GPU clouds / Crusoe alternative

A Crusoe alternative for the agent around your GPU work

Crusoe Cloud combines GPU and CPU VMs with persistent OS disks, separate durable data disks, and local ephemeral GPU storage. Its OS disk preserves installed software, while local scratch has a different lifecycle. Evaluate which state survives before moving an agent that returns between compute-heavy tasks.

Start buildingCompare capabilities ↓

Reviewed 2026-10-19 · Published by the Sandbox0 team

When Crusoe fits

Crusoe fits GPU training, inference, simulations and custom VM environments needing accelerator access and associated networking. Its persistent boot disk and custom images already support retaining system software.

When to consider Sandbox0

Consider Sandbox0 for compatible Linux agent orchestration, tools and user files that should persist while compute pauses and the RootFS grows online. Retain Crusoe or another accelerator service for GPU stages.

Capabilities

Crusoe and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeCrusoeNot 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 filesCrusoePersistent OS and data disks retain files independently of running compute; ephemeral GPU scratch is excluded.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSCrusoeSystem changes on the persistent OS disk survive normal stop/start, while local ephemeral mounts have separate loss conditions.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 stateCrusoeStopped VMs can be started with retained persistent-disk data; this is disk continuity rather than documented RAM restoration.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateCrusoeCustom boot-disk images do not establish memory or GPU execution-state checkpoints.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.

Crusoe currently does not resize OS disks; a persistent data disk must be detached before enlargement. Its shared-disk capacity is a separate mounted service. Sandbox0 instead supports automatic online growth of an enabled writable RootFS without restarting the agent.

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.

A retained Crusoe disk can be detached and used with other infrastructure, but that is a different workflow from supported Files API operations while the original Sandbox0 guest stays paused. Do not equate stopped disk retention with arbitrary offline guest-file mutation.

See paused file operations →

The persistent OS disk and ephemeral GPU disks differ

The storage guide describes a 128 GB persistent OS disk. Additional persistent disks have an independent lifecycle, while GPU local ephemeral disks can be erased on stop/start or a physical-host reboot or failure. The VM management overview’s broad retention statement must be read with those storage-specific exclusions.

Source 1 ↗Source 2 ↗

Stopping removes on-demand compute charges, with exceptions

On-demand VMs bill by running seconds, including running lifecycle scripts, and stopped VMs do not accrue on-demand compute fees. OS and other durable disks continue billing. Reserved instance commitments bill throughout the commitment regardless of VM state. Stop/start should not be treated as retaining GPU or process memory.

Source 1 ↗Source 2 ↗

Images and disk growth require explicit operations

Custom images capture a stopped VM’s boot disk and are scoped to its project and region. Additional persistent disks can grow only while detached; OS disks currently cannot resize. Shared disks provide another capacity model, rather than automatically expanding the VM’s boot filesystem.

Source 1 ↗Source 2 ↗

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.

Crusoe

Pricing reviewed 2026-10-19

H100 80 GB HGX on-demand reference
$4.50 per GPU-hour
Multiply by the configured GPU count; instance availability and shape matter.
Rate source ↗
B200 180 GB HGX on-demand reference
$8.50 per GPU-hour
Rate source ↗
Provisioned persistent disks, including OS
$0.08 per GiB-month
Billed even without an attached running VM.
Rate source ↗
Provisioned shared disks
$0.07 per GiB-month
Rate source ↗

These VM GPU rates differ from Crusoe’s separately priced managed inference endpoints. Confirm the complete VM shape and reservation terms. Ephemeral disk is included in VM pricing; separate persistent storage remains chargeable during on-demand stops. No inference-throughput or universal savings comparison is implied.

Check current Crusoe 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 Crusoe

  1. Inventory persistent OS and data disks separately from local ephemeral GPU scratch and copy indispensable scratch data before stopping.
  2. Export boot configuration, tools and files; port Linux agent stages while preserving GPU services, project boundaries, network rules and credentials.
  3. Validate retained system paths, growing outputs, paused files and restart requirements; compare complete running shapes and retained storage commitments.

Common questions

Is Sandbox0 a drop-in Crusoe replacement?

Evaluate API compatibility and product scope. These products serve different core workflows, so a change may require a different architecture.

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 ↗docs.crusoecloud.com/storage/disks/overview.md ↗docs.crusoecloud.com/compute/virtual-machines/managing-vms.md ↗docs.crusoecloud.com/compute/virtual-machines/overview.md ↗docs.crusoecloud.com/compute/images/overview.md ↗www.crusoe.ai/cloud/pricing ↗Sandbox0 RootFS capacity →

Related comparisons

Azure GPU VMs (NC_A100_v4 reference) vs Sandbox0 →Cerebrium Serverless Compute vs Sandbox0 →DigitalOcean GPU Droplets vs Sandbox0 →Fluidstack vs Sandbox0 →Google Compute Engine GPU VMs vs Sandbox0 →Hetzner GPU servers vs Sandbox0 →