Overview

smol machines provides a local microVM runtime, a managed cloud service, and an optional application SDK.

All three use the same basic machine model: an isolated Linux guest with explicit resources, storage, networking, and lifecycle controls.

Choose where machines run and how to control them:

Local CLI Run and manage Linux microVMs on your own device. Cloud API Managed, persistent machines on smol machines infrastructure. SDK Create and control machines from a TypeScript or Python application.

For more guidance on choosing the most suitable smol machines product for your use case, refer to the FAQ.

smolvm

smolvm is the open-source engine and CLI. It boots OCI images as Linux microVMs on a local macOS, Linux, or Windows host. Each machine has its own guest kernel and runs through the host’s native virtualization interface.

Use smolvm for local command-line workflows, self-managed infrastructure, portable .smolmachine artifacts, VM forks, and supported local GPU workloads. It does not require a Docker daemon.

smol cloud

smol cloud is the hosted service for managed smolvm-based machines. Clients create and manage machines through the Cloud API, the SDK, or supported CLI workflows.

Cloud clients do not need a local hypervisor. A local machine and a cloud machine can use the same configuration or compatible packaged environment, but they are separate runtime instances.

smol SDK and CLI

The public smol project provides the smolmachines packages for TypeScript and Python, plus the smol orchestration CLI. Its Machine API can embed smolvm for local use or connect to smol cloud.

The SDK is a control surface, not a prerequisite. Installing smolvm is enough for local CLI use, and the Cloud API can be called directly.

How the pieces relate

Layer	Runs the machine	Main interface
smolvm	Your computer or infrastructure	`smolvm` CLI and local engine
smol cloud	smol machines infrastructure	Cloud API, SDK, or CLI
smol SDK / CLI	Local engine or smol cloud	TypeScript, Python, or `smol`