boat vs Kedge

Kedge is a global app platform that also runs short sandboxes and coding agents in persistent workspaces. boat is built for the agent itself: a full VM with root, Docker, nested virtualization and a desktop, which your code drives through an API.

Where boat stands out

A whole machine the agent can own

Every boat sandbox is a full VM with root, systemd, Docker and /dev/kvm, so agents can run compose stacks, Android emulators and VM images. Kedge documents Docker and sudo on its CI runners, but its workspace docs do not, and no Kedge product documents nested virtualization.

Agents you drive from outside

boat ships seven coding agents and a prompt, steer and events API, so your code can start agents inside many sandboxes, redirect them and follow every step. Real API keys stay outside the sandbox in a credentials proxy that works with any HTTPS API.

A desktop and a browser built in

boat streams a 60fps Linux desktop with Chromium, and agents can drive it with computer use while you watch or embed it in your app. Kedge has no desktop or managed browser.

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.

Kedge

Kedge is a lightweight, globally distributed cloud platform for apps, with disposable sandboxes and coding-agent workspaces built in.

Kedge is mainly an app platform, closer to a global Heroku than to a sandbox API. You deploy sites, handlers, container services and databases, each app gets a replicated SQLite database and file tree, and services scale to zero by restoring VM snapshots taken after startup. It also offers one-shot and held sandboxes, GitHub Actions runners, and persistent workspaces where its own agent, Claude Code or Codex builds and deploys apps for you.

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 rankKedge: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
A software factory needs Docker, a browser for tests and long builds on one machine; boat documents all of these inside the sandbox, while Kedge documents Docker only on its CI runners.
3rd cheapest of 367
$634 a month
45th of 367, with limits
$275 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
Kedge workspaces sleep after ten quiet minutes by default and wake on the next request; boat keeps the machine running and lets your code prompt and steer the agent by API.
5th cheapest of 366
$657 a month
101st of 366, with limits
$441 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
Kedge's disposable sandboxes come from warm pools and bill only the CPU time they use, which suits short code runs; a held one lasts up to an hour.
186th of 345, with limits
$2,000 a month
57th of 345, with limits
$45 a month
Sandboxes for your users
300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month
Kedge can give each user an app with its own HTTPS URL that scales to zero; boat gives each user a full VM with hosting, a desktop and agents built in.
Cheapest of 361
$209 a month
32nd of 361, with limits
$154 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
49th of 360, with limits
$2,841 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
98th of 364, with limits
$213 a month
CI runners
40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month
Kedge runs GitHub Actions jobs in fresh 4 vCPU / 8 GiB VMs with Docker; boat adds nested virtualization for Android emulators and VM images.
4th cheapest of 377
$86 a month
149th of 377, with limits
$201 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
A Kedge machine that goes idle pauses and bills no CPU or memory, which suits a part-time dev box; boat gives root, systemd and /dev/kvm on a disk saved every minute.
18th cheapest of 368
$26 a month
116th of 368, with limits
$7.64 a month
Computer-use agents
50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month
Kedge has no desktop to drive; boat streams a 60fps Linux desktop with Chromium and computer use.
6th cheapest of 369
$200 a month
72nd of 369, with limits
$51 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
Kedge does not offer a browser for automation; boat ships Chromium on a desktop you can watch and embed.
7th cheapest of 367
$200 a month
153rd of 367, with limits
$41 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
171st of 346, with limits
$83 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
67th of 338, with limits
$1,702 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
134th of 360, with limits
$90 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

FeatureboatKedge
Kind of productAgent sandbox: a full Linux VM for each agent or userApp platform with sandboxes, CI runners and coding-agent workspaces
IsolationFull VM with its own kernel on dedicated AMD Ryzen 9 9950X hostsHardware-isolated VM with its own Linux kernel
Root, Docker and nested virtualizationRoot, systemd, Docker and /dev/kvm in every sandboxDocker and passwordless sudo documented on CI runners; no nested virtualization documented
Desktop and browser60fps Linux desktop with Chromium and computer use, embeddable in your appNone; the workspace UI is an agent session in a browser tab
Coding agentsSeven agents preinstalled, driven from outside with a prompt, steer and events APIBuilt-in app agent, Claude Code and Codex, started from the CLI, SSH or the web
Keeping API keys out of the sandboxCredentials proxy for model providers and any HTTPS API; switch a credential to another account when credits run outSecrets as environment variables or files mounted under /run/secrets inside the VM
Session lengthNo cap on paid plans; you set the time to liveWorkspaces and machines have no stated cap; held sandboxes last up to 1 hour, 10 minutes per command
Idle and wakeRuns until you stop it or its time to live ends; started again by APIWorkspaces sleep after ten quiet minutes and wake on the next request; idle machines bill no CPU or memory
Persistent diskDisk saved every minute and on stop; a stopped sandbox costs nothingWorkspace volume keeps the whole filesystem; storage is billed while it sleeps
ForkFork a running sandbox with its diskCopy an app and its current data with kedge fork; no fork of a running workspace or sandbox
Inbound accessHTTPS URLs for up to 50 ports, token-protected by default; raw TCPHTTPS at name.kedge.run or your own domain, with CDN and app login; no static addresses
App hosting extrasHost what the sandbox runs on its portsScale to zero, replicated SQLite in every app, domain registration, pull-request previews
RegionsEU only (Germany, Finland, France)Ten compute regions on six continents, including Germany and Finland
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)
FeatureboatKedge
Full VM (own kernel)YesYes
Docker insideYesNot published
Nested virtualizationYesNo
RootYesNot published
SystemdYesNot published
Desktop GUIYesNo
BrowserYesNo
Computer-use APIYesNo
SSHYesYes
Persistent diskYesYes
Pause/resumeYesYes
Automatic snapshotsYesYes
Fork/cloneYesYes
Live fork (no pause)YesNo
Memory forkNoNo
Custom image (Docker or snapshot)YesYes
≥24 h sessionsYesYes
Idle auto-stopNoYes
Wake on requestNoYes
HTTPS preview URLsYesYes
Raw TCP inboundYesNo
Public IPv4NoNo
Egress allowlistNoNo
Secret proxyYesNo
Secret proxy for any APIYesNo
Preinstalled agentsYesYes
Agent harness APIYesYes
MCP serverNoNo
GPUNoNo
WindowsNoNo
Arm64NoNo
Self-hostingNoNo
BYOCNoNo
Open sourceNoNo
Scoped API keysYesYes
Spend limitsYesYes
Usage per sandbox (API)YesNo
Audit logsNoNot published
SSONoNo
SOC 2NoNo
HIPAANoNo
EU data residencyYesYes
Zero data retentionYesNot published

Performance

boat26.69 runs/s
KedgeNot measured

There is no comparable measurement: Kedge has not been run on the hpc-sandbox-benchmarks CPU test, so this page does not compare CPU speed. boat did 26.69 runs per second at 4 vCPU / 8 GB on dedicated AMD Ryzen 9 9950X hosts. Kedge publishes start times instead: 0.6 ms for a fresh sandbox from its warm pool, measured on its own hosts.

Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Kedge 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 monthKedge, 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$6 · Public preview · with limits
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$26 · $20/mo$34 · Public preview · 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$480 · Public preview · 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$124 · Public preview · 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$510 · Public preview · 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.

Kedge: Kedge is in public preview, and its rates are estimates that may change: $15 per vCPU-month of CPU time actually used, $5 per GiB-month of recently touched memory, $0.05 per GiB-month of stored data and $0.01 per GiB of egress. Every account gets $5 of usage a month.

Limits in those scenarios

Every Kedge scenario is marked as not fitting because its prices are preview estimates, not a paid service you can sign up for today. Kedge also does not publish how many machines an account can run at once, and past its cap a new sandbox fails with an error instead of waiting in a queue, so 300 user sandboxes or 1,000 jobs at once are not confirmed. boat's plans fit every scenario except 1,000 short jobs at once, where even the $2,000 plan queues starts above 90 a minute.

ScenarioboatKedge
One coding agent, workdaysNone that bite
  • preview pricing published as estimates: final price may differ
One agent that never stopsNone that bite
  • preview pricing published as estimates: final price may differ
50 agents in parallelNone that bite
  • not confirmed: Docker inside
  • preview pricing published as estimates: final price may differ
  • how many machines can run at once: limit not published, probably fine at 50 (the typical published limit is 50)
Sandboxes for your usersNone that bite
  • not confirmed: how many machines can run at once (you need 300; the typical published limit is 50)
  • preview pricing published as estimates: final price may differ
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)
  • not confirmed: how many machines can run at once (you need 1,000; the typical published limit is 50)
  • preview pricing published as estimates: final price may differ

Which one should you use?

Use Kedge when

you want agents to build and ship web apps onto a global host that scales to zero, or need compute outside Europe.

  • Regions around the world. Useful when users are far from Europe: it runs in ten compute regions across six continents.
  • Scale to zero and wake on request. Good for agents and apps that sit idle most of the day: an idle machine bills no CPU or memory.
  • A full app platform. Fits teams that want agents to ship apps straight to production hosting with a database and domain.

Use boat when

the agent needs a real machine to work in: root, Docker, emulators and a browser, driven from your own code with keys kept out of reach.

  • A whole machine the agent can own. Agent tasks often end in docker compose up or an emulator boot. If the machine cannot do that, the agent stops and a person has to finish the job somewhere else.
  • Agents you drive from outside. Kedge starts its agents from a terminal, SSH or a web page, and secrets are mounted inside the VM. Running agents from your own backend means scripting the CLI, and one prompt injection can read the keys.
  • A desktop and a browser built in. Agents that test a web app, fill a form or use a site without an API need a screen. On Kedge you would have to bring and host that stack yourself.

Questions people ask

Is boat cheaper than Kedge?

It depends on the job. Kedge's estimates are lower for one workday agent, sandboxes for your users and 1,000 short jobs, because it bills only CPU time used. boat is lower for an agent that never stops and for 50 agents in parallel ($317 against $480 a month), and Kedge's rates are still preview estimates.

Is Kedge a sandbox or a hosting platform?

Mainly a hosting platform. Kedge deploys apps, services and databases worldwide, and adds disposable sandboxes, CI runners and coding-agent workspaces on the same VMs. boat is a sandbox first: a full machine for each agent.

Can I run Docker and Android emulators in each?

Kedge documents Docker and passwordless sudo on its GitHub Actions runners, and its workspace docs do not say either way. Neither nested virtualization nor /dev/kvm appears in Kedge's docs. boat gives root, Docker and /dev/kvm in every sandbox.

Can a Kedge agent use a browser or desktop?

Kedge does not offer one. Its workspace UI is a chat and terminal session in your browser. boat has a 60fps Linux desktop with Chromium that agents drive with computer use.

Is Kedge faster than boat?

There is no comparable CPU measurement for Kedge. boat did 26.69 runs per second on the hpc-sandbox-benchmarks test at 4 vCPU / 8 GB. Kedge reports fresh sandboxes ready in 0.6 ms from a warm pool, which measures start time, not CPU speed.

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