Agent sandboxes / smol machines alternative

Smol Machines alternative for a growing persistent agent RootFS

smol machines already retains installed packages and system paths across cloud stop/start, including /root, /etc, and /usr. Its durable cloud checkpoints are a separate mechanism that includes RAM. Compare its local machine state, external checkpoint artifacts, and growth contract with your agent’s needs.

Start buildingCompare capabilities ↓

Reviewed 2026-10-18 · Published by the Sandbox0 team

When smol machines fits

smol machines fits teams that want hardware-isolated machines, local/cloud artifact workflows, branchable running machines, and explicit networking/ports. Its documented cloud checkpoints can capture a branchable machine and restore it into a new running machine.

When to consider Sandbox0

Consider Sandbox0 when regional object storage should back the durable RootFS, online growth should expand it while the agent runs, and supported file access should leave compute paused. Check local portability, hardware virtualization, and architecture requirements independently.

Capabilities

smol machines and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimesmol machinesNot established in this reviewSandbox0ttl and hard_ttl default to 0 (disabled). Configure idle pause and deletion policies for the same persistent sandbox identity; an idle running sandbox still accrues compute charges.Read documentation →
Persistent filessmol machinesMachine filesystem persists through exec and stop/start; deletion and infrastructure failure have separate data-loss boundaries.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSsmol machinesThe guide explicitly includes /root, /etc, /usr and /workspace, excluding memory-backed runtime paths.View evidence ↗Sandbox0Persistence extends beyond /workspace to paths such as /usr/local, /etc, and /home. Runtime tmpfs and virtual filesystems are excluded.Read documentation →
Release compute, keep statesmol machinesStop/start keeps disk; branchable-machine checkpoints separately retain execution state.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process statesmol machinesCloud checkpoints include RAM/processes even though machine snapshot endpoints are not implemented.View evidence ↗Sandbox0Experimental memory checkpoints can preserve compatible process state. Use memory: true on both pause and resume on a compatible deployment. A filesystem checkpoint alone does not preserve running processes.Read documentation →

Online RootFS growth

On growth-enabled images, Sandbox0 automatically expands the writable RootFS as free space runs low, within a configured sparse logical ceiling up to 64 TiB. Agent processes keep running. Growth is enabled in the qualified Cloud region; confirm activation for other regions. Existing images and snapshots retain their captured storage geometry. The ceiling is logical capacity, not preallocated host disk.

The reviewed lifecycle guide establishes durable system paths and checkpoint artifacts, but not automatic online expansion of the whole mounted RootFS. Check effective disk limits separately from persistence and process restoration.

Files while compute is paused

Sandbox0's Files API reads persistent files without resuming compute. Filesystem-only pauses also permit edits; retained-memory pauses are read-only and reject mutations with 409. SSH, SFTP, file watching, and runtime-only mounts require running compute.

See paused file operations →

Stop/start retains more than the workspace

The storage guide retains both /workspace and the image filesystem, including /root, /etc, and /usr. Memory-backed /tmp, /run, and /dev/shm are emptied after stop/start. The guide also warns that stopping is not a backup and infrastructure failures can affect machine-local state.

Source 1 ↗

Cloud checkpoints and machine snapshots are different APIs

The current cloud snapshot routes return 501. That does not mean there is no durable execution capture: a branchable running machine supports checkpoints containing guest RAM and processes, stored in cloud storage and downloadable as a .smolcheckpoint artifact. Restore creates a new machine, and the format is not cross-architecture.

Source 1 ↗

Consumed resources still have a running base fee

Billing combines a running-machine base fee with active CPU, RSS memory, used disk, and egress. Stop removes base/CPU/RAM charges while held disk continues. Plan fees are additional to usage; the one-minute minimum also matters for short-lived work.

Source 1 ↗

Pricing

Compare billing models, then price your workload

CPU allocation, memory, billing minimums, disk, plan fees, and supported workload differ. These meters are not equal-performance machine quotes. Confirm region, resource units, and included allowances before estimating a total.

smol machines

Pricing reviewed 2026-10-18

Running machine base
$0.04 per machine-hour
One-minute minimum.
Rate source ↗
Consumed CPU
$0.05 per vCPU-hour
Rate source ↗
RSS memory
$0.0162 per GB-hour
Rate source ↗
Used disk
$0.0001 per GB-hour
Rate source ↗
Egress
$0.05 per GB
Rate source ↗
Standard plan
$0 monthly fee, plus usage
Rate source ↗
Pro / Startup plans
$20 / $150 monthly fee, plus usage
Rate source ↗

Plan fees buy limits/support and are not usage credits. Running idle retains the base fee; stopped disk remains billed. Signup credit is one-time. Preserve GB and consumption meters rather than treating them as allocated-capacity prices.

Check current smol machines pricing ↗

Sandbox0

$0.015/GiB-hour compute (billed by the second at approximately $0.000004167/GiB-second), plus $0.02/GiB-month of retained incremental RootFS and memory checkpoint data. Shared base images and platform recovery retention are excluded; paused sandboxes accrue storage charges only.

Compute / GiB-hour
$0.015
Retained data / GiB-month
$0.02
Paused compute
$0.00

A 4 GiB agent, 100 running hours

$6.00 compute

4 GiB × 100 hours, billed by the second. Keeping 20 GiB of billable incremental data for a full billing month adds $0.40. CPU is derived from memory by the platform; choose a tier that fits your workload.

Understand storage billing →

Before migrating from smol machines

  1. Inventory machine files, memory-backed paths, branchable configuration, checkpoints/artifacts, architecture, published ports, and external backups.
  2. Export needed artifacts and port compatible image inputs, machine lifecycle, exec/files, credentials, and networking.
  3. Validate system state across restarts, growth while running, paused file access, checkpoint expectations, and base plus consumed-resource/storage cost.

Common questions

Is Sandbox0 a drop-in smol machines replacement?

Evaluate API compatibility and product scope. A shared sandbox category does not make SDKs, lifecycle semantics, or limits interchangeable.

Does keeping files also keep a running agent?

Filesystem state and volatile process state are separate. Sandbox0 requires explicit compatible memory checkpoints to preserve running processes; its default filesystem resume starts fresh processes.

Sources and review scope

The Boat team's Battleships directory is the discovery source. References below describe the product and plan discussed on this page. Check current availability, regional terms and billing units before migrating. Where the cited material does not establish a capability or price, the comparison identifies that uncertainty.

Battleships research record ↗smolmachines.com/docs/cloud/lifecycle-storage-networking ↗smolmachines.com/pricing ↗Sandbox0 RootFS capacity →

Related comparisons

Agent 37 vs Sandbox0 →AgentComputer vs Sandbox0 →ainclave vs Sandbox0 →Alibaba Cloud Agent Sandbox / FC / AgentRun vs Sandbox0 →Alibaba Cloud AgentBay vs Sandbox0 →Amika vs Sandbox0 →