Vercel Sandbox vs Unikraft Cloud

Both run each sandbox in a Firecracker microVM. Vercel Sandbox bills each run as it goes and stops every session after 24 hours, while Unikraft Cloud sells a flat pool of running capacity whose microVMs sleep when idle and wake in milliseconds.

The companies and products

Vercel Sandbox

Vercel is a cloud platform for web apps and agents, and Vercel Sandbox is its product for running untrusted or agent-written code in isolated Linux microVMs.

Each sandbox is a Firecracker microVM with its own kernel, started from a Vercel image, your own OCI image, or a saved snapshot. You drive it with the JavaScript or Python SDK or the sandbox CLI: run commands, move files, open ports, and set network rules. It is aimed at code interpreters, app builders, test runs and agent tools, and Vercel's docs say it is not meant to run nonstop.

Unikraft Cloud

Unikraft Cloud is a microVM cloud from Unikraft Inc. that runs any Dockerfile-based workload as a minimal Linux microVM with millisecond cold starts and scale-to-zero.

Your Dockerfile becomes the VM image at build time, and each instance runs on a heavily customized Firecracker VMM, so it boots in milliseconds and can go into standby when no traffic arrives. It is a general compute platform built for dense, bursty work such as HTTP services, headless browsers and agent fleets, sold as flat self-serve plans plus Enterprise dedicated hosts, on-prem and bring your own cloud. It is not a dev machine you log into: instances are reached through services, the CLI and a sandbox plugin for commands and files.

Which use cases each one fits

For each use case, Battleships prices every sandbox provider on the same workload and sorts them from the cheapest monthly bill up. “22nd cheapest of 345” means 21 of the other 344 providers run that workload, with every feature it needs, for less. “With limits” means the provider runs it but misses something, so it is listed after every full match. Open a use case for its whole ranking and what each provider misses.

Use case and its workloadVercel Sandbox: price rankUnikraft Cloud: price rankboat: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
Neither has a desktop to check the UI an agent built, Vercel stops at 24 hours, and Unikraft documents no Docker inside an instance; boat has both.
244th of 367, with limits
$3,351 a month
Can't run it3rd cheapest of 367
$634 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
Vercel's docs say sandboxes are not meant to run nonstop, and Unikraft's Team pool is 8 GiB of memory for every running instance together.
257th of 366, with limits
$3,657 a month
Can't run it5th cheapest of 366
$657 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
Short code bursts suit both designs, but only Vercel Sandbox has a published rate for them; Unikraft's self-serve plans cap you at 24 running instances.
22nd cheapest of 345
$216 a month
Can't run it186th of 345, with limits
$2,000 a month
Sandboxes for your users
300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month
Vercel fits hundreds of sandboxes for your users on Pro; Unikraft's biggest self-serve pool holds 24 running instances, so this needs an Enterprise quote there, and boat ranks first.
15th cheapest of 361
$1,267 a month
Can't run itCheapest of 361
$209 a month
RL rollouts
1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month
Many short parallel rollouts suit Vercel's high concurrency; Unikraft has no self-serve pool big enough to run them at once.
3rd cheapest of 360
$15,761 a month
Can't run it34th of 360, with limits
$2,000 a month
Agent evals
250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month
69th cheapest of 364
$1,508 a month
Can't run it10th cheapest of 364
$200 a month
CI runners
40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month
CI jobs that build Docker images fit Vercel, which allows Docker with sudo; Unikraft documents no container runtime inside, and boat adds /dev/kvm for emulators.
107th cheapest of 377
$1,144 a month
Can't run it4th cheapest of 377
$86 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
Unikraft keeps a box up all month with no session cap for its $39 Team pool, while Vercel restarts it every 24 hours; boat does both for $26 a month with SSH.
276th of 368, with limits
$145 a month
219th of 368, with limits
$39 a month
18th cheapest of 368
$26 a month
Computer-use agents
50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month
Neither ships a desktop for an agent to see and click; boat streams one at 60fps with a computer-use API.
235th of 369, with limits
$952 a month
Can't run it6th cheapest of 369
$200 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
259th of 367, with limits
$486 a month
Can't run it7th cheapest of 367
$200 a month
Scraping & bots (startup scale)
50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month
277th of 346, with limits
$468 a month
Can't run it6th cheapest of 346
$90 a month
Mass scraping & bots (undetectable)
1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month
234th of 338, with limits
$9,960 a month
Can't run it71st of 338, with limits
$1,800 a month
Parallel red teaming
200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month
283rd of 360, with limits
$554 a month
Can't run it2nd cheapest of 360
$200 a month
GPU eval jobs
10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month
Can't run itCan't run itCan't run it
macOS agents
5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month
Can't run itCan't run itCan't run it

Key features

FeatureVercel SandboxUnikraft Cloudboat
IsolationFirecracker microVM with its own kernelMinimal Linux microVM on a customized Firecracker VMMFull VM with its own kernel on dedicated AMD Ryzen 9 9950X hosts
Longest session24 hours on Pro and Enterprise, 45 minutes on Hobby; resume starts a new sessionNo published maximum; instances run until stopped or put in standbyNo cap on paid plans (--no-auto-stop); default TTL 1 hour, changeable
Sleep and wakeFilesystem snapshot on stop; memory is not kept and there is no wake on requestStandby after a 1 second idle cooldown by default; stateful mode keeps memory and traffic wakes it in millisecondsDisk saved every minute and on stop; a stopped sandbox costs nothing and boots again on start
Machine sizeUp to 8 vCPU / 16 GB per sandbox on Pro, 32 vCPU / 64 GB on EnterpriseA plan pool shared by all running instances: 2 vCPU / 4 GiB, 8 vCPU / 8 GiB or 24 vCPU / 32 GiB4 vCPU / 8 GB by default, up to 16 vCPU / 32 GB per sandbox on the $100 plan and above
Docker insideYes, with sudoThe Dockerfile becomes the VM image; no container runtime inside is documentedYes, with root and systemd
Nested virtualizationNoYes, with a per-instance flag, Enterprise onlyYes, /dev/kvm on every plan
SSHNo; commands and files go through the SDK and CLINot yet; commands and files go through the sandbox plugin and CLIYes, into every sandbox
Desktop and browserNone; headless Linux onlyNone built in; you can run your own headless browser from a Dockerfile60fps desktop and Chromium, with computer-use and browser-control APIs
Coding agentsClaude Code, Codex, OpenCode and pi preinstalled; you start them as commandsNone preinstalled; you bring the agent in your own imageSeven agents preinstalled, with an API to prompt, steer and stream events
Inbound accessPublic URLs for up to 15 ports, no built-in auth; no raw TCP or public IPServices on its load balancer: HTTP, HTTPS and TLS-wrapped TCP; no public IP per instanceHTTPS URLs for up to 50 ports, token-protected by default; raw TCP; dedicated IPv4 or IPv6
Outbound rulesAllow-all, deny-all or a domain and CIDR allowlist, changeable while runningNetwork shield filters traffic, Enterprise only, limited accessOpen internet, no provider-side allowlist; firewall inside the VM
Keys kept outside the sandboxCredentials brokering injects headers for any HTTPS domain you allowNetwork shield can inject secrets, Enterprise only, limited accessCredentials proxy for model providers and any HTTPS API, on every plan; switch a credential to another account when one runs out
Regions19 regions across the US, Europe, Asia, South America, Africa and AustraliaDallas, Frankfurt, San Francisco, Singapore and Washington DC; more on requestEU only: Germany, Finland, France
Self-hosting and GPUsNo self-hosting and no GPUsDedicated hosts, on-prem, your own cloud, GPUs and arm64 on EnterpriseNo self-hosting, GPUs or arm64
Every feature Battleships tracks (43)
FeatureVercel SandboxUnikraft Cloudboat
Full VM (own kernel)YesYesYes
Docker insideYesNoYes
Nested virtualizationNoYesYes
RootYesYesYes
SystemdNot publishedNot publishedYes
Desktop GUINoNoYes
BrowserNoNoYes
Computer-use APINoNoYes
SSHNoNoYes
Persistent diskYesYesYes
Pause/resumeYesYesYes
Automatic snapshotsYesYesYes
Fork/cloneYesYesYes
Live fork (no pause)YesYesYes
Memory forkNoYesNo
Custom image (Docker or snapshot)YesYesYes
≥24 h sessionsYesYesYes
Idle auto-stopNoYesNo
Wake on requestNoYesNo
HTTPS preview URLsYesYesYes
Raw TCP inboundNoNot publishedYes
Public IPv4NoNoNo
Egress allowlistYesYesNo
Secret proxyYesYesYes
Secret proxy for any APIYesYesYes
Preinstalled agentsYesNoYes
Agent harness APINoNoYes
MCP serverYesNoNo
GPUNoYesNo
WindowsNoNoNo
Arm64NoYesNo
Self-hostingNoYesNo
BYOCNoYesNo
Open sourceNoNoNo
Scoped API keysNot publishedNot publishedYes
Spend limitsYesNot publishedYes
Usage per sandbox (API)NoNoYes
Audit logsNot publishedYesNo
SSOYesNoNo
SOC 2YesYesNo
HIPAAYesYesNo
EU data residencyYesYesYes
Zero data retentionNot publishedNot publishedYes

Performance

Vercel Sandbox12.96 runs/s
Unikraft CloudNot measured
boat26.69 runs/s

On the same hpc-sandbox-benchmarks Node.js loop at 4 vCPU / 8 GB, boat did 26.69 runs per second and Vercel Sandbox did 12.96, so boat finished about twice as many runs in the same time. Unikraft Cloud is not in that benchmark, so there is no comparable CPU score for it. Unikraft states cold starts under 10 ms, which is its own figure and measures start time, not CPU speed.

Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Unikraft Cloud has no published run in this suite yet, so there is no like-for-like number to compare.

Pricing in five scenarios

Each bill is priced from the providers' public price lists by the open Battleships engine, with the cheapest plan that runs the scenario.

ScenarioVercel Sandbox, a monthUnikraft Cloud, a monthboat, a month
One coding agent, workdays
4 vCPU / 8 GB, 50 GB disk, 8 hours a day, 22 days a month, files kept between days
$57 · Pro$39 · Hobby$20 · $20/mo
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$236 · Pro · with limits$39 · Hobby$26 · $20/mo
50 agents in parallel
50 machines of 4 vCPU / 8 GB at once, 8-hour sessions, 1,100 sessions a month, Docker inside
$2,884 · ProCan't run it$317 · $20/mo × 3 seats
Sandboxes for your users
2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once
$828 · ProCan't run it$200 · $100/mo × 2 seats
1,000 short jobs at once
2 vCPU / 4 GB, 1,000 at once, 3 minutes each, 200,000 jobs a month
$3,152 · ProCan't run it$2,000 · $2000/mo · with limits

Vercel Sandbox: Vercel charges for CPU only while your code uses it, but for memory for the whole time the sandbox runs, plus a small fee per sandbox created and a monthly fee per GB of snapshots. The $20 Pro credit is shared with your other Vercel usage.

Unikraft Cloud: Unikraft charges a flat monthly fee for a pool of running capacity: free for 2 vCPU / 4 GiB, $39 for 8 vCPU / 8 GiB, $199 for 24 vCPU / 32 GiB, with sleeping instances not counted. It publishes no per-hour or egress rate, so anything bigger is an Enterprise quote.

Unikraft Cloud: boat charges a flat price per size, per second, only while the machine runs; a stopped sandbox costs nothing. Plans start at $20 a month, and all of the fee is usage credit.

Limits in those scenarios

For one agent that never stops, Vercel ends each session at 24 hours and running processes start over, while Unikraft's $39 Team pool keeps it up. For 50 agents, sandboxes for your users and 1,000 short jobs at once, Unikraft has no self-serve pool big enough and no published rate, while Vercel fits all three on Pro. boat fits every scenario except the 1,000-job burst, where even its $2,000 plan queues starts above 600 an hour.

ScenarioVercel SandboxUnikraft Cloudboat
One coding agent, workdaysNone that bite
  • flat pool: the tier is billed whether used or not
  • disk beyond 0 GiB: price unknown
  • egress price not published
None that bite
One agent that never stops
  • Pro: sessions capped at 24 h
  • flat pool: the tier is billed whether used or not
  • disk beyond 0 GiB: price unknown
  • egress price not published
None that bite
50 agents in parallelNone that bite
  • no pool tier with 200 vCPU / 400 GiB for 50 VMs
  • no published per-resource rate
None that bite
Sandboxes for your usersNone that bite
  • no pool tier with 600 vCPU / 1,200 GiB for 300 VMs
  • no published per-resource rate
None that bite
1,000 short jobs at onceNone that bite
  • no pool tier with 2,000 vCPU / 4,000 GiB for 1,000 VMs
  • no published per-resource rate
  • max 2,400 starts/day (needs ~6,667)
  • max 600 starts/hour (needs ~1,000)
  • bursts above 90 starts/min are queued (a burst of 333 takes ~4 min to start)

Which one should you use?

Use Vercel Sandbox when

you run many short bursts of untrusted code at high concurrency, want pay-as-you-go rates, or need regions all over the world.

  • Big fleets without a sales call. Useful if you run hundreds of sandboxes at once: Vercel Pro allows high concurrency, and Unikraft's self-serve plans stop at 24 running instances.
  • Regions all over the world. Useful if your users are spread out: Vercel Sandbox runs in 19 regions, from Virginia to Tokyo, Sydney and Cape Town.
  • A firewall on every paid plan. A running sandbox can switch to deny-all or a domain allowlist; on Unikraft, egress filtering is Enterprise only.

Use Unikraft Cloud when

you run small services or headless browsers that sleep most of the time and must wake in milliseconds, or need GPUs, on-prem or your own cloud under an Enterprise contract.

  • Millisecond wake and scale-to-zero. Best for many small services that sleep most of the time and must answer the next request in milliseconds, with memory kept.
  • No 24-hour session cap. Useful for a box that stays up all month: Unikraft instances run until stopped, while Vercel ends each session at 24 hours.
  • Your own hardware on Enterprise. Enterprise teams can get GPUs, arm64, dedicated hosts, on-prem or their own cloud, which Vercel Sandbox does not offer.

Where boat fits

Vercel Sandbox and Unikraft Cloud are both built for short or sleepy work in small microVMs. If your agents work for hours on one machine, with Docker, emulators, a desktop and a browser, look at boat. It also runs fleets of full-size sandboxes on self-serve plans.

A whole machine with SSH

Agents install packages, start databases with Docker and run emulators. Vercel has no /dev/kvm or SSH, and Unikraft has no SSH, no documented Docker inside and nested virtualization only on Enterprise. boat gives root, systemd, Docker, /dev/kvm and SSH on every plan.

A desktop and agents you drive

Neither has a screen, so an agent cannot open the app it built to check it, and you build the layer that prompts and reads each agent yourself. boat streams a 60fps desktop with Chromium and drives seven agents inside from outside through one API.

Long runs at full size

Vercel stops each session at 24 hours and running processes start over. Unikraft has no cap, but its Team pool is 8 GiB in total and 50 agents need an Enterprise quote. boat has no session cap and runs 100 sandboxes from its first paid plan.

Use boat when your agents need a real machine for hours: Docker, emulators, SSH, a desktop and a browser, coding agents you drive from outside, and fleets of full-size sandboxes without a sales call. Compare it one to one: boat vs Vercel Sandbox and boat vs Unikraft Cloud.

Questions people ask

Is Vercel Sandbox cheaper than Unikraft Cloud?

For one agent, Unikraft is cheaper: one coding agent on workdays costs $39 a month on Unikraft against $57 on Vercel, and one that never stops costs $39 against $236. For 50 agents or sandboxes for your users, Vercel fits at $2,884 and $828, while Unikraft needs an Enterprise quote. boat costs $20, $26, $317 and $200 for those four.

Which is faster, Vercel Sandbox or Unikraft Cloud?

For CPU speed there is no fair answer: Vercel did 12.96 runs per second in the hpc-sandbox-benchmarks loop and Unikraft was not measured. For start time, Unikraft is built to cold boot in milliseconds and wake on the next request. boat did 26.69 runs per second in the same benchmark.

Can Vercel Sandbox or Unikraft Cloud run Docker?

Vercel Sandbox can run Docker inside with sudo. Unikraft uses your Dockerfile to build the VM image, and its docs do not describe a container runtime inside a running instance. boat runs Docker with root and systemd, and adds /dev/kvm.

Can either run 24/7?

Unikraft instances have no published session limit and run until stopped or put to sleep. Vercel sessions stop after 24 hours on Pro, and its docs say sandboxes are not designed to run continuously. boat has no session cap on paid plans.

Do they have SSH or a desktop?

No. Both are headless, and you run commands through an SDK, plugin or CLI; Unikraft has announced SSH as coming soon. boat has SSH into every sandbox and a 60fps desktop with Chromium.

Is boat SOC 2 compliant?

SOC 2 is in progress. We're open about where we stand: live status at status.ascii.dev and our data retention policy at docs.boat.dev/data-retention.

Try boat

Install the CLI, then run boat new.

7-day free trial

More comparisons