Core Engine

Embed it in the application.

SMILE Core Engine is the set of algorithms you call from Java, Scala, or Kotlin. A commercial license covers proprietary distribution, without the GPLv3 source-disclosure obligation.

Start Read the license

The license to keep shipping what was already built continues. Updates, support, and critical and high CVE windows are the annual maintenance term.

What you embed

Classification, regression, clustering, deep learning, and the data layer live in the same process as the product. There is no separate service to operate.

Classic ML & NLP

Hundreds of algorithms — classification, regression, clustering, manifold learning, and classic NLP — behind a clean API.

Deep Learning

Train and deploy neural nets on CPU or GPU. And an idiomatic Java API for running ONNX and ONNX-GenAI models.

Large Language Models

Java-native LLM inference with tiktoken, RoPE, transformer primitives, KV cache management with Radix, quantization, tool calling, chat templates, and OpenAI-compatible REST server integration.

Math, Statistics & Graphs

Dense and sparse matrices and tensors, matrix decompositions, statistical distributions, hypothesis tests, interpolation, wavelet, compressed sensing, and graph data structure.

var iris = Read.arff("iris.arff");
var model = RandomForest.fit(Formula.lhs("class"), iris);
System.out.println(model.metrics());
System.out.println(model.predict(iris.get(0)));

val iris = read.arff("iris.arff")
val model = randomForest("class" ~ ".", iris)
println(model.metrics)
println(model.predict(iris(0)))

val iris = read.arff("iris.arff")
val model = randomForest(Formula.lhs("class"), iris)
println(model.metrics())
println(model.predict(iris[0]))

Speed

Advanced data structures deliver state-of-the-art performance — often faster than R, Python, Spark, or H2O.

Interactive visualization

Interactive Swing plots and Vega-Lite charts in the browser.