GPU clouds / Hyperstack alternative

A Hyperstack alternative for the agent around your GPU workload

Hyperstack already supports retaining a VM’s root disk and configuration through hibernation, with compute resources deallocated. Ordinary Stop is different: SHUTOFF keeps the allocation and its full bill. Examine root, ephemeral and shared storage separately before planning a returning agent’s lifecycle.

Start buildingCompare capabilities ↓

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

When Hyperstack fits

Hyperstack fits GPU applications needing configurable VMs, GPU flavors, shared storage, snapshots and related networking or AI services. Hibernation can remove GPU and CPU charges while retaining a saved root disk.

When to consider Sandbox0

Consider Sandbox0 for compatible Linux agents whose complete durable RootFS should grow automatically online, with supported file access while sandbox compute stays paused. Keep GPU processing on the appropriate accelerator infrastructure.

Capabilities

Hyperstack and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeHyperstackNot 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 filesHyperstackHibernation retains root disk data; shared storage volumes have an independent retained-data lifecycle.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSHyperstackInstalled software and configuration on the root disk are saved; ephemeral storage is excluded and must be backed up separately.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 stateHyperstackHibernation releases compute and preserves the root disk, but restore depends on availability of the same flavor.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateHyperstackPublished hibernation scope names configuration and root disk data; durable application/GPU RAM checkpoint scope 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.

Hyperstack flavor changes use a snapshot and replacement VM rather than changing the running flavor in place. Root, ephemeral and shared storage are distinct; these mechanisms differ from automatic online growth of Sandbox0’s enabled whole writable RootFS.

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.

Hibernation retains root data without GPU compute billing, but the reviewed guides do not establish arbitrary file reads and writes against that saved whole root through a paused-file API. Sandbox0 supports those RootFS operations under its filesystem-only and retained-memory restrictions.

See paused file operations →

Hibernation preserves the root disk, with restore constraints

Hibernation saves configuration and root disk data and releases CPU, GPU, RAM and ephemeral storage. Copy needed ephemeral data to durable storage first. Restoring requires the original flavor to be back in stock. Public IP is released by default unless explicitly retained, and saved root data, retained IPs and shared volumes continue billing.

Source 1 ↗Source 2 ↗

A stopped VM still bills at its full resource rate

SHUTOFF keeps CPU, RAM, GPU and local storage reserved. Hibernation and deletion have different data and billing outcomes. Transitional and failed-operation ERROR states are listed as not billed; retained storage can still explain a decreasing account balance without an active VM.

Source 1 ↗

Snapshots and flavor changes are explicit workflows

A snapshot records configuration and bootable disk data. An ACTIVE VM temporarily shuts off for consistency and restarts afterward. Flavor changes use a snapshot and a new VM rather than an in-place resource change. That workflow caps snapshots at 100 GB, excludes Spot, Kubernetes-node and bootable-volume VMs, and stays in one region. Detach data volumes and save ephemeral data first.

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.

Hyperstack

Pricing reviewed 2026-10-19

H100 SXM on-demand reference
$3.20 per GPU-hour
Rate source ↗
H100 PCIe on-demand reference
$2.50 per GPU-hour
Different hardware from SXM; do not assume equal throughput.
Rate source ↗
H200 SXM on-demand reference
$3.99 per GPU-hour
Rate source ↗
Shared Storage Volumes
$0.000096774 per GB-hour
Retained volumes bill even when detached or RESERVED.
Rate source ↗
Retained public IP
$0.00672043 per hour
Applies when an IP remains allocated, including explicit retention during hibernation.
Rate source ↗

On-demand is prepaid and billed per minute; reserved, spot and AI Studio token prices describe separate offers. Hibernated root storage and snapshots are additional costs. Confirm the chosen flavor, effective storage and IP terms, and balance-related termination or deletion policies.

Check current Hyperstack 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 Hyperstack

  1. Identify root, ephemeral and shared-volume data, flavor availability, retained IPs and SHUTOFF versus HIBERNATED state.
  2. Back up ephemeral files and verify any snapshot before deleting or replacing a source VM; port compatible Linux agent stages and retain GPU jobs separately.
  3. Validate installed tools, growth, supported paused files and restore timing, and compare compute plus saved-root, volume, snapshot and IP costs.

Common questions

Is Sandbox0 a drop-in Hyperstack 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.hyperstack.cloud/docs/virtual-machines/hibernation/ ↗docs.hyperstack.cloud/docs/faq/ ↗docs.hyperstack.cloud/docs/virtual-machines/snapshots/ ↗docs.hyperstack.cloud/docs/virtual-machines/change-vm-flavor/ ↗www.hyperstack.cloud/gpu-pricing ↗Sandbox0 RootFS capacity →

Related comparisons

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