boat vs Hyperbeam
Hyperbeam streams a cloud Chromium into your web app so many people can watch and control it together. boat gives your agent a whole Linux machine with a browser and desktop inside, where it can also run code, Docker and servers and keep its files.
Where boat stands out
A whole machine around the browser
A boat sandbox is a full Linux VM with root, a shell, Docker, systemd and /dev/kvm, with Chromium and a desktop already inside. Your agent can browse a site, save what it found, run a script on it, build an app and serve it on HTTPS ports, all on one machine.
Its work is still there tomorrow
On boat the whole disk is saved every minute and when the sandbox stops, a stopped sandbox costs nothing, and you can fork a running one to try several paths at once. Your agent comes back to the same files, tools and browser profile.
Coding agents and keys built in
Claude Code, Codex and other coding agents come preinstalled, and you prompt, steer and follow them through one API. A credentials proxy keeps real API keys outside the sandbox, for model providers and any HTTPS API, and can switch a credential to another account when credits run out.
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
Hyperbeam
Hyperbeam sells virtual computers you embed in a web page: a cloud Chromium, and a few other session types, streamed to many participants at once.
Hyperbeam is a different kind of product from boat. Your backend starts a session with a REST call, and your page shows it through an embed URL and a JavaScript SDK, with synced audio and video, multi-user control and per-user roles. It is built for co-browsing and watch-together features in apps, and you can drive the browser from code with mouse, keyboard and navigation calls, but there is no shell or general machine underneath that you can use.
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 | Hyperbeam: price rank |
|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A software factory needs a shell, Docker and a place to build and test, and a Hyperbeam session offers a browser only. | 3rd cheapest of 367 $634 a month | 282nd of 367, with limits $4,994 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB An assistant that runs all month needs a machine that keeps its state; Hyperbeam keeps cookies, history and bookmarks between sessions, not files. | 5th cheapest of 366 $657 a month | 348th of 366, with limits $15,260 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 | 313th of 345, with limits $1,307 a month |
| Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month If your users need to watch or share a live browser together, Hyperbeam is built for that; if they need their own machine to run code, boat fits. | Cheapest of 361 $209 a month | 321st of 361, with limits $4,253 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 | 313th of 360, with limits $20,813 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 | 298th of 364, with limits $1,610 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 | 276th of 377, with limits $434 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB A dev box needs SSH and a shell, and a Hyperbeam session is reached only through its embed URL and API. | 18th cheapest of 368 $26 a month | 309th of 368, with limits $237 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Both fit; Hyperbeam streams a browser session, while boat streams a full Linux desktop at 60fps so an agent can use apps beyond the browser. | 6th cheapest of 369 $200 a month | 33rd cheapest of 369 $1,442 a month |
| Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month Both fit; Hyperbeam is a browser you drive with input and navigation calls, while boat puts the browser on a whole machine where the agent can also save and process what it finds. | 7th cheapest of 367 $200 a month | 42nd cheapest of 367 $1,442 a month |
| Scraping & bots (startup scale) 50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month Both can run startup-scale bots; on boat each bot keeps its files and tools on a disk that survives stops, while Hyperbeam keeps only the browser profile. | 6th cheapest of 346 $90 a month | 81st cheapest of 346 $1,978 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 | 315th of 338, with limits $40,880 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 | 83rd of 360, with limits $928 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 | Hyperbeam |
|---|---|---|
| What you get | A full Linux VM with its own kernel and root, on dedicated AMD Ryzen 9 9950X hosts | A streamed Chromium or emulator session; the machine behind it is not exposed |
| Shell, code and Docker | SSH, any language, Docker and systemd | No shell or SSH; the API controls the browser, not the machine |
| Nested virtualization | Yes, /dev/kvm for Android emulators and VMs you run yourself | Not exposed; Android (private beta) and console emulators are session types Hyperbeam runs |
| State after a session | The whole disk is saved every minute and on stop; a stopped sandbox costs nothing | Opt-in browser profile (cookies, history, passwords, bookmarks) saved to S3 when the session ends; deleted after three months unused |
| Fork | Fork a running sandbox with its disk | Load one browser profile and save it as a new one when a new session starts |
| Session length | No cap on paid plans; you set the time to live | No absolute cap by default; ends after 1 hour with nobody connected unless you change it |
| Live view | A full Linux desktop streamed at 60fps, or Chromium alone; embed it in your app | Chromium streamed over WebRTC at 24 to 60 fps, up to 1920 x 1080, embedded with an SDK |
| Many people in one session | Not what it is built for | Up to 10,000 participants with synced audio and video, roles and per-user control |
| Driving it from code | Claude Code, Codex and other coding agents preinstalled, with a prompt, steer and events API; desktop control for computer use | Mouse, keyboard and wheel events, tab and navigation calls, video frames and webhooks; no agent included |
| Hosting | HTTPS URLs for up to 50 ports, token-protected by default; raw TCP | No ports of your own; only the embed URL, with token or webhook auth |
| Keeping API keys from the agent | Credentials proxy: real keys never enter the sandbox, for any HTTPS API | No credentials proxy; you can route browser HTTP traffic through your own proxy |
| Outbound IP | Open internet from the VM; run any proxy client inside it | Your own HTTP proxy through a PAC file; no fixed IP range to allowlist |
| Regions | EU only (Germany, Finland, France) | North America, Europe and Asia |
| Compliance and transparency | SOC 2 is in progress. We're open about where we stand: live status and data retention policy. | Not published |
Every feature Battleships tracks (43)
| Feature | boat | Hyperbeam |
|---|---|---|
| Full VM (own kernel) | Yes | Not published |
| Docker inside | Yes | No |
| Nested virtualization | Yes | No |
| Root | Yes | Not published |
| Systemd | Yes | Not published |
| Desktop GUI | Yes | Yes |
| Browser | Yes | Yes |
| Computer-use API | Yes | Yes |
| SSH | Yes | No |
| Persistent disk | Yes | No |
| Pause/resume | Yes | No |
| Automatic snapshots | Yes | No |
| Fork/clone | Yes | No |
| Live fork (no pause) | Yes | No |
| Memory fork | No | No |
| Custom image (Docker or snapshot) | Yes | Not published |
| ≥24 h sessions | Yes | Yes |
| Idle auto-stop | No | Yes |
| Wake on request | No | No |
| HTTPS preview URLs | Yes | No |
| Raw TCP inbound | Yes | Not published |
| Public IPv4 | No | No |
| Egress allowlist | No | No |
| 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 | No |
| Windows | No | No |
| Arm64 | No | Not published |
| 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 | Yes |
| Audit logs | No | Not published |
| SSO | No | Not published |
| SOC 2 | No | Not published |
| HIPAA | No | Not published |
| EU data residency | Yes | Yes |
| Zero data retention | Yes | Not published |
Performance
There is no comparable measurement. Hyperbeam sells participant-minutes and does not publish the CPU or memory behind each session, and it is not in the CPU benchmark. boat scored 26.69 runs per second on the hpc-sandbox-benchmarks Node.js test 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. Hyperbeam 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 | Hyperbeam, 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 | $0 · 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 | $237 · 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 | $3,256 · 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 | $1,820 · 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 | $4,107 · Pay as you go · 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.
Hyperbeam: Hyperbeam gives 10,000 free minutes a month, then bills $0.007 per participant-minute, with volume discounts that apply on their own. It counts participants, not machines, so the more people connected to a session, the more minutes it uses.
Limits in those scenarios
None of the five scenarios fits Hyperbeam as written, because each one asks for a general machine and Hyperbeam sells a browser session with no shell, no persistent disk and no Docker, of a size it does not publish. One coding agent on workdays shows $0 because it stays inside the free minutes, not because it fits. Hyperbeam does not publish how many sessions can run at once, so 300 sandboxes for your users and 1,000 short jobs at once are not confirmed. boat fits four scenarios; at 1,000 short jobs at once its $2,000 plan queues the starts.
| Scenario | boat | Hyperbeam |
|---|---|---|
| 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 Hyperbeam when
you want to put a live browser or emulator inside your own app, so many users can watch and control it together.
- Many people in one live browser. For co-browsing and watch-together features: up to 10,000 participants with synced audio and video, and roles for who can control it.
- Made for embedding in your app. For apps that show a remote browser or emulator to their own users, with JavaScript and Unity SDKs and Chrome extensions.
- Regions in America and Asia. Sessions run in North America, Europe or Asia, picked per session to keep the stream close to the people watching.
Use boat when
your agent does more than click through pages: it needs a shell, files, Docker, a desktop or a server next to the browser, and it has to keep its work between runs.
- A whole machine around the browser. A Hyperbeam session is a browser with no shell, so the agent's code, files and tools have to live on another service and send every input event across the network.
- Its work is still there tomorrow. Hyperbeam saves only the browser profile, and only when the session ends. Downloads, installed tools and anything else the agent made start over in the next session.
- Coding agents and keys built in. Hyperbeam gives you input events and navigation calls, so you build the agent loop yourself and keep its keys somewhere else.
Questions people ask
Is Hyperbeam a sandbox like boat?
No. Hyperbeam streams a cloud Chromium or emulator into a web page and lets you control it with input and navigation calls, with no shell on the machine. boat gives you a whole Linux VM with Chromium, a desktop, Docker and root, so the browser is one tool among many.
Is boat cheaper than Hyperbeam?
It depends on the scenario. One coding agent on workdays costs $20 on boat and $0 on Hyperbeam, inside its free minutes. One agent running all month costs $26 against $237, 50 agents in parallel $317 against $3,256, and 300 sandboxes for your users $200 against $1,820. Hyperbeam fits none of them as a general machine.
Can an AI agent control a Hyperbeam browser?
Yes, with your own code. The SDK sends mouse, keyboard and wheel events, controls tabs and navigation, and hands you the video frames, so you can build a computer-use loop on top. It does not include an agent; boat ships coding agents and a desktop-control tool inside each sandbox.
Does Hyperbeam keep files between sessions?
It keeps an opt-in browser profile: cookies, history, passwords, bookmarks and tabs, saved when the session ends and deleted after three months unused. A boat sandbox keeps its whole disk, saved every minute and when it stops.
Can several people share one session?
That is Hyperbeam's strength. One session can have up to 10,000 participants with synced audio and video, and you choose per user who can control it. boat is built for an agent's own machine, not for crowds watching one browser.
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