GPU clouds / Verda CPU and GPU Instances alternative

A Verda alternative for the Linux agent around GPU workloads

Verda, formerly DataCrunch, provides CPU/GPU instances, clusters, serverless containers and independent storage. Its OS block volume can already grow live and be retained separately from compute. The current instance lifecycle removed the Hibernate label, so older lifecycle assumptions need to be checked against today’s shutdown and deletion actions.

Start buildingCompare capabilities ↓

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

When Verda CPU and GPU Instances fits

Verda fits accelerator workloads needing full instances, cluster interconnects and independent block/shared storage. Its retained OS block volumes and explicit live root-volume resizing can support returning agent workflows, with separate guest partition and filesystem steps.

When to consider Sandbox0

Consider Sandbox0 for compatible Linux agents around GPU jobs, with a complete durable writable RootFS, automatic free-space-triggered expansion and supported files while guest compute stays paused. Required accelerators and cloud services remain separate.

Capabilities

Verda CPU and GPU Instances and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeVerda CPU and 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 filesVerda CPU and GPU InstancesOS/data block volumes can remain after compute deletion and be attached to another instance; retained storage remains billed.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSVerda CPU and GPU InstancesThe OS root volume is a separately managed block device, preserving installed software and configuration outside separate mounts.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 stateVerda CPU and GPU InstancesShutdown retains the instance but remains billed; delete-with-retained-storage supports a new instance with the saved disk rather than saved process memory.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateVerda CPU and GPU InstancesThe removed Hibernate label described retained storage, not an established durable RAM/GPU-memory checkpoint contract.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.

Verda already supports live OS-volume enlargement followed by partition/filesystem growth. Sandbox0’s distinction is automatically growing an enabled writable RootFS as free space runs low, without making each capacity increment a separate administrator operation.

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 →

Shutdown continues billing; deleting compute can retain storage

The current guide says Shutdown keeps instance charges running. Hibernate was removed because it was equivalent to deleting compute while keeping storage. During deletion, no storage is selected for deletion by default; retained volumes continue billing and can attach to a new instance. Confirm each disk choice before deleting compute.

Source 1 ↗Source 2 ↗

The root volume already supports live enlargement

Verda explicitly supports enlarging the running instance’s OS block device. The administrator then grows its partition and filesystem before using the extra capacity. This is real online boot-disk expansion, with explicit sizing and filesystem operations rather than automatic free-space-triggered growth.

Source 1 ↗

Instance, cluster and serverless prices cover different shapes

The GPU instance list specifies CPU, host RAM and VRAM alongside each rate, and separate Spot and confidential-computing rows. Instant clusters and serverless containers have their own specifications and tariffs. Region and the exact deployment configuration affect the effective quote.

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.

Verda CPU and GPU Instances

Pricing reviewed 2026-10-19

1x H100 SXM5 80 GB instance
$4.296 per instance-hour
Published 30-CPU/120-GB-RAM shape; verify selected region and configuration.
Rate source ↗
1x H200 SXM5 141 GB instance
$5.602 per instance-hour
Rate source ↗
1x B200 SXM6 180 GB instance
$8.029 per instance-hour
The instance row differs from the Instant Cluster rate.
Rate source ↗
CPU-TURIN.4V.16G instance
$0.04800 per instance-hour
4 CPUs and 16 GB RAM.
Rate source ↗
NVMe block capacity
$0.2000 per GiB-month
Rate source ↗

These are published USD on-demand instance rows, not Spot or confidential variants. Retained storage continues billing, including after compute deletion. Confirm region, selected shape, storage and commitments; explicit filesystem growth is separate from block-device resizing.

Check current Verda CPU and 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 Verda CPU and GPU Instances

  1. Record current instance versus cluster/container product, OS/data volumes, region, Spot/commitment terms and retained-storage choices.
  2. Export required data and retain GPU services; use the current delete/storage lifecycle rather than assuming Hibernate saves RAM.
  3. Test compatible agent system retention, automatic growth and paused files separately from GPU configuration and retained-volume billing.

Common questions

Is Sandbox0 a drop-in Verda CPU and 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.verda.com/cpu-and-gpu-instances/shutdown-hibernate-and-delete/index.md ↗docs.verda.com/storage/block-volumes/attaching-a-block-volume/index.md ↗docs.verda.com/storage/block-volumes/resizing-a-block-volume/index.md ↗verda.com/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 →