boat vs Prized
Prized is a cloud dev box built to feel like your laptop, with ports and files mirrored to your machine. boat is a VM for agent work: nested virtualization, AI keys that stay outside the machine, and a desktop agents can drive.
Where boat stands out
/dev/kvm inside every machine
Both products give you a real VM with root and Docker. boat also exposes /dev/kvm, so an agent can run Android emulators, Firecracker or a nested VM inside its sandbox. Prized's docs say nested virtualization is off and KVM is not available inside a box.
Real keys never enter the sandbox
boat's AI credentials proxy runs outside the sandbox and adds the real key on the way out, for model providers and any HTTPS API you register. Prized's broker runs on the box itself as root, and its security page says root can read anything on the box. boat can also switch a credential to another account when one runs out of credits.
A desktop agents can drive
boat streams a 60fps desktop with Chromium and has a computer-use API, so an agent can see the screen and click, and you can embed the stream in your app. Prized streams an XFCE desktop over VNC through a private tunnel and says it is not for full-motion video. It has no computer-use API; you script xdotool on the box.
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
Prized
Prized is a cloud dev box that feels like localhost: a dedicated Linux VM that syncs with your laptop and keeps your agents running after you close the lid.
Each Prized box is an AWS EC2 virtual machine with its own kernel, running Ubuntu 24.04, in seven sizes from Nano (2 vCPU / 2 GB) to Extra Large (32 vCPU / 128 GB). Ports on the box open on your laptop at the same number, files sync both ways, and Claude Code and Codex come preinstalled. It is built for developers and small teams who want a persistent remote machine; a newer beta adds short-lived Firecracker sandboxes your code creates through an API.
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 | Prized: price rank |
|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Both fit with a full VM, Docker and root; boat adds /dev/kvm for emulators and nested VMs, which Prized turns off. | 3rd cheapest of 367 $634 a month | 7th cheapest of 367 $1,652 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB | 5th cheapest of 366 $657 a month | 7th cheapest of 366 $1,075 a month |
| Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month Neither is a perfect fit for one-minute runs; Prized's new sandbox beta targets this kind of short, throwaway job. | 186th of 345, with limits $2,000 a month | 112th of 345, with limits $213 a month |
| Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month A paid Prized workspace holds at most 100 boxes, counting paused ones, so 300 sandboxes for your users do not fit there. | Cheapest of 361 $209 a month | 127th of 361, with limits $1,338 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 | 103rd of 360, with limits $6,464 a month |
| Agent evals 250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month Many parallel eval runs hit Prized's 100-box cap and minutes-long forks, while boat forks a running sandbox for each run. | 10th cheapest of 364 $200 a month | 155th of 364, with limits $1,040 a month |
| CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month Prized boxes are dev machines whose forks take minutes, so short CI jobs are a compromise; boat starts fresh VMs for each job. | 4th cheapest of 377 $86 a month | 167th of 377, with limits $624 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB Both fit; Prized is built for this, with ports and files mirrored to your laptop, and boat keeps the disk with snapshots every minute. | 18th cheapest of 368 $26 a month | 64th cheapest of 368 $86 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Prized streams a VNC desktop and leaves automation to xdotool scripts on the box, while boat has a 60fps desktop with a computer-use API. | 6th cheapest of 369 $200 a month | 63rd of 369, with limits $493 a month |
| Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month | 7th cheapest of 367 $200 a month | 123rd of 367, with limits $123 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 | 94th of 346, with limits $317 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 | 51st of 338, with limits $1,336 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 | 16th of 360, with limits $81 a month |
| GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month | Can't run it | Can't run it |
| 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 | Prized |
|---|---|---|
| Isolation | Every sandbox is a full VM with its own kernel | Every box is an EC2 VM with its own kernel; sandboxes are Firecracker microVMs |
| Hardware | Dedicated AMD Ryzen 9 9950X (Zen 5) hosts | AWS t3a (burstable) for Nano and Micro, m6a for Extra Small and up |
| Docker and nested virtualization | Docker, root, systemd and /dev/kvm on every machine | Docker, root and systemd; nested virtualization off, no KVM inside a box |
| Biggest machine | Up to 16 vCPU / 32 GB | Up to 32 vCPU / 128 GB (Extra Large) |
| Desktop and computer use | 60fps desktop and Chromium, computer use and browser control, embeddable stream | XFCE desktop and Chrome over VNC through a private tunnel; no computer-use API, scripts use xdotool |
| Pause and memory | Stop keeps the disk; resume reboots from disk | Nano to Small hibernate with memory kept; Medium and up restart processes on resume |
| Snapshots and fork | Snapshots every minute and on stop; fork a running sandbox | Scheduled disk snapshots every 1 to 168 hours; fork a running box in minutes |
| Files after a stop | A stopped sandbox costs nothing and keeps its disk | A paused box pays for its whole disk at $0.0001 per GB hour |
| Reaching the machine | HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4 | Ports mirrored to your laptop; share links for HTTP and websockets, up to 10 public or anyone links per workspace; public IP changes on pause |
| Keeping keys out of the machine | AI credentials proxy outside the sandbox for model providers and any HTTPS API; switch a credential to another account when credits run out | Broker on the box swaps placeholders for allowlisted HTTPS hosts; it runs as root, and root can read anything on the box |
| Coding agents | Claude Code, Codex, OpenCode, pi and more preinstalled, with a prompt, steer and events API | Claude Code and Codex preinstalled; prompt, events and interrupt API for Claude Code, Codex, OpenCode and pi |
| Laptop workflow | SSH, web terminal and hosted URLs | Ports open on your laptop at the same number; two-way file sync including .git |
| Regions | EU only: Germany, Finland, France, with EU data residency | Four US regions and Frankfurt; account data and control plane in Oregon |
| Compliance and transparency | SOC 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)
| Feature | boat | Prized |
|---|---|---|
| Full VM (own kernel) | Yes | Yes |
| Docker inside | Yes | Yes |
| Nested virtualization | Yes | No |
| Root | Yes | Yes |
| Systemd | Yes | Yes |
| Desktop GUI | Yes | Yes |
| Browser | Yes | Yes |
| Computer-use API | Yes | No |
| SSH | Yes | Yes |
| Persistent disk | Yes | Yes |
| Pause/resume | Yes | Yes |
| Automatic snapshots | Yes | Yes |
| Fork/clone | Yes | Yes |
| Live fork (no pause) | Yes | Yes |
| Memory fork | No | No |
| Custom image (Docker or snapshot) | Yes | Yes |
| ≥24 h sessions | Yes | Yes |
| Idle auto-stop | No | Yes |
| Wake on request | No | Yes |
| HTTPS preview URLs | Yes | Yes |
| Raw TCP inbound | Yes | No |
| Public IPv4 | No | Yes |
| Egress allowlist | No | No |
| Secret proxy | Yes | Yes |
| Secret proxy for any API | Yes | Yes |
| Preinstalled agents | Yes | Yes |
| Agent harness API | Yes | Yes |
| MCP server | No | No |
| GPU | No | No |
| Windows | No | No |
| Arm64 | No | No |
| Self-hosting | No | No |
| BYOC | No | No |
| Open source | No | No |
| Scoped API keys | Yes | Not published |
| Spend limits | Yes | Yes |
| Usage per sandbox (API) | Yes | No |
| Audit logs | No | Yes |
| SSO | No | No |
| SOC 2 | No | No |
| HIPAA | No | No |
| EU data residency | Yes | No |
| Zero data retention | Yes | Not published |
Performance
There is no comparable measurement for Prized. In StarSling's public run of 2026-09-29, boat did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB, and Prized was not in that run. Prized runs on AWS t3a instances for Nano and Micro and m6a instances for larger sizes.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Prized 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 | Prized, 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 | $22 · Paid credit plan |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $26 · $20/mo | $90 · Paid credit plan |
| 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 | $1,085 · Paid credit plan |
| 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 | $154 · Paid credit plan · 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 | $341 · Paid credit plan · 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.
Prized: You pick a prepaid plan from $10 to $1,000 a month, all of it credit, and each awake box draws its size's monthly price spread over 730 hours, never more than that monthly price. A paused box still pays for its whole disk, and Prized publishes no price for outbound traffic. Its sizes pair 4 vCPU with 16 GB, so the table compares boat's 4 vCPU / 8 GB with Prized's Small.
Limits in those scenarios
A paid Prized workspace can hold 100 boxes, and paused and deep-sleeping boxes count, so 300 sandboxes for your users and 1,000 short jobs at once do not fit there. Creating, waking or resizing a box also needs enough balance to run the whole fleet for four hours. On boat the limit is a running sandbox count (100 to 2,000 by plan) plus starts per minute, so the 1,000-job burst gets queued and needs more starts per day than the $2,000 plan allows.
| Scenario | boat | Prized |
|---|---|---|
| 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 Prized when
you want a cloud dev box that acts like your laptop, with local ports, two-way file sync and a paused box that keeps its memory, or when your boxes must run in the US.
- Feels like your laptop. Useful for developers who code on a laptop: ports open locally at the same number and files sync both ways.
- Pause with memory kept. Useful if you want a paused box to come back with tmux, processes and half-done builds still running.
- US regions and bigger boxes. Useful if your boxes must run in the US, or one task needs up to 32 vCPU and 128 GB.
Use boat when
agents run on their own at scale: emulators and nested VMs with /dev/kvm, real keys kept outside the machine, a desktop they can drive, and hundreds of sandboxes forked from a running one.
- /dev/kvm inside every machine. Agents that test mobile apps, build VM images or run their own sandboxes need KVM. On Prized that work falls back to slow software emulation or cannot run at all.
- Real keys never enter the sandbox. Prized gives the box's user passwordless sudo, so code an agent runs with root can reach the broker's memory. On boat there is no real key inside the VM to take.
- A desktop agents can drive. Without a computer-use API you write the screenshot, click and type loop yourself on top of xdotool, and VNC frames make fast pages hard for a model to follow.
Questions people ask
Is boat cheaper than Prized?
In most of the scenarios we priced, yes: one agent running all month comes to $26 on boat and $90 on Prized, and 50 agents in parallel come to $317 versus $1,085. One agent on workdays is close, at $20 versus $22. Prized's 300 sandboxes for users would be $154 against boat's $200, but they do not fit Prized's 100-box limit.
Are Prized boxes VMs or containers?
VMs. Each Prized box is a dedicated AWS EC2 virtual machine with its own kernel, and its newer sandboxes are Firecracker microVMs. boat sandboxes are full VMs too, with /dev/kvm on top, which Prized turns off.
Can I host a public app on Prized?
Partly. Prized shares a port as a link for HTTP and websockets, which can be public on a paid plan, with up to 10 public or anyone links per workspace; those links expire after 7 days by default, and the box's own IP changes on each pause. boat gives HTTPS URLs on up to 50 ports per sandbox, plus raw TCP and a dedicated IP.
Does Prized have computer use?
Prized has a desktop and Chrome streamed over VNC, and an agent on the box can drive them with xdotool or Playwright scripts. It has no computer-use API of its own. boat streams a 60fps desktop and has a computer-use API agents call directly.
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. Prized runs boxes in four US regions and Frankfurt; a Frankfurt box keeps its disk and backups in Germany, but account data and the control plane stay in Oregon.
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. Prized's security page also says its SOC 2 is in progress.
Start a sandbox
Install the CLI, then run boat new.
7-day free trial