Toka SDK

AI Cost Optimizer SDK for Developers
Track token usage • Estimate costs in real-time • Reduce AI API spend • Optimize model usage

        

---

🚀 What is Toka?

Toka SDK is a lightweight, developer-first AI Cost Optimization SDK built to help you:

• Track token usage automatically
• Estimate API costs in real-time
• Prevent overspending with cost limits
• Reduce redundant API calls using caching
• Automatically fallback to cheaper AI models

It works seamlessly with modern AI models like OpenAI and other LLM providers.

If you're building AI-powered apps, Toka helps you control cost without sacrificing performance.

---

📦 Installation

npm install toka-sdk

---

⚡ Quick Start

1️⃣ Import the SDK

import { TokaClient } from 'toka-sdk';

---

2️⃣ Initialize the Client

const client = new TokaClient({
  apiKey: process.env.TOKA_API_KEY,
  models: ['gpt-4', 'gpt-4o-mini', 'gpt-3.5'],
  maxCostPerRequest: 0.05,   // Maximum allowed cost per request (USD)
  cache: true                // Enable built-in caching
});

---

3️⃣ Make an API Call with Cost Tracking

const response = await client.chat({
  messages: [
    { role: 'user', content: 'Hello world' }
  ]
});

console.log(`Text: ${response.text}`);
console.log(`Cost: $${response.cost}`);
console.log(`Tokens used: ${response.tokens}`);
console.log(`Model used: ${response.modelUsed}`);
console.log(`Cache hit: ${response.cacheHit}`);

---

✨ Core Features

🔍 Token Usage Tracking

Automatically tracks token usage for every request.

💰 Real-Time Cost Estimation

Know exactly how much each API call costs before and after execution.

🧠 Intelligent Cost Optimization

If a request exceeds your defined budget, Toka automatically falls back to a cheaper model.

⚡ Built-In Caching

Reduce redundant API calls with:

• In-memory caching
• Optional Redis integration

🔄 Multi-Model Support

Define multiple models and let Toka dynamically choose the optimal one.

📊 Logging & Analytics

Gain visibility into:

• Tokens used
• Cost per request
• Cache hits
• Fallback events
• Model selection

---

🧩 Optional Enhancements

Toka is modular and extensible.

🗄 Redis Caching

Use Redis for scalable, production-grade caching.

📈 Dashboard Middleware

Mountable Express middleware to monitor AI usage and costs.

🖥 CLI Tool

Interact with AI APIs directly from your terminal.

🎣 Event Hooks

Listen to lifecycle events:

• onRequest
• onFallback
• onCacheHit
• onComplete

---

🧪 Testing

Run unit tests:

npm test

---

🛠 Development

Build the project:

npm run build

Lint:

npm run lint

Format:

npm run format

---

🌍 Use Cases

Toka SDK is ideal for:

• AI SaaS platforms
• Chatbot applications
• AI-powered web apps
• Prompt engineering workflows
• High-volume AI API environments
• Startups monitoring burn rate

---

🤝 Contributing

We welcome contributions from the community!

Please read CONTRIBUTING.md before submitting a pull request.

Ways to contribute:

• Bug fixes
• Feature improvements
• Documentation
• Test coverage
• New integrations

---

📄 License

MIT License © 2026
Abreham Wondimu Shiferaw

---

⭐ Support the Project

If Toka SDK helps your project:

• Star the repository
• Share it with other developers
• Open issues or feature suggestions
• Contribute improvements

Open source grows through community support ❤️