GPU clouds / Vast.ai GPU Marketplace alternative

A Vast.ai alternative for the agent around GPU work

Vast.ai supplies marketplace GPU compute with host-specific prices and storage. A stopped container retains its disk while the instance exists, and separate local volumes survive instance destruction. These are real persistence features, with physical-machine and capacity boundaries that matter when an agent needs to return or recover elsewhere.

Start buildingCompare capabilities ↓

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

When Vast.ai GPU Marketplace fits

Vast.ai fits GPU training, inference and development that can choose hosts, rental types, templates and data-movement policies. Retained container files and independent local volumes are useful existing capabilities.

When to consider Sandbox0

Consider Sandbox0 for compatible Linux agents requiring a complete durable writable RootFS backed independently of disposable compute caches, automatic online growth and supported paused RootFS files. Keep required GPU execution separately.

Capabilities

Vast.ai GPU Marketplace and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeVast.ai GPU MarketplaceNot 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 filesVast.ai GPU MarketplaceContainer files remain while the instance exists, including stop; independent local volumes can outlive instance destruction.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSVast.ai GPU MarketplaceThe retained container allocation preserves its mutable container root, not just one workspace folder; destruction removes it and separate mounts have different lifecycles.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 stateVast.ai GPU MarketplaceInstances can stop and restart with retained disk, continuing storage charges and capacity-dependent restart.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateVast.ai GPU MarketplaceDocumented instance stop and retained local disk do not establish durable RAM/GPU-process checkpoint restoration.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.

Vast.ai explicitly fixes both container allocation and local-volume size at creation. Sandbox0 can automatically extend an enabled whole writable RootFS online as free space runs low, up to its configured logical ceiling.

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.

Vast.ai documents Cloud Copy between local volumes and connected cloud storage. Local volumes remain machine-bound and are separate from the complete instance root. Sandbox0 supports direct paused durable RootFS files, with mutations only for filesystem-only generations.

See paused file operations →

A stopped container keeps its complete container disk

The default container allocation retains files while stopped, including installed tools and configuration within that filesystem. Destroying the instance removes that allocation. Its size is fixed at creation and cannot be resized, so increasing accumulated agent state requires a separate migration or storage plan.

Source 1 ↗

Independent volumes remain tied to one physical machine

Local volumes survive instance destruction but can only attach to instances on the machine where they were created. Their size is also fixed. The current volume guide limits them to Docker instances rather than VM instances and documents Cloud Copy to connected cloud storage; retaining a volume is not cross-node whole-machine recovery.

Source 1 ↗

Compute, storage and bandwidth form the actual quote

Marketplace rates depend on the selected host and on-demand, reserved or interruptible rental. Storage continues while stopped and can cost more in the stopped state; upload and download charges also vary by host. Restart depends on compatible GPU availability. A marketed entry price is not a guaranteed offer for every host or resource configuration.

Source 1 ↗Source 2 ↗Source 3 ↗

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.

Vast.ai GPU Marketplace

Pricing reviewed 2026-10-19

Advertised H100 entry reference
As little as $0.90 per hour
Marketing entry reference; verify actual host, rental type and complete resource quote.
Rate source ↗
Compute billing
Host-specific rate, metered per second
Rate source ↗
Retained storage
Host-specific rate, including stopped periods
Storage remains chargeable while the allocation exists.
Rate source ↗
Upload and download traffic
Host-specific per-byte rates
Rate source ↗

On-demand, prepaid reserved and interruptible capacity use different terms. The advertised H100 entry rate does not guarantee current inventory under a particular filter. Compare the selected host’s compute, disk and traffic quote together, and account for retained storage after stopping compute.

Check current Vast.ai GPU Marketplace 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 Vast.ai GPU Marketplace

  1. Map container-root files, local volumes, host placement, rental type, disk limits and bandwidth rates.
  2. Export indispensable state before instance destruction and retain suitable GPU capacity and cloud-copy integrations.
  3. Port compatible Linux agents and test whole-system retention, online growth, paused files and recovery separately from GPU performance.

Common questions

Is Sandbox0 a drop-in Vast.ai GPU Marketplace 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.vast.ai/guides/instances/storage/types.md ↗docs.vast.ai/guides/instances/storage/volumes.md ↗docs.vast.ai/guides/instances/pricing.md ↗docs.vast.ai/guides/instances/manage-instances.md ↗vast.ai/products/gpu-cloud ↗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 →