boat vs Clever Cloud

Clever Cloud is a platform for hosting apps: you push code or a Docker image, and it runs on fresh, read-only VMs that it rebuilds on every deploy. boat gives each agent its own VM it can change, with root, Docker, /dev/kvm, a disk that keeps its files and coding agents built in.

Where boat stands out

A machine the agent can change

On boat the agent has root, systemd, Docker and /dev/kvm, so it can install packages, start services and run emulators. On Clever Cloud you log in as a non-root user on a read-only instance, and only one container runs per app unless you open the Docker socket.

Files that stay, and forks

A boat sandbox keeps its disk, takes snapshots every minute and on stop, and costs nothing while stopped. You can fork a running sandbox to try two paths at once. Clever Cloud rebuilds the VM on every restart, so the agent's work is lost unless it was written to an add-on.

Agents, desktop and keys built in

boat ships Claude Code, Codex, OpenCode and other agents inside each VM, which you drive from outside with a prompt, steer and events API. Every VM also has a 60fps desktop with Chromium, and the credentials proxy keeps real API keys out of the machine.

The companies and products

boat

boat.dev runs sandboxes for AI agents: each one is a full Linux VM with its own kernel.

You start a machine with one command or one API call, then SSH in, run Docker, open a desktop, or let a built-in agent (Claude Code, Codex, OpenCode and others) work in it. The disk stays between sessions, and a stopped sandbox costs nothing.

Clever Cloud

Clever Cloud is a European platform as a service (PaaS) based in Nantes, France: it hosts web apps, workers and databases for development teams.

You deploy code in one of many runtimes, or a Docker image, and Clever Cloud runs it on instances called scalers, each in its own virtual machine, with autoscaling, managed add-ons and HTTPS routing. Instances are read-only: every deploy or restart builds a fresh VM, and anything that must last goes into an add-on such as a database, an FS Bucket or Cellar object storage. It is a different kind of product from an agent sandbox, so this page compares the parts where the two overlap.

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. “3rd cheapest of 367” means 2 of the other 366 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 workloadboat: price rankClever Cloud: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
Clever Cloud is a compromise here: an agent's installs and edits vanish when the instance restarts, and it logs in as a non-root user, while boat keeps the disk and gives root.
3rd cheapest of 367
$634 a month
239th of 367, with limits
$3,233 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
An assistant that runs for weeks needs files that last; on Clever Cloud they must live in a database or storage add-on, while boat keeps them on the VM's disk.
5th cheapest of 366
$657 a month
246th of 366, with limits
$3,352 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
Clever Tasks fit short jobs well: Clever Cloud starts a VM, runs the command, stops the VM and bills by the second, with no plan fee.
186th of 345, with limits
$2,000 a month
88th of 345, with limits
$129 a month
Sandboxes for your users
300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month
Each user sandbox on Clever Cloud is a separate app built from a deploy, with no fork or snapshot, so creating hundreds on demand is slow work; boat creates and forks VMs by API.
Cheapest of 361
$209 a month
249th of 361, with limits
$1,068 a month
RL rollouts
1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month
34th of 360, with limits
$2,000 a month
83rd of 360, with limits
$4,592 a month
Agent evals
250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month
10th cheapest of 364
$200 a month
147th of 364, with limits
$918 a month
CI runners
40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month
Both bill by the second; boat's CI jobs get root, /dev/kvm and a persistent cache disk, while a Clever Cloud task starts from a fresh build each time.
4th cheapest of 377
$86 a month
160th of 377, with limits
$441 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
18th cheapest of 368
$26 a month
124th of 368, with limits
$134 a month
Computer-use agents
50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month
Clever Cloud has no desktop, browser or computer-use tooling; boat streams a 60fps desktop with Chromium on every VM.
6th cheapest of 369
$200 a month
232nd of 369, with limits
$918 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
On Clever Cloud you would package your own headless browser in a Docker image; boat ships Chromium and browser control on every VM.
7th cheapest of 367
$200 a month
254th of 367, with limits
$459 a month
Scraping & bots (startup scale)
50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month
6th cheapest of 346
$90 a month
221st of 346, with limits
$193 a month
Mass scraping & bots (undetectable)
1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month
71st of 338, with limits
$1,800 a month
132nd of 338, with limits
$3,867 a month
Parallel red teaming
200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month
2nd cheapest of 360
$200 a month
215th of 360, with limits
$230 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 it
macOS agents
5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month
Can't run itCan't run it

Key features

FeatureboatClever Cloud
Kind of productAgent sandbox: a VM per agent, created and controlled by APIApp hosting platform: deploy code or a Docker image, scale it, add managed databases
IsolationEvery sandbox is a full VM with its own kernelEvery app instance runs in its own VM; Docker containers run inside it
Root and system accessRoot, systemd and /dev/kvm on every machineSSH in as a non-root user on a read-only instance; no nested virtualization documented
Docker insideDocker and Docker Compose run inside the VM like on any Linux serverYour image runs as one container; the Docker socket is opt-in (CC_MOUNT_DOCKER_SOCKET) and Docker Compose is not supported
Files after a restartThe disk persists, with snapshots every minute and on stop; a stopped sandbox costs nothingChanges on an instance are lost on redeploy or restart; data goes in add-ons (FS Buckets are not available to Docker apps)
Fork and snapshotsFork a running sandbox; named snapshots as templatesNo instance snapshot or fork; new instances always start from a build
Reaching the machineHTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4HTTPS on one port (8080) with custom domains; one TCP port, mapped to a random public port above 5000
Run-to-completion jobsStart a VM, run the job, stop it; starts per minute are capped by planClever Tasks start a VM, run one command and stop the VM, billed by the second
Managed databases and add-onsNone; you run databases yourself inside the VMManaged databases, object storage, FS Buckets and more, linked to apps by environment variables
Desktop and computer use60fps desktop and Chromium, computer use and browser control, embeddable streamNone
Coding agentsClaude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events APINone inside apps; an MCP server and CLI let agents manage the platform from outside
Keeping keys out of the machineAI credentials proxy for model providers and any HTTPS API you register; switch a credential to another account when credits run outCredentials are injected as environment variables into the instance
RegionsEU only: Germany, Finland, FranceFrance, London and Warsaw, plus Montreal, Singapore and Sydney
On your own serversNoClever Cloud On Premises, on request
Compliance and transparencySOC 2 is in progress. We're open about where we stand: live status and data retention policy.No SOC 2 report published
Every feature Battleships tracks (43)
FeatureboatClever Cloud
Full VM (own kernel)YesYes
Docker insideYesYes
Nested virtualizationYesNo
RootYesNot published
SystemdYesNot published
Desktop GUIYesNo
BrowserYesNo
Computer-use APIYesNo
SSHYesYes
Persistent diskYesNo
Pause/resumeYesYes
Automatic snapshotsYesNo
Fork/cloneYesNo
Live fork (no pause)YesNo
Memory forkNoNo
Custom image (Docker or snapshot)YesYes
≥24 h sessionsYesYes
Idle auto-stopNoNo
Wake on requestNoNo
HTTPS preview URLsYesYes
Raw TCP inboundYesYes
Public IPv4NoNo
Egress allowlistNoNo
Secret proxyYesNo
Secret proxy for any APIYesNo
Preinstalled agentsYesNo
Agent harness APIYesNo
MCP serverNoYes
GPUNoNo
WindowsNoNo
Arm64NoNo
Self-hostingNoYes
BYOCNoNo
Open sourceNoNo
Scoped API keysYesNot published
Spend limitsYesNot published
Usage per sandbox (API)YesNo
Audit logsNoNot published
SSONoNo
SOC 2NoNo
HIPAANoNo
EU data residencyYesYes
Zero data retentionYesNot published

Performance

boat26.69 runs/s
Clever CloudNot measured

There is no comparable measurement for Clever Cloud: it was not part of the public StarSling benchmark run. boat did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB on dedicated AMD Ryzen 9 9950X (Zen 5) hosts. Clever Cloud notes that its smallest nano and pico instances run with reduced CPU priority.

Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Clever 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.

Scenarioboat, a monthClever Cloud, 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
$20 · $20/mo$32 · Pay as you go · with limits
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$26 · $20/mo$134 · Pay as you go · with limits
50 agents in parallel
50 machines of 4 vCPU / 8 GB at once, 8-hour sessions, 1,100 sessions a month, Docker inside
$317 · $20/mo × 3 seats$1,616 · Pay as you go · with limits
Sandboxes for your users
2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once
$200 · $100/mo × 2 seats$459 · Pay as you go · with limits
1,000 short jobs at once
2 vCPU / 4 GB, 1,000 at once, 3 minutes each, 200,000 jobs a month
$2,000 · $2000/mo · with limits$918 · Pay as you go · with limits

boat: You pick a monthly plan from $20, and all of it becomes usage credit spent per second on running sandboxes. A stopped sandbox costs nothing and keeps its disk.

Clever Cloud: Clever Cloud bills each instance by the second of running, at a fixed price per size from nano to 3XL; the sizes priced here are L (6 vCPU / 8 GiB) and M (4 vCPU / 4 GiB), because no size matches 4 vCPU / 8 GiB. Disk and egress prices are not published per instance, and Premium adds a 490 euro monthly licence with consumption billed at 1.8 times and a 12-month commitment.

Limits in those scenarios

None of the Clever Cloud scenarios fully fit, mainly because it is an app platform: an instance has no disk that survives a restart, so the coding-agent scenarios lose their work. For 300 user sandboxes and 1,000 short jobs at once, Clever Cloud does not publish how many instances can run at the same time, and the typical published limit is 50. On boat the 1,000-job burst is limited by starts: the $2,000 plan allows 90 starts a minute and 2,400 a day, so a big burst gets queued.

ScenarioboatClever Cloud
One coding agent, workdaysNone that bite
  • missing persistent disk
  • different kind of product: app platform
  • disk beyond 0 GiB: price unknown
  • egress price not published
One agent that never stopsNone that bite
  • missing persistent disk
  • different kind of product: app platform
  • disk beyond 0 GiB: price unknown
  • egress price not published
50 agents in parallelNone that bite
  • missing persistent disk
  • different kind of product: app platform
  • disk beyond 0 GiB: price unknown
  • egress price not published
Sandboxes for your usersNone that bite
  • different kind of product: app platform
  • not confirmed: how many machines can run at once (you need 300; the typical published limit is 50)
  • disk beyond 0 GiB: price unknown
  • egress price not published
1,000 short jobs at once
  • 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)
  • different kind of product: app platform
  • not confirmed: how many machines can run at once (you need 1,000; the typical published limit is 50)
  • disk beyond 0 GiB: price unknown
  • egress price not published

Which one should you use?

Use Clever Cloud when

you need to host a finished web app in production with managed databases and autoscaling, need regions outside the EU, or must run the platform on your own servers.

  • Hosting the app your agent built. Useful once the app is done: Clever Cloud runs it in production with autoscaling, managed databases and custom domains.
  • Regions outside the EU and your own servers. Useful if you serve users in Canada, Singapore or Australia, or must run the platform in your own data center.
  • Short jobs with no plan fee. Useful for bursts of short scripts: Clever Tasks start a VM, run one command, stop, and bill by the second.

Use boat when

agents need a real machine to work in: root, Docker, a disk that keeps their work, forks, a desktop and coding agents you drive from outside.

  • A machine the agent can change. Coding agents install tools, edit system files and start services as they go. On Clever Cloud each of those must be baked into the image and redeployed first, which turns one agent step into a build.
  • Files that stay, and forks. An agent that restarts loses its checkout, its installs and its half-done work on Clever Cloud. The workaround is a database or object storage add-on, which the agent's tools do not write to by default.
  • Agents, desktop and keys built in. Clever Cloud says its AI tooling is not an AI-specific product, and it injects secrets as environment variables. You would build the agent runner, the browser and the key handling yourself.

Questions people ask

Can I use Clever Cloud as a sandbox for AI agents?

Partly. You can deploy a Docker image and SSH in, but instances are read-only, you log in as a non-root user and changes are lost when the instance restarts. Clever Cloud is built to host apps, so an agent that installs things and keeps files fits a sandbox like boat better.

Is boat cheaper than Clever Cloud?

In most of the scenarios we priced, yes: one coding agent on workdays comes to $20 a month on boat and $32 on Clever Cloud, and one agent running all month comes to $26 versus $134. 50 agents in parallel come to $317 versus $1,616. For 1,000 short jobs at once, Clever Cloud comes to $918 against $2,000 on boat, and neither fits as priced.

Do Clever Cloud apps run in VMs or containers?

In VMs. Clever Cloud runs each application instance in its own virtual machine, and Docker images run as a container inside that VM. boat also gives every sandbox its own VM, with root and a disk that persists.

Does data stay in the EU on both?

boat runs only in the EU (Germany, Finland and France), with EU data residency and a zero data retention option. Clever Cloud has European zones in France, Warsaw and London, and also runs in Montreal, Singapore and Sydney, so you pick the zone when you create an app.

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. Clever Cloud lists ISO 27001 and HDS on its security page.

Start a sandbox

Install the CLI, then run boat new.

7-day free trial

More comparisons