ToonUse
v1.0 is now live

Convert JSON to TOON
Save 50% on Tokens

Stop wasting context window on redundant syntax. Optimize your LLM data payloads instantly with our high-performance API.Get 1000 Free Credits on Sign Up.

See the Magic in Action

Swipe through different data scenarios to see how much you can save.

User Profiles

Typical user data structure in applications

Input (JSON)
128 bytes
{
  "users": [
    {
      "id": 1,
      "name": "Alice",
      "role": "admin",
      "active": true
    },
    {
      "id": 2,
      "name": "Bob",
      "role": "user",
      "active": false
    }
  ]
}
Output (TOON)-45% Size
70 bytes
users[2]{id,name,role,active}:
1,Alice,admin,true
2,Bob,user,false

E-commerce Products

Product listings with repeated keys

Input (JSON)
180 bytes
{
  "products": [
    { "id": 101, "name": "Laptop", "price": 999, "stock": 50 },
    { "id": 102, "name": "Mouse", "price": 29, "stock": 200 },
    { "id": 103, "name": "Keyboard", "price": 59, "stock": 150 }
  ]
}
Output (TOON)-52% Size
86 bytes
products[3]{id,name,price,stock}:
101,Laptop,999,50
102,Mouse,29,200
103,Keyboard,59,150

Server Logs

High volume repetitive log entries

Input (JSON)
165 bytes
{
  "logs": [
    { "level": "INFO", "msg": "User login", "ts": 1630001 },
    { "level": "WARN", "msg": "High latency", "ts": 1630002 },
    { "level": "INFO", "msg": "Request ok", "ts": 1630003 }
  ]
}
Output (TOON)-48% Size
85 bytes
logs[3]{level,msg,ts}:
INFO,User login,1630001
WARN,High latency,1630002
INFO,Request ok,1630003

How It Works

Three simple steps to optimizing your data pipeline.

1. Send JSON

Post your JSON data payload to our high-speed API endpoint.

2. We Compress

Our engine strips redundant keys and formats data into TOON.

3. Send to LLM

Send the optimized string to OpenAI, Anthropic, or any LLM.

Perfect For...

RAG Pipelines

Compress retrieved context chunks before feeding them to the generator.

Data Analysis

Fit more rows of CSV/JSON data into a single prompt window.

Chat History

Store and send longer conversation histories for better context.

Agent Logs

Reduce the overhead of verbose agent thought traces and logs.

Engineered for Efficiency

Built for developers who care about token optimization and API latency.

Token Efficient

Reduce your LLM API costs by 30-60% by stripping away redundant JSON syntax.

Developer First

Integrate seamlessly with our REST API. Libraries available for Python and Node.js.

Analytics Ready

Track your token savings and API usage in real-time via our comprehensive dashboard.

Frequently Asked Questions

Does TOON lose any data?

No. TOON is a lossless format. It restructures the data schema to avoid repeating keys, but all values and relationships are preserved.

Which LLMs support TOON?

Any LLM! Since TOON is text-based, you can feed it into GPT-4, Claude 3, Llama 3, or any other model. You just need to instruct the model on how to read it (we provide system prompts).

Is my data secure?

Yes. We do not store your payload data. It is processed in memory for conversion and immediately discarded.

Ready to Optimize Your AI Stack?

Join thousands of developers saving tokens today.