boat vs Coasty

Coasty sells a computer-use model and an agent loop that runs on its servers, with machines rented by the hour to run it on. boat sells the machine itself: a full VM with its own kernel, a desktop, and coding agents inside that you drive from outside.

Where boat stands out

A full VM you know the shape of

Every boat sandbox is a VM with its own kernel on dedicated AMD Ryzen 9 9950X hosts, with root, systemd, Docker and /dev/kvm for emulators. Sizes are published, and the CPU was measured at 26.69 runs per second at 4 vCPU / 8 GB. Coasty places a machine on a Daytona sandbox, an EC2 VM or an Azure container, and does not publish the default size.

Coding agents inside, driven from outside

Claude Code, Codex and other coding agents come preinstalled, and you prompt, steer and follow them through one API. Real API keys stay outside the sandbox in a credentials proxy, and a credential can switch to another account when it runs out of credits.

State that saves itself

boat saves the disk every minute and when the sandbox stops, a stopped sandbox costs nothing, and you can fork a running sandbox into copies. Coasty snapshots only when you call its API, restores only the latest snapshot on Linux, and has no fork.

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.

Coasty

Coasty is a computer-use AI company: it builds models, environments, data and evaluations for AI labs, and sells a computer-use API.

Coasty's API turns a screenshot and an instruction into clicks and key presses, and can run a whole task for you on a machine it rents out, with checks, human takeover and webhooks. Those machines are Linux or Windows desktops placed on AWS EC2, Daytona sandboxes or Azure containers, billed per minute from a prepaid wallet. It is built for teams that want an agent to operate apps through the screen, and for labs that train and grade such agents. It is not a general sandbox product, so this page compares the machines and 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 rankCoasty: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
A software agent needs Docker, systemd and build speed, which boat documents on every machine and Coasty does not publish for its machines.
3rd cheapest of 367
$634 a month
34th of 367, with limits
$874 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
An assistant that lives on one machine all month needs saved state and free stops; Coasty bills a stopped machine by the hour and snapshots only on request.
5th cheapest of 366
$657 a month
70th of 366, with limits
$1,819 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
186th of 345, with limits
$2,000 a month
231st of 345, with limits
$161 a month
Sandboxes for your users
300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month
boat gives each user a VM with HTTPS URLs on its ports, while Coasty machines are reached by IP, SSH and VNC.
Cheapest of 361
$209 a month
218th of 361, with limits
$594 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
243rd of 360, with limits
$2,494 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
238th of 364, with limits
$244 a month
CI runners
40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month
CI jobs need a known machine size and Docker; Coasty publishes neither for its default machine.
4th cheapest of 377
$86 a month
218th of 377, with limits
$54 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
18th cheapest of 368
$26 a month
143rd of 368, with limits
$31 a month
Computer-use agents
50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month
Both fit: Coasty brings its own computer-use model and agent loop, while boat gives your agent a full VM with a 60fps desktop it can drive.
6th cheapest of 369
$200 a month
8th cheapest of 369
$244 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
Both fit; Coasty has browser control and a managed residential proxy, and boat has Chromium with computer use on a VM you can also host from.
7th cheapest of 367
$200 a month
12th cheapest of 367
$244 a month
Scraping & bots (startup scale)
50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month
Both fit; Coasty can route a machine through its residential proxy, and boat runs a Chromium fork on a full VM with its own IP.
6th cheapest of 346
$90 a month
21st cheapest of 346
$244 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
160th of 338, with limits
$4,994 a month
Parallel red teaming
200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month
Parallel red teaming fits boat's forkable VMs with nested virtualization; on Coasty you boot fresh machines from your latest snapshot.
2nd cheapest of 360
$200 a month
51st of 360, with limits
$119 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

FeatureboatCoasty
Kind of productSandbox VMs for agents, with coding agents built inComputer-use model API and agent runs, with machines rented to run them
IsolationFull VM with its own kernel on dedicated AMD Ryzen 9 9950X hostsDaytona sandbox, AWS EC2 VM or Azure container, depending on the task
Machine sizeUp to 16 vCPU / 32 GB on the $100 plan and above, sizes published1 to 16 cores and 1 to 64 GB, capped by tier; default size not published
Operating systemsUbuntu 24.04 LinuxLinux or Windows; your own macOS machine can be enrolled
Desktop and computer use60fps desktop and Chromium, computer use and browser control, embeddable streamXFCE desktop over VNC, screenshots, action and browser APIs
AgentClaude Code, Codex and other coding agents preinstalled, with a prompt, steer and events APIIts own computer-use agent runs on Coasty's servers and drives the machine through the screen
Docker, systemd and rootYes, documented on every machineSSH and a terminal API as the ubuntu user; Docker and root not documented
Nested virtualizationYes, /dev/kvm for Android emulators and VMsNot documented
Saving stateDisk saved every minute and on stop; a stopped sandbox costs nothingSnapshot on request for $0.01 each, Linux only; a stopped machine bills $0.01 an hour
ForkFork a running sandbox with its diskNo fork; a new machine can boot from your latest snapshot
Session lengthNo cap on paid plans; you set the time to liveNo cap by default; optional auto-destroy from 5 minutes to 7 days
Inbound accessHTTPS URLs for up to 50 ports, token-protected by default; raw TCP; dedicated IPv4 or IPv6Public IP with SSH, VNC, WebSocket and browser debug ports; no managed HTTPS URLs
Outbound proxyOpen internet from the sandbox's own IPOptional managed residential proxy, or your own HTTP or SOCKS5 proxy, checked before use
Keeping API keys out of the sandboxCredentials proxy for model providers and any HTTPS API; switch a credential to another account when credits run outModel keys for its own agent go to Coasty in a header; no proxy for keys used inside the machine
Compliance and transparencySOC 2 is in progress. We're open about where we stand: live status and data retention policy.SOC 2
Every feature Battleships tracks (43)
FeatureboatCoasty
Full VM (own kernel)YesNot published
Docker insideYesNot published
Nested virtualizationYesNo
RootYesNot published
SystemdYesNot published
Desktop GUIYesYes
BrowserYesYes
Computer-use APIYesYes
SSHYesYes
Persistent diskYesYes
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 URLsYesNo
Raw TCP inboundYesNot published
Public IPv4NoYes
Egress allowlistNoNo
Secret proxyYesNo
Secret proxy for any APIYesNo
Preinstalled agentsYesNo
Agent harness APIYesNo
MCP serverNoYes
GPUNoNo
WindowsNoYes
Arm64NoNo
Self-hostingNoNo
BYOCNoYes
Open sourceNoNo
Scoped API keysYesYes
Spend limitsYesYes
Usage per sandbox (API)YesYes
Audit logsNoNot published
SSONoYes
SOC 2NoYes
HIPAANoYes
EU data residencyYesNo
Zero data retentionYesNot published

Performance

boat26.69 runs/s
CoastyNot measured

There is no comparable measurement. Coasty does not publish a default machine size or a CPU benchmark, and its machines were not in the hpc-sandbox-benchmarks runs. For reference, boat did 26.69 runs per second on that Node.js build-and-test loop at 4 vCPU / 8 GB.

Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Coasty 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 monthCoasty, 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$9 · API 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$31 · Developer API · 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$434 · Developer API · 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$244 · Developer API · 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$494 · Developer API · with limits

boat: Plans start at $20 a month, and the whole fee is usage credit. You pay per second while a sandbox runs, and a stopped sandbox costs nothing and keeps its disk.

Coasty: Coasty bills machines per minute from a prepaid wallet: $0.05 an hour running on Linux, $0.09 on Windows, $0.01 an hour stopped and $0.01 per snapshot, for a size it does not publish. Its agent bills per step on top, $0.05 per managed run step, and the Developer plan is $19 a month with $25 of API credit.

Limits in those scenarios

Coasty does not publish its default machine size, its concurrent-machine cap per plan, or prices for disk and egress, so Coasty's fit cannot be confirmed in any scenario, and its prices here are the runtime rate for a machine of unknown size. Agent steps, if you use Coasty's agent, are billed on top. boat fits four of the five; at 1,000 short jobs at once its $2,000 plan queues starts above 90 a minute and allows 600 starts an hour.

ScenarioboatCoasty
One coding agent, workdaysNone that bite
  • machine size not published
  • machine size per session not published
  • disk beyond 0 GiB: price unknown
  • egress price not published
One agent that never stopsNone that bite
  • machine size not published
  • machine size per session not published
  • disk beyond 0 GiB: price unknown
  • egress price not published
50 agents in parallelNone that bite
  • not confirmed: Docker inside
  • machine size not published
  • machine size per session not published
  • disk beyond 0 GiB: price unknown
Sandboxes for your usersNone that bite
  • machine size not published
  • not confirmed: how many machines can run at once (you need 300; the typical published limit is 50)
  • machine size per session not published
  • disk beyond 0 GiB: price unknown
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)
  • machine size not published
  • not confirmed: how many machines can run at once (you need 1,000; the typical published limit is 50)
  • machine size per session not published
  • disk beyond 0 GiB: price unknown

Which one should you use?

Use Coasty when

you want a ready computer-use model that operates apps through the screen for you, on Windows or on machines you already own.

  • Its own computer-use model and agent loop. For teams that want a ready computer-use agent, with checks, human takeover, workflows and schedules, rather than building their own.
  • Windows desktops and your own machines. Useful when an agent must work in Windows apps, or on a Windows or macOS machine you already run.
  • Managed residential proxy. Helpful for browser agents on sites that block datacenter IPs: one flag routes the machine through a residential upstream.

Use boat when

your agents need a real machine to work in for hours: Docker, builds, emulators, a desktop and a browser on a full VM, with coding agents and credentials built in.

  • A full VM you know the shape of. An agent that installs packages, runs Docker or starts an emulator needs to know what machine it gets. On Coasty the backend can change by task and the default size is not published, so builds may behave differently between runs.
  • Coding agents inside, driven from outside. Coasty's agent works through the screen and runs on Coasty's servers. To run a coding agent inside a Coasty machine you install it yourself and put its API key on a machine the agent controls.
  • State that saves itself. Agents that run for weeks lose work when a machine fails between manual snapshots. On Coasty you also pay by the hour for each stopped machine you keep.

Questions people ask

Is Coasty a sandbox?

Not mainly. Coasty sells a computer-use model and agent runs, and rents machines for them to run on. boat sells the machine: a full VM for any agent, with coding agents preinstalled.

Is boat cheaper than Coasty?

Coasty's hourly rate is lower on paper, and it comes out lower in the workday and burst scenarios. boat costs less in the always-on, 50-agent and user-sandbox scenarios. Coasty does not publish the size of the machine behind its rate, and its agent steps cost extra, so the numbers are not like for like.

Can I use Coasty's computer-use model with a boat sandbox?

Coasty's docs describe a predict API that takes a screenshot and returns actions for you to run, and a way to enroll a Linux machine you operate. Both paths would work with a desktop you run yourself; we have not tested them on boat.

Does Coasty run Windows?

Yes. Coasty rents Windows machines at a higher hourly rate than Linux, and can drive a Windows or macOS machine you enroll. boat runs Linux only.

Can I run Docker and Android emulators in each?

boat runs Docker on every sandbox and exposes /dev/kvm for Android emulators and VMs. Coasty gives SSH and a terminal API, but does not document Docker or nested virtualization, and some of its machines are Azure containers.

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.

Start a sandbox

Install the CLI, then run boat new.

7-day free trial

More comparisons