GPU clouds / DigitalOcean GPU Droplets alternative

A DigitalOcean GPU Droplets alternative for persistent agent orchestration

DigitalOcean GPU Droplets are Linux VMs with GPU hardware, a persistent boot disk, and non-persistent scratch on selected configurations. Powering a Droplet off preserves its allocation and continues the bill. For a returning agent, compare system-disk retention and the idle-cost policy separately from accelerator selection.

Start buildingCompare capabilities ↓

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

When DigitalOcean GPU Droplets fits

GPU Droplets fit workloads needing direct NVIDIA or AMD accelerator access in a self-managed VM, with driver-ready images and familiar SSH and infrastructure controls.

When to consider Sandbox0

Consider Sandbox0 for Linux agents that orchestrate external GPU jobs and should retain their complete writable RootFS between tasks without paying paused compute charges. GPU execution still requires suitable accelerator infrastructure.

Capabilities

DigitalOcean GPU Droplets and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeDigitalOcean GPU DropletsNot 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 filesDigitalOcean GPU DropletsThe boot disk is persistent; scratch disks are non-persistent and are separate storage.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSDigitalOcean GPU DropletsOperating-system and framework files on the boot disk persist through ordinary power off/on; scratch mounts are excluded.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 stateDigitalOcean GPU DropletsPower off/on retains the VM allocation and disk, but continues resource billing and does not establish memory continuity.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateDigitalOcean GPU DropletsDocumented software snapshots are reusable machine images; RAM or GPU-state restoration is not established.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.

DigitalOcean currently disallows GPU Droplet resizing. Its persistent boot images and separate storage are useful, but differ from automatic online expansion of Sandbox0’s enabled writable RootFS while agent processes continue.

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 powered-off GPU Droplet still incurs its reserved resource charges. The reviewed snapshot and storage features do not establish arbitrary boot-file editing with the original guest stopped. Sandbox0 provides supported paused RootFS operations under its filesystem-only and retained-memory rules.

See paused file operations →

Boot persistence is broader than a project directory

The local persistent boot disk stores the OS and installed ML frameworks. Some GPU configurations also have a local non-persistent scratch disk; useful outputs must be copied to durable storage. A configured software environment can be snapshotted and used as the image for another GPU Droplet.

Source 1 ↗Source 2 ↗

Power off keeps the full GPU allocation bill

DigitalOcean explicitly bills powered-off GPU Droplets because CPU, RAM, disk and IP remain reserved. Billing ends only when the instance is destroyed. GPU usage is per second with a five-minute minimum and continues beyond the standard 672-hour monthly limit; the ordinary bundled CPU plan’s monthly cap should not be applied here.

Source 1 ↗

Changing capacity means planning another GPU machine

The current GPU limits prohibit resizing GPU Droplets. Preserving a boot image and moving needed files to another instance is distinct from growing a running agent’s RootFS automatically. Spot GPU capacity is a separate preview offer with variable price and reclamation risk.

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.

DigitalOcean GPU Droplets

Pricing reviewed 2026-10-19

H100 on-demand reference
$4.41 per GPU-hour
Verify configured GPU count and complete Droplet shape.
Rate source ↗
L40S on-demand reference
$1.57 per GPU-hour
Rate source ↗
Minimum usage bill
5 minutes, then per-second usage
Rate source ↗
Powered-off GPU Droplet
Continues billing until destroyed
GPU billing has no standard 672-hour monthly cap.
Rate source ↗

Prices refer to on-demand GPU Droplets, not spot GPU capacity or a GPU-free CPU Droplet. Included boot disk, RAM, CPU and transfer vary by shape. Confirm additional storage, snapshot, transfer and tax charges. Stopping a VM and deleting it have different data and billing consequences.

Check current DigitalOcean GPU Droplets 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 DigitalOcean GPU Droplets

  1. Copy durable results off non-persistent scratch and record driver, GPU, boot image, networking and allocation details.
  2. Preserve required images and backups before destroying a billed source VM; migrate compatible Linux agent stages and keep GPU processing separate.
  3. Test retained packages, growing data, supported paused-file operations and task recovery, then compare the complete machine and retained-storage bill.

Common questions

Is Sandbox0 a drop-in DigitalOcean GPU Droplets 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.digitalocean.com/products/droplets/details/features/ ↗docs.digitalocean.com/products/droplets/getting-started/recommended-gpu-setup/ ↗www.digitalocean.com/pricing/gpu-droplets ↗docs.digitalocean.com/products/droplets/details/limits/ ↗Sandbox0 RootFS capacity →

Related comparisons

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