smolvm
A portable microVM runtime for running AI agents in real virtual machines with their own kernel, embeddable in an app, checkpointable, and branchable into copy-on-write clones.
smolvm provides branchable microVMs meant for running AI agents locally, embeddable in an app, shippable as a single file, and free to run on your own machine. Machines boot in under a second, use elastic memory so the host only commits what the guest actually uses, persist across restarts, and can run any OCI image, including ones built locally.
Its SDKs let code drive machines directly from Node, Python, or Rust with no separate daemon, running locally or on a hosted cloud. A running machine can be checkpointed to capture its RAM, CPU state, and disks, then rewound or branched into copy-on-write children that keep running from the same point, and a whole machine can be packed into a single file and pushed to any OCI registry.