Quickstart

import Feedback from "/snippets/page-feedback.mdx";

Choose the interface that matches what you're building. Each path takes you from install to a working model in a few minutes.

<CardGroup cols={2}> <Card title="Command Line" href="/en/run/cli/quickstart" icon="terminal"> Run models from a terminal on Windows ARM64, or via Docker on Linux ARM64. Best for trying things out fast. </Card> <Card title="Local Server" href="/en/run/cli/local-server" icon="server"> OpenAI-compatible API on Windows ARM64 and Linux ARM64. </Card> <Card title="Python SDK" href="/en/run/python/quickstart" icon="python"> Huggingface style API for scripting and notebooks, available on Windows ARM64 and Linux ARM64. </Card> <Card title="Linux (Docker)" href="/en/run/linux/install" icon="docker"> Docker image for Linux ARM64 with NPU access. Best for Dragonwing IoT and similar platforms. </Card> <Card title="Android SDK" href="/en/run/android/quickstart" icon="android"> Kotlin SDK from Maven Central, plus a prebuilt demo APK for Snapdragon 8 Elite. </Card> </CardGroup>

Before you start

• Platform & runtime — pick where you'll run (Windows ARM64, Android, Linux ARM64) and which runtime fits your model. See Platforms & runtimes.
• Models — see Models for tested examples on each runtime.
• Bring your own model — already have a GGUF or AI Hub bundle on disk? See Run a local model.

<Feedback/>