cat > config.tqn <<EOF
app_name "MyService"
version "1.0.0"
port 8080
debug true
features [api websockets metrics]
EOF
# Parse to JSON
tauq build config.tqn --prettyPerfect for Every Use Case
AI/LLM Applications
Reduce token costs by up to 54% vs JSON, 14% vs TOON on average. Perfect for prompt engineering, RAG systems, and LLM training data.
Learn more →Data Processing
Stream large datasets efficiently. ETL pipelines, data transformations, and batch processing.
Learn more →Configuration Management
Human-readable configs with schema validation. Replace YAML, JSON, or TOML with Tauq.
Learn more →Multi-Language Support
Use Tauq in your favorite programming language.
Python
Native bindings
JavaScript
Native bindings
Rust
Native bindings
Go
Native bindings
Example: Tabular Data
!def User id name email role
1 Alice "alice@example.com" admin
2 Bob "bob@example.com" user
3 Carol "carol@example.com" userReady to Save Tokens?
Join developers who are already using Tauq to reduce AI costs and improve performance.