boat vs Sakura Cloud
Sakura Cloud is a Japanese cloud that rents plain virtual servers you set up yourself. boat gives each agent its own VM with Docker, /dev/kvm, a desktop and coding agents already inside, and it costs less in most of the scenarios below.
Where boat stands out
Coding agents you drive from outside
boat ships Claude Code, Codex, OpenCode and other agents inside each sandbox, and you drive them from outside with a prompt, steer and events API. Its credentials proxy keeps real AI keys out of the sandbox and can switch to another account when one runs out of credits. A Sakura Cloud server starts as a bare OS.
A desktop and nested virtualization ready
Every boat machine has root, systemd, Docker and /dev/kvm, plus a 60fps desktop with Chromium and computer use you can embed in your app. Sakura Cloud gives you root on a plain VM: Docker you install yourself, no desktop tools, and no documented /dev/kvm.
Snapshots every minute, fork while running
boat snapshots each sandbox's disk every minute and on stop, a stopped sandbox costs nothing, and you can fork a running sandbox into new ones. On Sakura Cloud you make disk images by hand, the manual advises stopping the server first, and the disk keeps billing while it is off.
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.
- Isolation
- Full VM per sandbox, own kernel
- Regions
- EU: Germany, Finland, France
Sakura Cloud
Sakura Cloud is the general-purpose cloud of Sakura Internet, a Japanese hosting company: you rent virtual servers, disks, networks and GPUs by the hour, day or month.
It is a classic cloud (IaaS), not an agent sandbox. You pick a server size from 1 vCPU up to 192 vCPU and 1,024 GB of memory, attach disks, choose Linux or Windows Server, and install everything yourself through the control panel, an API, a CLI (usacloud) or Terraform. It is built for Japanese companies and public bodies that want servers, GPUs and data kept in Japan.
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 workload | boat: price rank | Sakura Cloud: price rank |
|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Both give a full VM, but on Sakura Cloud you install Docker, the agents and your own backups, and nested virtualization for emulators is not documented. | 3rd cheapest of 367 $634 a month | 285th of 367, with limits $5,154 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB A Sakura Cloud server can run for months with no session cap, but it has no agent, desktop or key proxy built in, so each assistant needs its own setup. | 5th cheapest of 366 $657 a month | 180th of 366, with limits $1,562 a month |
| Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month Neither fits short bursts cleanly: Sakura Cloud bills a server by the hour, and boat's start limits queue big bursts. | 186th of 345, with limits $2,000 a month | 180th of 345, with limits $1,549 a month |
| Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month Sakura Cloud's default quota is 20 servers per zone for personal accounts and 100 for companies, with 300 GB of memory per zone, so hundreds of user sandboxes need a raise by ticket. | Cheapest of 361 $209 a month | 172nd of 361, with limits $4,096 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 | 190th of 360, with limits $29,828 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 | 86th cheapest of 364 $4,090 a month |
| CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month | 4th cheapest of 377 $86 a month | 134th cheapest of 377 $3,345 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB Both fit: a Sakura Cloud server runs all month at a fixed monthly price in Japan, and a boat machine keeps snapshots and costs nothing while stopped. | 18th cheapest of 368 $26 a month | 50th cheapest of 368 $65 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Sakura Cloud has no desktop or computer-use tools, so you would install a desktop and remote access yourself; boat streams a 60fps desktop with Chromium. | 6th cheapest of 369 $200 a month | 326th of 369, with limits $2,799 a month |
| Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month On Sakura Cloud you install and run the browser yourself on a plain server; boat ships Chromium and browser control in every machine. | 7th cheapest of 367 $200 a month | 327th of 367, with limits $1,430 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 | 255th of 346, with limits $322 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 | 179th of 338, with limits $6,442 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 | 299th of 360, with limits $698 a month |
| GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month Only Sakura Cloud fits: it rents H100 GPUs by the hour, and boat has no GPUs. | Can't run it | 85th of 98, with limits $2,876 a month |
| macOS agents 5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month | Can't run it | Can't run it |
Key features
| Feature | boat | Sakura Cloud |
|---|---|---|
| What it is | A sandbox cloud for AI agents: each sandbox is a ready Linux VM | A general cloud: you rent virtual servers and set them up yourself |
| Isolation | Every sandbox is a full VM with its own kernel | Every server is a full VM with its own kernel |
| Docker and nested virtualization | Docker, root, systemd and /dev/kvm ready on every machine | Root on the VM, so you can install Docker; nested virtualization (/dev/kvm) is not documented |
| Desktop and computer use | 60fps desktop and Chromium, computer use and browser control, embeddable stream | None built in; you install a desktop and remote access yourself |
| Coding agents | Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API | None; you install and run agents yourself |
| Keeping keys out of the sandbox | AI credentials proxy for model providers and any HTTPS API you register; real keys never enter the sandbox | Secrets Manager stores secrets that the server reads through an API, so the key ends up on the machine |
| Files after a stop | Snapshots every minute and on stop; a stopped sandbox costs nothing | The disk stays and is billed on its own; disk images (archives) are made by hand, best with the server stopped |
| Fork | Fork a running sandbox into new ones | Clone a server or make one from a disk image; copying a running disk can be inconsistent |
| Reaching the sandbox | HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4 | A public IPv4 per server and any port you open; you run your own HTTPS |
| Outbound network | Open internet on every plan; no provider-side allowlist | Outbound capped at 100 Mbps per server; IP and port rules through a paid VPN router appliance |
| Biggest machine | 16 vCPU / 32 GB | Up to 192 vCPU and 1,024 GB, plus plans with dedicated physical cores |
| GPUs | None | NVIDIA H100 by the hour (V100 until 2027) |
| Windows | No Windows guests; Wine for Windows apps | Windows Server with Remote Desktop licences |
| Regions | EU only: Germany, Finland, France | Japan only: Ishikari and Tokyo |
| Compliance and transparency | SOC 2 is in progress. We're open about where we stand: live status and data retention policy. | SOC 2 |
Every feature Battleships tracks (43)
| Feature | boat | Sakura Cloud |
|---|---|---|
| Full VM (own kernel) | Yes | Yes |
| Docker inside | Yes | Yes |
| Nested virtualization | Yes | No |
| Root | Yes | Not published |
| Systemd | Yes | Not published |
| Desktop GUI | Yes | No |
| Browser | Yes | No |
| Computer-use API | Yes | No |
| SSH | Yes | Yes |
| Persistent disk | Yes | Yes |
| Pause/resume | Yes | Yes |
| Automatic snapshots | Yes | No |
| Fork/clone | Yes | Yes |
| Live fork (no pause) | Yes | No |
| Memory fork | No | No |
| Custom image (Docker or snapshot) | Yes | Yes |
| ≥24 h sessions | Yes | Yes |
| Idle auto-stop | No | No |
| Wake on request | No | No |
| HTTPS preview URLs | Yes | Yes |
| Raw TCP inbound | Yes | Not published |
| Public IPv4 | No | Yes |
| Egress allowlist | No | Yes |
| Secret proxy | Yes | No |
| Secret proxy for any API | Yes | No |
| Preinstalled agents | Yes | No |
| Agent harness API | Yes | No |
| MCP server | No | No |
| GPU | No | Yes |
| Windows | No | Yes |
| Arm64 | No | No |
| Self-hosting | No | No |
| BYOC | No | No |
| Open source | No | No |
| Scoped API keys | Yes | Not published |
| Spend limits | Yes | Not published |
| Usage per sandbox (API) | Yes | No |
| Audit logs | No | Not published |
| SSO | No | Yes |
| SOC 2 | No | Yes |
| HIPAA | No | No |
| EU data residency | Yes | No |
| Zero data retention | Yes | Not published |
Performance
No comparable measurement. Sakura Cloud was not in StarSling's public benchmark run of 2026-09-29, so there is no CPU number to put next to boat's 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB on dedicated AMD Ryzen 9 9950X (Zen 5) hosts.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Sakura 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.
| Scenario | boat, a month | Sakura 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 | $58 · Pay as you go |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $26 · $20/mo | $65 · Pay as you go |
| 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 | $2,569 · Pay as you go |
| 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 | $4,090 · Pay as you go |
| 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 | $27,267 · Pay as you go |
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.
Sakura Cloud: Sakura Cloud bills each server by the hour, caps a day at its daily price, and fixes the monthly price once the server has run 20 days in a month. Disks are billed on top, there are no data transfer fees, and listed prices include Japan's 10% consumption tax (our table uses the price before tax).
Limits in those scenarios
On Sakura Cloud every scenario has the same limit: outbound traffic from a server is capped at 100 Mbps, and servers that send far more than about 1.6 TB a month may be slowed. Default quotas are 20 servers per zone for personal accounts and 100 for companies, with 300 GB of memory per zone, so 50 agents of 8 GB (400 GB) already need a second zone or a raise by ticket, and 300 user sandboxes or 1,000 jobs need a much bigger one. On boat the limit is a sandbox count per plan plus starts per minute, so 1,000 short jobs at once get queued and need more starts per day than the $2,000 plan allows.
| Scenario | boat | Sakura Cloud |
|---|---|---|
| One coding agent, workdays | None that bite |
|
| One agent that never stops | None that bite |
|
| 50 agents in parallel | None that bite |
|
| Sandboxes for your users | None that bite |
|
| 1,000 short jobs at once |
|
|
Which one should you use?
Use Sakura Cloud when
you need GPUs, Windows Server, very large machines, or servers and data kept in Japan.
- GPUs by the hour. Useful if your agents train or test models: Sakura Cloud rents NVIDIA H100 GPUs by the hour in Japan.
- Windows Server and very big machines. Useful if you need Windows Server or one machine with up to 192 vCPU and 1,024 GB of memory.
- Servers and data in Japan. Useful if your data or users must stay in Japan: Sakura Cloud runs in Ishikari and Tokyo with no data transfer fees.
Use boat when
you run AI agents: full VMs with Docker, /dev/kvm, a desktop and coding agents built in, keys kept out, and snapshots and forks without setup.
- Coding agents you drive from outside. On Sakura Cloud you write the code that installs, starts, prompts and watches every agent, and its Secrets Manager hands the real key to the server, where code the agent runs can read it.
- A desktop and nested virtualization ready. Computer-use and browser agents need a screen to look at, and Android emulators or microVMs need /dev/kvm. On Sakura Cloud you build and maintain the desktop stack yourself, and emulators may not run at all.
- Snapshots every minute, fork while running. Agents break things. Without automatic snapshots you lose the last good state, and without a live fork you stop the agent before you can copy its machine to try several paths.
Questions people ask
Is boat cheaper than Sakura Cloud?
In most of the scenarios we priced, yes: one coding agent on workdays comes to $20 a month on boat and $58 on Sakura Cloud, and one agent running all month comes to $26 versus $65. 50 agents in parallel come to $317 versus $2,569, and 300 sandboxes for your users to $200 versus $4,090. For 1,000 short jobs at once, Sakura Cloud comes to $27,267 and boat's $2,000 plan does not fit its start limits.
Is Sakura Cloud an AI sandbox?
No. Sakura Cloud is a general cloud that rents virtual servers, disks and GPUs, much like a smaller hyperscaler. You can run agents on it, but you install and manage everything yourself; boat is built for agents and ships the desktop, coding agents and snapshots ready.
Can I run Docker and emulators on Sakura Cloud?
You get root on a full VM, so you can install Docker yourself. Sakura Cloud does not document nested virtualization (/dev/kvm), which Android emulators and microVMs need. On boat, Docker, systemd and /dev/kvm are ready on every machine.
Does boat have GPUs or Windows like Sakura Cloud?
No. boat has no GPUs and no Windows guests; it runs Linux VMs on dedicated AMD Ryzen 9 9950X hosts and has Wine docs for Windows apps. If you need H100 GPUs or Windows Server, Sakura Cloud has both.
Where does my data run?
boat runs only in the EU (Germany, Finland and France), with EU data residency and a zero data retention option. Sakura Cloud runs only in Japan, in its Ishikari and Tokyo zones.
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. Sakura Internet says a SOC 2 Type 1 report covers Sakura Cloud, and the service is listed under Japan's ISMAP scheme.
Start a sandbox
Install the CLI, then run boat new.
7-day free trial