The idea

There are three storage objects in this market. The root disk is part of the machine and disappears or is parked with it. A volume is a separate disk you attach, which can outlive the machine and sometimes attach to another. Object storage is a bucket reached over the network, not a filesystem.

What persists is the second question. Some providers keep the whole filesystem, including packages installed outside the home directory. Others keep one folder, usually /workspace or /data, and reset the rest from the template. The difference shows up the first time you install a database in /var/lib.

Retention is the third. A stopped machine's disk may be kept indefinitely, kept free for a fixed period, or deleted with the machine. A volume may be node-local and lost with the host. Read the retention line before treating a sandbox as storage.

Where providers split

The claim on the left, the provider's own words on the right

The whole disk, or one folder

Agent 37, boxd, Boat and Sprites persist the whole disk. Lizard persists only /workspace unless you pause rather than stop, and InstaBox gives you one disk at /data with the root disk owned by the template. E2B's snapshots cover the filesystem, but its separate volumes are the object that outlives a sandbox.

The narrow case is easy to miss because the marketing says “persistent”. At Lizard, a stop deletes everything outside /workspace; only a pause keeps the installed packages.

Volumes are a separate product

Runtime attaches up to four volumes per sandbox, and a volume can also be mounted read-only as a snapshot. Blaxel allows one volume per sandbox, attached to one sandbox at a time, and sells Agent Drive for shared access. Vercel's drives allow one read-write mount at a time. Morph's volumes are S3-compatible object storage, not an attached block disk.

Several providers have no volume product at all: Agent 37, Boat, Sandboxer and InstaBox. E2B's volumes exist but are in private beta.

Retention is where “persistent” ends

Boat keeps a stopped sandbox's disk for 14 days, then deletes it. Lizard's volumes are node-local, and there is no trash: deleting a sandbox or a project removes its data at once. Vercel's snapshots expire after 30 days by default unless you remove the expiry.

Agent 37 keeps the disk until you delete the instance, and charges $0.09 per GB-month for it. Sprites caps storage at 100 GB per Sprite and bills the bytes you write.

One worked case

The same job at every provider, from the documented defaults

Install Postgres in /var/lib, stop overnight

The agent installs Postgres with apt, initialises a database in /var/lib/postgresql, and writes a week of test data. You stop the machine at 22:00 and resume at 08:00.

  1. 21:00

    apt install writes outside the home folder, which is where the whole-disk and one-folder designs diverge.

  2. 22:00

    Stop. The disk is either kept with the machine, parked for a fixed period, or deleted.

  3. 08:00

    Resume and start Postgres. If the disk was reset to the template, the database is gone and the install has to run again.

A provider that persists the whole disk keeps the database; one that persists only /workspace or a single /data disk does not, and one that wipes the disk on stop loses it entirely. Retention is the second half: a 14-day window turns a long weekend into data loss.

What each provider does

Agent 37
The whole disk persists and bills until you delete the instance, but the nightly backup covers only /home/node, so the database is not in it.
Blaxel
The sandbox disk goes with the sandbox; only a volume survives deletion, and one volume attaches at a time.
Boat
The whole disk persists, including /var/lib, and is kept free for 14 days after a stop.
boxd
The fixed 100 GB disk persists in every state and bills on bytes written, not provisioned size.
Cloudflare Sandbox
Postgres in /var/lib is gone when the instance stops; only a snapshot or a mounted R2 bucket keeps the data.
Daytona
Postgres in /var/lib survives a stop because the whole disk persists, but it bills while stopped and cannot exceed 10 GiB.
E2B
Snapshots keep the filesystem, but the sandbox disk does not outlive the sandbox; separate volumes are in private beta.
InstaBox
One disk at /data persists and grows live; the root disk is owned by the template, so a package install outside /data is at risk.
Lizard
Only /workspace persists; a stop deletes everything else, including the database in /var/lib.
Modal
The database lives on the fixed 512 GiB root disk, which cannot be resized and is billed as memory at 20:1; a Volume is the durable option.
Morph
The instance disk holds the machine; Morph Volumes is S3-compatible object storage reached through a gateway, not an attached block disk.
Runtime
Up to four volumes mount at chosen paths, so the database belongs on a volume rather than the root disk.
Sandboxer
The whole disk persists while stopped, free within a storage quota; going over blocks new sandboxes rather than charging.
Sprites
The whole filesystem persists on durable local storage, capped at 100 GB, and bills the bytes written.
Vercel Sandbox
64 GB of ephemeral NVMe per sandbox; a drive bills on logical used size and only one sandbox can mount it read-write.
Tensorlake
The root disk is ephemeral, so anything that must outlive the sandbox goes on a mounted volume or a hosted Git repository.

What to watch for

The details that decide the answer in practice

  1. 01

    Whether the whole filesystem survives or only one mounted folder.

  2. 02

    Whether a volume can attach to more than one machine, and whether sharing is read-only.

  3. 03

    How long a stopped machine's disk is kept before deletion.

  4. 04

    Whether storage bills on provisioned size or written bytes.

  5. 05

    Whether snapshots include credentials and installed packages outside the home folder.

Compare all 16

Docs read 28 October 2026

“Persistent” can mean the whole disk, one mounted folder, or fourteen days. It can also mean a separate volume that outlives the machine, which is a different product from the disk the machine boots with.
ProviderDisk, and how long it is keptDoes it?What that means
Agent 37Whole disk$0.09/GB-moPartlyThe whole disk persists and bills at $0.09 per GB-month until you delete the instance. No volumes. Backups cover only /home/node, so a database in /var/lib is not in them.Agent 37 docs: Billing — disk rate, cold storage and what bills while stopped ↗
Agent 37's own wordsdocs-confirmed
“Disk | $0.09 per GB per month”
Agent 37docs-confirmed
Blaxel1 volumesurvives deleteYesA volume survives sandbox deletion, but only one volume attaches to a sandbox and a volume attaches to one sandbox at a time. Agent Drive is the shared filesystem, priced separately.Blaxel docs: Volumes for sandboxes ↗
Blaxel's own wordsdocs-confirmed
“you can only attach one volume to a sandbox. A volume can also only be attached to one sandbox at a time.”
Blaxeldocs-confirmed
Boat14 days after stopPartlyThe whole disk persists, with 51 GB on the smaller sizes and no volume product. A stopped sandbox's disk is kept free for 14 days, then deleted.Boat docs: Data retention and deletion ↗
Boat's own wordsdocs-confirmed
“boat keeps the disk for 14 days after the stop”
Boatdocs-confirmed
boxd100 GBbilled on writeYesEvery machine has a fixed 100 GB disk, and disk bills on bytes written in every state. Separate disks now have a guide: they outlive the machine they are attached to, and attach to one machine at a time.boxd docs: Disks ↗
boxd's own wordsdocs-confirmed
“Disk | 100 GB, fixed (copy-on-write)”
boxddocs-confirmed
Cloudflare Sandbox20 GBwiped on stopPartly20 GB maximum disk, deleted when the instance stops. A snapshot keeps the root filesystem for 30 days; an R2 directory backup or S3/R2 mount outlives every instance.Cloudflare Sandbox docs: Work with files in a sandbox ↗
Cloudflare Sandbox's own wordsdocs-confirmed
“Stopping the instance deletes the files on its disk.”
Cloudflare Sandboxdocs-confirmed
Daytona10 GiB capvolumes freePartlyThe whole disk persists and bills in stopped and paused states at $0.000108/GiB-hour. Volumes are S3-backed FUSE mounts, 100 per organisation, free, and not usable for block storage such as database tables.Daytona docs: Volumes ↗
Daytona's own wordsdocs-confirmed
“Daytona volumes are included at no additional cost. Each organization can create up to 100 volumes, and volume data does not count against your storage quota.”
Daytonadocs-confirmed
E2B10–20 GiBvolumes in betaPartlySnapshots keep the filesystem, and separate volumes persist across sandboxes, but volumes are in private beta and cannot be snapshotted. Included disk is 10 GiB on Hobby and 20 GiB on Pro.E2B docs: Volumes overview ↗
E2B's own wordsdocs-confirmed
“Disk | Included (10 GiB Hobby / 20 GiB Pro)”
E2Bdocs-confirmed
InstaBox20 GiB at /dataPartlyOne disk at /data, 20 GiB by default and growable, with the root disk owned by the template. No volumes and no second disk; the free plan's disk is fixed.InstaBox docs: Limits — disk sizes, storage caps and outbound bandwidth ↗
InstaBox's own wordsdocs-confirmed
“a 20 GiB persistent ext4 disk at /data”
InstaBoxdocs-confirmed
Lizard/workspace only1–50 GBPartlyThe local disk is lost on stop; only /workspace persists, through a named volume of 1–50 GB attached to one sandbox at a time and stored on one node. There is no trash.Lizard docs: Persistent Volumes — keep /workspace across runs ↗
Lizard's own wordsdocs-confirmed
“New volumes allow 1 to 50 GB, in whole GB, with 5 GB as the default.”
Lizarddocs-confirmed
Modal512 GiBbilled as memoryPartlyA fixed 512 GiB root filesystem, not resizable, billed by raising the memory request at a 20:1 ratio. Volumes are separate, survive the sandbox, are excluded from snapshots, and cost $0.09/GiB-month after 1 TiB.Modal docs: Sandbox files and Volumes ↗
Modal's own wordsdocs-confirmed
“On the VM runtime, the root filesystem is a fixed 512 GiB.”
Modaldocs-confirmed
MorphS3 volumes, not a diskPartlyThe instance disk holds the machine, and Morph Volumes is separate S3-compatible object storage reached through a gateway and CLI, not an attached block disk. EFS is a third option.Morph docs: Getting started with Morph Volumes ↗
Morph's own wordsdocs-confirmed
“Morph Volumes are persistent S3-compatible storage for build caches, release artifacts, checkpoints, and shared datasets.”
Morphdocs-confirmed
RuntimeUp to 4 volumesYesUp to four volumes per sandbox, each mounted at a path, with a read-only snapshot mode for sharing. Buckets from S3, R2 or GCS can appear as a directory without the key entering the sandbox.Runtime docs: Storage and backups ↗
Runtime's own wordsdocs-confirmed
“A create attaches up to four volumes, each at a path such as /data”
Runtimedocs-confirmed
SandboxerWhole diskquota, not billPartlyThe whole disk persists while stopped, free within a storage quota, and there is no separate volume or image registry. Going over quota blocks new sandboxes rather than charging.Sandboxer docs: Billing: storage ↗
Sandboxer's own wordsdocs-confirmed
“is space for your files, added on top of the OS image”
Sandboxerdocs-confirmed
Sprites100 GB capbytes writtenYesThe whole filesystem persists on durable local storage, capped at 100 GB, and checkpoints keep filesystem blocks that bill while retained. No attachable or shared volumes are documented.Sprites docs: Lifecycle and Persistence — Capacity and billing ↗
Sprites's own wordsdocs-confirmed
“you pay for the bytes you actually write, not the full 100 GB”
Spritesdocs-confirmed
Vercel Sandbox64 GB ephemeraldrivesPartly64 GB of ephemeral NVMe per sandbox, plus drives billed on logical used size. Only one sandbox at a time can mount a drive read-write; snapshots are the sharing mechanism.Vercel Sandbox docs: Drives ↗
Vercel Sandbox's own wordspricing-page
“Enterprise | 32 | 64GB | 15 | 64 GB”
Vercel Sandboxpricing-page
Tensorlake10–100 GB root8 volumesYesRoot disk 10–100 GB, ephemeral and growth-only. Cloud Volumes mount as ordinary POSIX directories, are durable and versioned with snapshots and time travel, and up to 8 can be mounted per sandbox, read-only, prefetched or pinned to a snapshot. Hosted Git repositories mount without a clone.Tensorlake docs: Mount Filesystems — durable volumes on an ephemeral root disk ↗
Tensorlake's own wordsdocs-confirmed
“The sandbox's root disk is ephemeral; a mounted filesystem is durable, survives the sandbox, and can be shared by many sandboxes at once.”
Tensorlakedocs-confirmed

The specific capabilities

Each one checked for every provider, with the catch where a version falls short. All capabilities

Storage: capabilities by provider
CapabilityAgent 37BlaxelBoatboxdCloudflare SandboxDaytonaE2BInstaBoxLizardModalMorphRuntimeSandboxerSpritesTensorlakeVercel Sandbox
Files survive a stopAgent 37: Yes. The disk bills at $0.09 per GB-month until you delete the instance.Blaxel: YesBoat: Yes. 14 days. The disk is deleted 14 days after a stop unless you saved it.boxd: YesCloudflare Sandbox: NoDaytona: YesE2B: NoInstaBox: YesLizard: NoModal: NoMorph: NoRuntime: Yes. A stopped disk is billed and kept until you delete the sandbox.Sandboxer: YesSprites: Yes. 100 GBTensorlake: With a catch. 48 h after terminate. A terminated machine can be brought back for 48 hours; after that its files are gone.Vercel Sandbox: Yes
Separate volumesAgent 37: NoBlaxel: Yes. Only one volume attaches to a sandbox, and it attaches to one sandbox at a time.Boat: Noboxd: Yes. A disk attaches to one machine at a time and lives on one host.Cloudflare Sandbox: With a catch. 20 GB disk. There is no separate disk that outlives the machine; a bucket can be mounted, and a snapshot lasts 30 days.Daytona: Yes. Volumes are network filesystems, not block disks, so a database table store is a poor fit.E2B: With a catch. Volumes exist but are in private beta and need access from support.InstaBox: NoLizard: Yes. 1–50 GB. A volume attaches to one sandbox at a time and lives on one node.Modal: Yes. Up to 1 TiB freeMorph: With a catch. The volume is object storage reached over S3, not a disk attached to the machine.Runtime: Yes. Up to 4Sandboxer: NoSprites: NoTensorlake: YesVercel Sandbox: Yes
One volume, several machinesAgent 37: NoBlaxel: With a catch. Shared storage is a separate product called Agent Drive; a volume itself serves one machine at a time.Boat: Noboxd: NoCloudflare Sandbox: With a catch. The shared option is a bucket, not a disk, so files are not locked and two machines can overwrite each other.Daytona: Yes. Writes are not transactional, so two machines writing one file means the last write wins.E2B: With a catch. One volume can serve several sandboxes, but volumes are in private beta.InstaBox: NoLizard: NoModal: Yes. Hundreds of writers. Two machines writing the same file means the last write wins.Morph: With a catch. Sharing happens through an S3 bucket, not a disk several machines mount.Runtime: With a catch. Only one machine can write a volume at a time; others can hold a read-only copy taken when they start.Sandboxer: NoSprites: NoTensorlake: Yes. Same-path writes are last-writer-wins, so two machines editing one file can clash.Vercel Sandbox: With a catch. Only one machine can write a drive; others get a read-only copy taken when they mount it.
Mount outside storageAgent 37: NoBlaxel: NoBoat: Noboxd: NoCloudflare Sandbox: YesDaytona: YesE2B: With a catch. Mounting a bucket needs a custom template with the right filesystem tool installed; it is not built in.InstaBox: NoLizard: NoModal: YesMorph: NoRuntime: YesSandboxer: NoSprites: NoTensorlake: NoVercel Sandbox: No
Large diskAgent 37: No. 80 GB max. The largest shape offers 80 GB of disk, short of 100 GB.Blaxel: Not publishedBoat: Yes. 251 GBboxd: Yes. 100 GBCloudflare Sandbox: No. 20 GB max. The largest disk is 20 GB, short of 100 GB.Daytona: No. 10 GiB cap. The organisation ceiling is 10 GiB of disk per sandbox, though some Windows presets list 50 GiB.E2B: No. 20 GiB max. The largest included disk is 20 GiB, short of 100 GB.InstaBox: Yes. Up to 1024 GiBLizard: No. 50 GB max. The largest size offers 50 GB of disk, short of 100 GB.Modal: Yes. 512 GiBMorph: Not publishedRuntime: Yes. 400 GiBSandboxer: Yes. 251 GBSprites: Yes. 100 GBTensorlake: Yes. 100 GB rootVercel Sandbox: Yes. 1 TiB drive

YesWith a catchNoNot publishedHover or select a mark to see the catch.

What keeping files costs

The price of keeping a machine's files between sessions, from each provider's rate card. Every rate card

ProviderCostDetail
Agent 37$0.09 per GiB-monthWhile the machine is stopped.
Blaxel$0.12 per GiB-monthKept as a separate volume.
BoatIncludedPart of the size or plan; no separate disk price.
boxd$0.0701 per GiB-monthWhile the machine is stopped.
Cloudflare SandboxLost on stopA stopped instance loses its disk; files have to be backed up to R2 or a snapshot and restored.
Daytona$0.0788 per GiB-monthWhile the machine is stopped.
E2BIncludedPart of the size or plan; no separate disk price.
InstaBox$0.0803 per GiB-monthWhile the machine is stopped.
LizardIncludedPart of the size or plan; no separate disk price.
ModalLost on stopModal sandboxes have no stopped state; keeping files needs a Volume or a snapshot.
Morph$0.0073 per GiB-monthCharged on disk plus memory, because memory is kept too.
Runtime$0.08 per GiB-monthWhile the machine is stopped.
SandboxerIncludedPart of the size or plan; no separate disk price.
Sprites$0.0197 per GiB-monthWhile the machine is stopped.
Tensorlake$0.07 per GiB-monthWhile the machine is stopped.
Vercel Sandbox$0.08 per GiB-monthWhile the machine is stopped.

Questions to ask before you sign

Each one targets something the documentation does not answer

  1. Does the whole filesystem survive a stop, or only /workspace or /data, and what resets from the template?
  2. How long is a stopped machine's disk kept before deletion, and is that in the docs or only in the terms?
  3. Does storage bill on provisioned capacity, written bytes or used bytes, and does a paused machine still pay for it?
  4. Can a volume attach to more than one machine, and is sharing read-only?
  5. Do snapshots include packages installed outside the home folder, and do they carry saved credentials?

Common questions

Answered from the capability checks; each links to the provider-by-provider detail and its sources

Which sandbox providers offer “files survive a stop”?
Stopping the machine keeps its disk; starting it again finds the files where they were. Fully: Agent 37, Blaxel, Boat, Daytona, InstaBox, Runtime, Sandboxer, Sprites, Vercel Sandbox and boxd. With a catch: Tensorlake. Not at all: Cloudflare Sandbox, E2B, Lizard, Modal and Morph. See each provider's catch
Which sandbox providers offer “separate volumes”?
Storage that lives on after the machine is deleted and can be attached to a new one. Fully: Blaxel, Daytona, Lizard, Modal, Runtime, Tensorlake, Vercel Sandbox and boxd. With a catch: Cloudflare Sandbox, E2B and Morph. Not at all: Agent 37, Boat, InstaBox, Sandboxer and Sprites. See each provider's catch
Which sandbox providers offer “one volume, several machines”?
Several machines can use the same storage at once. Fully: Daytona, Modal and Tensorlake. With a catch: Blaxel, Cloudflare Sandbox, E2B, Morph, Runtime and Vercel Sandbox. Not at all: Agent 37, Boat, InstaBox, Lizard, Sandboxer, Sprites and boxd. See each provider's catch
Which sandbox providers offer “mount outside storage”?
Attach an S3-style bucket or another outside filesystem as a folder. Fully: Cloudflare Sandbox, Daytona, Modal and Runtime. With a catch: E2B. Not at all: Agent 37, Blaxel, Boat, InstaBox, Lizard, Morph, Sandboxer, Sprites, Tensorlake, Vercel Sandbox and boxd. See each provider's catch
Which sandbox providers offer “large disk”?
Disks of 100 GiB or more are available. Fully: Boat, InstaBox, Modal, Runtime, Sandboxer, Sprites, Tensorlake, Vercel Sandbox and boxd. Not at all: Agent 37, Cloudflare Sandbox, Daytona, E2B and Lizard. See each provider's catch

Part of State and storage. Next: put two or three providers side by side.