GPU clouds / Hyperbolic alternative

A Hyperbolic GPU alternative for returning agent orchestration

Hyperbolic rents GPU VMs and bare-metal capacity, with separate reserved and private-cloud offers. Onboard storage survives a supported platform reboot but is erased on reprovisioning or termination. Optional persistent volumes can outlive instances in supported regions. Compare those paths with the durable state a returning Linux agent needs.

Start buildingCompare capabilities ↓

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

When Hyperbolic fits

Hyperbolic fits training, fine-tuning, inference and multi-node GPU workloads needing direct accelerator access and selected interconnects. Available hardware, persistent storage and reboot support vary by region and machine type.

When to consider Sandbox0

Consider Sandbox0 for the compatible Linux agent around GPU jobs, with durable system packages and files, automatic online RootFS growth and supported paused-file access. Keep accelerator workloads on a service that supplies the required GPUs.

Capabilities

Hyperbolic and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeHyperbolicNot 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 filesHyperbolicOptional regional network volumes survive instance termination; onboard storage does not.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSHyperbolicReturning persistent storage is a separate volume; the onboard OS/user storage is erased on termination or reprovisioning.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 stateHyperbolicSupported platform reboot is documented; a general durable stopped-instance pause/resume contract is not established.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateHyperbolicML model checkpoints and onboard disk retention during reboot are not durable process/RAM 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.

Hyperbolic supports separate volumes and custom storage configurations. Those operations do not establish automatic online expansion of the whole mounted OS RootFS; onboard storage and independently retained data must be managed separately.

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 →

Onboard storage and persistent volumes have separate lifetimes

Included onboard NVMe is erased at instance termination or reprovisioning. Persistent network volumes are created separately, can attach to running instances and can remain after every instance terminates. Storage availability and hourly rates are region-specific; a separate mounted volume does not preserve every changed OS path.

Source 1 ↗

Use the supported platform lifecycle

Where available, platform Reboot preserves onboard storage and platform network state. The guide cautions that in-OS reboot can leave some machine types unrecoverable, and hardware-failure recovery can require a fresh instance. Current lifecycle guidance covers running, reboot and termination; it does not establish general RAM-preserving pause or compute-free whole-system retention.

Source 1 ↗

Advertised starting prices are not fixed universal tariffs

The current homepage lists starting GPU-hour rates and says they are refreshed weekly from supplier availability. Review the exact machine, GPU count, region, interconnect and storage quote. Running instances are billable; pending, starting, failed and terminating states are listed as not billed.

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.

Hyperbolic

Pricing reviewed 2026-10-19

H100 SXM starting reference
From $3.19 per GPU-hour
Rate source ↗
H200 starting reference
From $3.99 per GPU-hour
Rate source ↗
B200 starting reference
From $5.99 per GPU-hour
Rate source ↗
Persistent storage
Region-specific hourly rate in console
Additional to instance pricing and retained independently.
Rate source ↗

Starting prices change with weekly supplier pricing and do not guarantee availability. Confirm GPU count, complete VM or bare-metal shape, reservation terms and extra volume costs. The provider’s comparative savings table is not evidence of equal workload performance or Sandbox0 savings.

Check current Hyperbolic 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 Hyperbolic

  1. Inventory onboard OS changes and scratch separately from persistent volumes, regions, GPU topology and reserved capacity.
  2. Export required onboard data before reprovisioning or termination; keep GPU work on suitable capacity and port compatible Linux agent stages.
  3. Test durable system paths, increasing outputs, supported paused files and the full instance-plus-storage bill, including recovery from host failure.

Common questions

Is Sandbox0 a drop-in Hyperbolic 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 ↗www.hyperbolic.ai/docs/on-demand/storage-and-ports.md ↗www.hyperbolic.ai/docs/on-demand/managing-instances.md ↗www.hyperbolic.ai/marketplace ↗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 →