GPU clouds / OVHcloud GPU instances alternative

An OVHcloud GPU instances alternative for returning Linux agents

OVHcloud GPU instances are full cloud VMs with accelerator-specific storage and network shapes. Its control panel and OpenStack use different lifecycle names: control-panel Suspend means shelve, while OpenStack suspend and pause keep resource billing. Compare the selected GPU configuration’s boot and passthrough storage separately.

Start buildingCompare capabilities ↓

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

When OVHcloud GPU instances fits

OVHcloud fits workloads needing GPU VMs, cloud networking, custom images and an OpenStack-oriented operating model. Snapshot-backed shelving can preserve boot-disk data while releasing the original compute resources.

When to consider Sandbox0

Consider Sandbox0 for compatible Linux agents around GPU work, with the whole durable writable RootFS, automatic online expansion and supported paused-file access. Retain GPU execution and required cloud infrastructure separately.

Capabilities

OVHcloud GPU instances and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeOVHcloud GPU instancesNot 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 filesOVHcloud GPU instancesShelving retains the instance disk in a snapshot; memory and separate passthrough or ephemeral data require their own preservation plan.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSOVHcloud GPU instancesInstance disk backups can preserve the boot filesystem and installed software; separate mounts are outside that disk image.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 stateOVHcloud GPU instancesShelve/unshelve provides disk continuity; OpenStack suspend and pause are different, still-billed operations.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateOVHcloud GPU instancesGeneric VM suspend/pause can retain VM state, but durable GPU/RAM restore compatibility is not established for every GPU shape.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.

OVHcloud instance shapes, separate storage and snapshot migration are explicit infrastructure controls. Confirm the selected GPU disk and filesystem resize workflow; those features alone do not establish automatic free-space-triggered expansion of a whole durable agent 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.

Snapshot-backed shelving retains disk data without the original running instance. It does not by itself establish direct arbitrary file operations on that saved whole root without another runtime. Sandbox0 provides supported paused RootFS operations, including the read-only restriction for memory-paused generations.

See paused file operations →

Choose the lifecycle operation by its actual behavior

The current guide says shelve saves the disk to a snapshot, releases compute and preserves the main IP. Memory is not retained. OpenStack suspend stores VM state on disk while keeping resources reserved, and pause freezes state in RAM; both remain billed at the same instance rate. GPU execution-state compatibility still needs confirmation for the chosen accelerator shape.

Source 1 ↗

Disk snapshots are not every mounted storage device

The documented shelve backup retains the instance disk, while data in memory and elsewhere is not retained. The guide explicitly warns about losing NVMe passthrough data on selected instance families. The H100 reference shape includes additional passthrough NVMe: copy indispensable scratch to durable storage and verify its exact lifecycle before shelving.

Source 1 ↗Source 2 ↗

Hourly and monthly lifecycle bills differ

Under hourly shelving, snapshot storage remains chargeable. A monthly-billed instance shelved for the whole month has only snapshot charges under the guide, but unshelving even briefly triggers the full monthly instance charge. The guide also contains broad wording about continued charges while an instance exists; confirm the exact operation and billing model in the account.

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.

OVHcloud GPU instances

Pricing reviewed 2026-10-19

Worldwide h100-380 list reference
$2.99 per instance-hour, excluding VAT
1 H100, 30 vCores, 380 GB memory, 200 GB NVMe plus 3.84 TB passthrough NVMe; confirm region and current quote.
Rate source ↗
Worldwide l4-90 list reference
$1.00 per instance-hour, excluding VAT
1 L4, 22 vCores and 90 GB memory; different accelerator and throughput.
Rate source ↗
Shelved hourly instance
Snapshot and applicable retained-resource charges
Monthly instance billing follows a different full-month rule.
Rate source ↗

References are published worldwide list entries for complete instance shapes, not universal GPU device-hour rates or a selected-region availability guarantee. Verify the account region, currency, storage, public networking, billing model and commitment. AI Training, AI Deploy and AI Notebooks are separate offers.

Check current OVHcloud GPU instances 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 OVHcloud GPU instances

  1. Record exact OpenStack and control-panel operation names, hourly versus monthly terms, boot disk and passthrough data, GPU shape and IP policy.
  2. Preserve required files and boot images before shelving or replacing a source, and port compatible Linux agent stages while keeping GPU jobs separate.
  3. Test system retention, storage growth, supported paused-file access and restart requirements, then compare the complete configuration and retained-resource bill.

Common questions

Is Sandbox0 a drop-in OVHcloud GPU instances 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.ovhcloud.com/en/guides/public-cloud/compute/suspend-or-pause-an-instance.md ↗www.ovhcloud.com/en/public-cloud/prices/ ↗docs.ovhcloud.com/en/guides/public-cloud/compute/create-restore-a-virtual-server-with-a-backup.md ↗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 →