Muna helps you run AI models on any device. Specifically:
  • Any device: Muna compiles and runs AI models for cloud, desktop, mobile, and web.
  • Zero dependencies: Muna compiles these models into self-contained binaries, meaning you can make predictions with zero dependencies.
  • Hardware acceleration: Muna is designed to use all available compute acceleration hardware (e.g. GPU, NPU) and instruction sets (e.g. CoreML, DirectML, CUDA).

​
Installing Muna

We provide SDKs for common development environments: 
# Run this in Terminal
$ npm install muna
Most of our client SDKs are open-source. Star them on GitHub!

​
Creating an Access Key

Head over to Muna to create an account by logging in. Once you do, generate an access key: create key

​
Making your First Prediction

Let’s run the @fxn/greeting predictor which accepts a name and returns a friendly greeting: 
import { Muna } from "muna"

// 💥 Create a Muna client
const muna = new Muna({ accessKey: "..." });

// 🔥 Make a prediction
const prediction = await muna.predictions.create({
  tag: "@fxn/greeting",
  inputs: { name: "Yusuf" }
});

// 🚀 Print the result
console.log(prediction.results[0]);
You should see a friendly greeting. Congrats on your first prediction 🎉