Which sandbox providers offer “pause keeps memory”?

Pausing keeps memory and running programs, so the machine resumes exactly where it was. Fully: Blaxel, E2B, Lizard, Morph, Runtime, Sprites, Tensorlake and boxd. With a catch: Agent 37, Daytona, InstaBox and Sandboxer. Not at all: Boat, Cloudflare Sandbox, Modal and Vercel Sandbox.

Which sandbox providers offer “pre-started pools”?

Keep machines ready in advance so new work starts without waiting for a boot. Fully: Tensorlake. With a catch: Daytona, Modal, Runtime and Sandboxer. Not at all: Agent 37, Blaxel, Boat, Cloudflare Sandbox, E2B, InstaBox, Morph, Sprites, Vercel Sandbox and boxd.

Which sandbox providers offer “starts in under a second”?

The provider documents new or resumed machines being ready in under a second. Fully: Blaxel, Daytona, Lizard, Modal, Runtime, Sandboxer, Sprites, Tensorlake, Vercel Sandbox and boxd. With a catch: InstaBox. Not at all: Boat.

Lifecycle: stop, pause and sleep

The idea

Every provider offers a way to make a machine cheaper while you are not using it. They fall into four behaviours. Disk only: the machine is destroyed and its disk is kept, so files survive and processes do not. Disk and memory: RAM is saved, so the agent's process tree can resume mid-task. Warm memory: the machine is frozen in the host's RAM, which resumes fastest and usually keeps billing for memory. Delete: the disk goes too.

The vocabulary is not standardised. InstaBox calls a memory-preserving freeze “pause” and a power-off “stop”. Agent 37 has both, as sleep and stop. boxd has three parked states plus destroy. Lizard's stop deletes the disk. Boat's API calls a stopped sandbox archived, and its homepage still says “pause billing” even though the docs state there is no pause.

Memory-preserving resume is usually best-effort. Sandboxer restores memory only if a matching machine is free within two minutes; InstaBox reports processesPreserved false when it cold-boots instead; Agent 37 says memory usually survives a wake but not always. Anything that must outlive a pause belongs in a file.

Where providers split

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

Paused, frozen in place

boxd keeps the machine's memory in the host's RAM and resumes it in under a millisecond. Nothing is written to disk, which is why it can come back instantly, and why the memory keeps billing while the machine sits paused.

“They freeze the VMM process and keep memory, running processes, and open sockets intact.”

boxd, CLI commandsdocs-confirmed

Paused, written to disk

Sandboxer writes memory and device state to disk when a sandbox stops, then reads it back on resume. The read-back depends on a host of the same CPU type being free within two minutes; if none is, the sandbox boots cold with its files intact.

“Memory is restored on a machine with the same CPU type as the one the sandbox stopped on.”

Sandboxer, docs/snapshotsdocs-confirmed

Paused, fast pause keeps files only

Lizard's ordinary pause keeps the whole disk, the memory and the running processes. Its fast pause is the files-only variant: it keeps only the files and boots fresh when you resume. Both are called pause, and the difference is whether a half-finished task survives.

“keeps only the files: the whole disk and /workspace. It takes about a second at any memory size … resume() boots the sandbox fresh with its files, like a restart”

Lizard, sandboxes/quickstartdocs-confirmed

No pause at all

Boat has no paused state. A stop removes the virtual machine and keeps a copy of the disk; every resume is a new machine with that disk. The disk itself is kept for fourteen days, after which a resume fails and the work is gone.

“There is no pause. A stop removes the virtual machine and keeps a copy of the disk.”

Boat, snapshots/stopdocs-confirmed

One worked case

The same job at every provider, from the documented defaults

Stop at 18:00, resume at 09:00

The agent has a REPL with loaded data, a dev server on port 3000, a half-written file, and a cron job it installed. You stop the machine for the night and come back the next morning.

18:00
You stop the sandbox. What happens next is the provider's definition of stop, not yours.
09:00
You resume and ask the agent to continue. The question is whether the REPL still holds its data.
09:01
You check the dev server, the cron job and the IP address the agent had written down.

Twelve of the sixteen can restore memory in at least one mode, but not all make it the default. Boat and Cloudflare keep nothing, and Lizard's stop deletes the disk.

What each provider does

Agent 37: Sleep keeps memory and processes as a checkpoint and bills disk only; stop keeps the disk and loses the process tree.
Blaxel: Standby keeps process and filesystem state and stops memory charges, but the dev server's connections are gone.
Boat: Stop saves the disk and destroys the machine, so tomorrow is a reboot with a new IP and no running process.
boxd: Standby keeps RAM frozen and bills it; hibernate writes memory to disk and wakes in about 85 ms.
Cloudflare Sandbox: The instance stops, the disk and every process go with it, and tomorrow starts from a snapshot that restores files but not the REPL.
Daytona: The container stops: files kept, REPL gone. The VM class would pause with memory intact, and either way the disk keeps billing.
E2B: Pause saves the filesystem and RAM, so the process tree resumes; the default timeout would have killed it instead.
InstaBox: Pause keeps memory and running programs and resumes in about a second; stop loses them, and a paused machine cannot be stopped at all.
Lizard: Ordinary pause keeps the whole disk, memory and processes; fast pause keeps files only and boots fresh, under the same name.
Modal: There is nothing to pause: the sandbox is terminated and tomorrow starts from a filesystem snapshot, so the REPL is gone.
Morph: Pause snapshots memory and disk and suspends the VM; starting the snapshot restores memory, booting it is a cold boot.
Runtime: Pause freezes the VM with memory and programs intact; stop keeps files and drops memory, and both keep billing storage.
Sandboxer: Stop writes memory and device state to disk; the resume restores processes only on a matching CPU within two minutes.
Sprites: A warm pause freezes memory and processes and resumes in 100–500 ms; a longer stop drops memory and starts processes fresh.
Vercel Sandbox: Stopping ends the session; the filesystem is restored into a new session and background services restart through the onResume hook.
Tensorlake: Suspend freezes memory, processes and files, and resume returns the same sandbox in about a second; a terminated one can be restarted for 48 hours.

What to watch for

The details that decide the answer in practice

01
Whether the provider's “pause” keeps RAM, keeps only files, or destroys the machine.
02
Whether a memory restore is guaranteed or a fallback that silently cold-boots.
03
What resets: IP address, open sockets, cron timers, systemd units, Docker containers.
04
Whether stopping a paused machine is allowed; at InstaBox it is an error.
05
Whether the disk survives a stop at all; at Lizard and Morph, stop deletes it.

Compare all 16

Docs read 28 October 2026

Providers use seven words for four behaviours. The only question that matters is what is still running when the machine comes back.
Provider	What a pause keeps	Does it?	What that means
Agent 37	Sleep: RAM + disk	Partly	Sleep keeps memory and processes as a checkpoint and bills disk only; stop keeps the disk and loses memory. A wake that has to rebuild boots fresh, and stopping a sleeper discards its checkpoint.Agent 37 docs: Instance lifecycle — start, stop, restart, resize, update ↗ Agent 37's own wordsdocs-confirmed “Sleep and stop both release CPU and memory and bill disk alone. A sleeping instance keeps its memory and wakes on the next request; a stopped one loses its memory and stays down until start.” Agent 37docs-confirmed
Blaxel	Standby: RAM, no sockets	Yes	Standby preserves process and filesystem state and stops memory charges, but external network connections do not survive. Archive exports filesystem changes only and restarts processes with new IDs.Blaxel docs: Sandboxes ↗ Blaxel's own wordsdocs-confirmed “The snapshot preserves process and filesystem state, but not external network connections” Blaxeldocs-confirmed
Boat	No pausedisk only	No	There is no pause. Stop saves the disk, removes the machine and loses memory and hand-started processes; resume boots a new machine from the disk with a new IP.Boat docs: Concepts ↗ Boat's own wordsdocs-confirmed “There is no pause. A stop removes the virtual machine and keeps a copy of the disk.” Boatdocs-confirmed
boxd	RAM frozen85 ms wake	Yes	Four distinct states: standby keeps RAM frozen and bills it, hibernate writes memory to disk with an ~85 ms wake, stop keeps the disk only, destroy removes everything.boxd docs: Suspend, resume, and hibernate ↗ boxd's own wordsdocs-confirmed “Waking from hibernation takes about 85ms and loads the memory image back into RAM.” boxddocs-confirmed
Cloudflare Sandbox	No pausestop wipes disk	No	The instance runs only until something stops it; its processes and the files on its disk end with it. A snapshot carries files into a new instance, not memory. There is no paused state.Cloudflare Sandbox docs: Sandbox lifetime ↗ Cloudflare Sandbox's own wordsdocs-confirmed “An instance runs only until something stops it, and its processes and the files on its disk end with it.” Cloudflare Sandboxdocs-confirmed
Daytona	VM pause: RAMcontainer stop	Partly	Pause exists only on Linux VM and Windows sandboxes; a container stop keeps the filesystem and loses memory. Stopped and paused sandboxes bill reserved disk only, and archive is container-only.Daytona docs: Sandboxes: lifecycle states ↗ Daytona's own wordsdocs-confirmed “The filesystem and memory state are preserved, and CPU is no longer consumed.” Daytonadocs-confirmed
E2B	RAM + disk	Partly	Pause saves the filesystem and RAM, so processes resume; a filesystem-only pause is opt-in. Kill is terminal, and the default timeout kills the sandbox after five minutes.E2B docs: Sandbox lifecycle ↗ E2B's own wordsdocs-confirmed “When you pause a sandbox, both the sandbox's filesystem and memory state will be saved.” E2Bdocs-confirmed
InstaBox	RAM1 s resume	Partly	Pause keeps memory and running programs and resumes in about a second; stop loses them, and a paused machine cannot be stopped at all. A resume can cold-boot and report processesPreserved false.InstaBox docs: Machines — lifecycle, retention, auto-pause, sizes ↗ InstaBox's own wordsdocs-confirmed “carries on where it left off, usually in about a second” InstaBoxdocs-confirmed
Lizard	Pause: RAMfast pause: files	Partly	Pause keeps the whole disk, memory and processes; a fast pause keeps files only and resumes like a restart; stop or expiry deletes the sandbox disk, leaving only /workspace and snapshots.Lizard docs: Sandboxes — lifecycle, pause/resume, kill ↗ Lizard's own wordsdocs-confirmed “A stop (kill or expiration) deletes the sandbox's disk: only files on an attached Persistent Volume at /workspace and the snapshots you took remain.” Lizarddocs-confirmed
Modal	No pauseterminate	No	No paused or stopped state. Terminate sends SIGKILL; a memory snapshot terminates the source and restores a new sandbox within seven days. The nearest thing to a resume is an automatic exit snapshot, which is filesystem only.Modal docs: Sandboxes: lifecycle events and timeouts ↗ Modal's own wordsdocs-confirmed “Snapshotting a Sandbox will currently cause it to terminate.” Modaldocs-confirmed
Morph	Pause: RAM + disk	Partly	Pause snapshots memory and disk and suspends the VM; stop releases the instance and deletes it. Booting a snapshot is a cold boot, while starting one restores memory.Morph docs: Basic Operations (start, save, stop) ↗ Morph's own wordsdocs-confirmed “Use when you need to preserve the running state of applications, the memory contents, and all processes.” Morphdocs-confirmed
Runtime	Pause: RAM frozen	Yes	Pause freezes the VM with memory and programs intact; stop keeps files and drops memory. Both keep billing storage, and only delete ends the charge.Runtime docs: Time and lifetime: pause, stop or delete ↗ Runtime's own wordsdocs-confirmed “wake: it carries on where it was” Runtimedocs-confirmed
Sandboxer	RAM to disk2 min window	Partly	A stop writes memory and device state to disk, so a resume can restore running processes — but only on a machine of the same CPU type within two minutes, otherwise it cold-boots.Sandboxer docs: Snapshots (stop, resume, fork) ↗ Sandboxer's own wordsdocs-confirmed “If a machine with that CPU type is online but busy, the resume waits up to 2 minutes for it” Sandboxerdocs-confirmed
Sprites	Warm pause100–500 ms	Yes	A warm pause freezes memory and processes and resumes in 100–500 ms; a longer idle period stops the VM, drops memory and starts processes fresh. Open TCP connections do not survive either.Sprites docs: Lifecycle and Persistence ↗ Sprites's own wordsdocs-confirmed “The VM is suspended with everything in memory frozen in place. Compute billing stops. The next request resumes it in 100–500ms” Spritesdocs-confirmed
Vercel Sandbox	Filesystem only	Partly	Stopping ends the session. Persistent sandboxes save their filesystem and restore it into a new session, not RAM; restart background services with the onResume hook.Vercel Sandbox docs: Persistent sandboxes: sessions, stop and resume ↗ Vercel Sandbox's own wordsdocs-confirmed “Persistent sandboxes automatically save their filesystem state when stopped and restore it when resumed.” Vercel Sandboxdocs-confirmed
Tensorlake	RAM + processes~1 s resume	Yes	Named sandboxes suspend on idle or on request and resume under the same ID with the process tree alive. Ephemeral sandboxes cannot suspend and are deleted at their timeout. Terminated sandboxes stay restartable for 48 hours, restoring from the latest snapshot when one exists.Tensorlake docs: Lifecycle — ephemeral and named sandboxes, suspend, resume, restart ↗ Tensorlake's own wordsdocs-confirmed “A resumed sandbox continues from the exact memory and file system state, it was suspended.” Tensorlakedocs-confirmed

The specific capabilities

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

Lifecycle: capabilities by provider
Capability	Agent 37	Blaxel	Boat	boxd	Cloudflare Sandbox	Daytona	E2B	InstaBox	Lizard	Modal	Morph	Runtime	Sandboxer	Sprites	Tensorlake	Vercel Sandbox
Pause keeps memory	Agent 37: With a catch. A sleep usually comes back where it left off, but a wake that has to rebuild starts fresh and loses what was running.	Blaxel: Yes. Standby: 25 ms. Network connections do not survive; only files and processes do.	Boat: No	boxd: Yes. 85 ms wake. A hibernated machine wakes in about 85 ms; standby is faster still.	Cloudflare Sandbox: No	Daytona: With a catch. Only the full virtual machine can be paused with memory; a container sandbox loses what was running.	E2B: Yes	InstaBox: With a catch. RAM; 1 s resume. A resume can come back without the running programs, and the machine reports that it did.	Lizard: Yes. A fast pause keeps only the files, so it is the option to avoid if work must survive.	Modal: No	Morph: Yes	Runtime: Yes	Sandboxer: With a catch. 2 min window. Memory comes back only if a machine of the same processor type is free within two minutes; otherwise it boots fresh with the files.	Sprites: Yes. 100–500 ms. After a longer idle spell it is stopped and memory is dropped, and you are not told which happened.	Tensorlake: Yes. Only named sandboxes can be suspended; an unnamed one runs until it ends.	Vercel Sandbox: No
Pre-started pools	Agent 37: No	Blaxel: No	Boat: No	boxd: No	Cloudflare Sandbox: No	Daytona: With a catch. Pools exist, but they must be switched on for your organisation by asking support.	E2B: No	InstaBox: No	Lizard: Not published	Modal: With a catch. You have to build and run the pool yourself; there is no setting that keeps machines waiting.	Morph: No	Runtime: With a catch. There is no pool to size or refill; you keep spare paused machines yourself and wake one for each task.	Sandboxer: With a catch. The provider keeps machines ready, but you cannot say how many or which setup; a size the region does not keep boots instead.	Sprites: No	Tensorlake: Yes	Vercel Sandbox: No
Starts in under a second	Agent 37: Not published	Blaxel: Yes. Under 25 ms	Boat: No. A few seconds. A resume or a new machine takes a few seconds, not under one.	boxd: Yes. Sub-millisecond	Cloudflare Sandbox: Not published	Daytona: Yes. Under 90 ms	E2B: Not published	InstaBox: With a catch. About a second. A resume usually takes about a second, which is not under one.	Lizard: Yes. 228 ms median	Modal: Yes. Sub-second	Morph: Not published	Runtime: Yes. About 0.1 s	Sandboxer: Yes. 0.18 s median	Sprites: Yes. 100–500 ms	Tensorlake: Yes. Hundreds of ms	Vercel Sandbox: Yes. Sub-second

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

Questions to ask before you sign

Each one targets something the documentation does not answer

Does your pause keep RAM, keep only files, or destroy the machine — and what does your documentation call each state?
Is the memory restore guaranteed, or a best-effort that silently cold-boots when no matching host is free?
What changes on resume: IP address, open sockets, cron timers, systemd units, Docker containers?
Can a paused machine be stopped, and does stop delete the disk or only the machine?
How long is a stopped disk kept before deletion, and is that in the docs or only in the terms?

Common questions

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

Which sandbox providers offer “pause keeps memory”?: Pausing keeps memory and running programs, so the machine resumes exactly where it was. Fully: Blaxel, E2B, Lizard, Morph, Runtime, Sprites, Tensorlake and boxd. With a catch: Agent 37, Daytona, InstaBox and Sandboxer. Not at all: Boat, Cloudflare Sandbox, Modal and Vercel Sandbox. See each provider's catch
Which sandbox providers offer “pre-started pools”?: Keep machines ready in advance so new work starts without waiting for a boot. Fully: Tensorlake. With a catch: Daytona, Modal, Runtime and Sandboxer. Not at all: Agent 37, Blaxel, Boat, Cloudflare Sandbox, E2B, InstaBox, Morph, Sprites, Vercel Sandbox and boxd. See each provider's catch
Which sandbox providers offer “starts in under a second”?: The provider documents new or resumed machines being ready in under a second. Fully: Blaxel, Daytona, Lizard, Modal, Runtime, Sandboxer, Sprites, Tensorlake, Vercel Sandbox and boxd. With a catch: InstaBox. Not at all: Boat. See each provider's catch

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