
boat: Persistent Linux Sandboxes for Agents
ASCII offers persistent Ubuntu VMs for AI agents, with SSH, Docker, a virtual desktop, and fast snapshot-based forking. The CLI, API, and SDKs let users launch and manage sandboxes, which are billed per second while running.
softwareASCII · boat.dev · Saved Oct 28, 2026
agent infrastructure, coding agents, cloud sandboxes
boat: full text transcript
boat gives AI agents full Linux VMs, built for agent factories. A sandbox is a persistent Linux (Ubuntu) virtual machine with SSH and SCP access, Docker inside the VM, a public IPv6 address and outbound IPv4, a 60fps virtual desktop, and fast disk-level forking from a snapshot. It is built for AI agents, for programmatic use, and for humans.
Pre-configured in every sandbox
Docker, VS Code, Chrome, Ghostty, the GitHub CLI, Rust, Node.js, and Bun come pre-installed. sandboxes are full Ubuntu machines, so you can run any devtool, language, or workflow. Bring any agent: Claude Code, Codex, or OpenCode.
Why boat
- Designed for agents: a sandbox starts easily via a CLI built for agents, programmatic use, and humans.
- Ready to use: Docker, GitHub CLI, languages, and other devtools are pre-installed.
- Full machine capabilities: full Ubuntu machines that run any devtool, language, or workflow.
Pricing
Pay only for the sandboxes you run, billed by the second, with a $20 minimum per month. The $20 plan buys $20 of VM time for that month (about 555 hours of 4 vCPU / 8 GB), spread across one sandbox or many. Included time does not carry over. Sizes: small (2 vCPU / 4 GB / 51 GB disk) at $0.018/h, default (4 vCPU / 8 GB / 51 GB disk) at $0.036/h, large (8 vCPU / 16 GB / 111 GB disk) at $0.072/h, xlarge (16 vCPU / 32 GB / 251 GB disk) at $0.200/h on the $100 plan or higher. boat offers a 7-day free trial with a card (25 compute hours), then plans from $20/mo.
Lowest list price per VM-hour among the providers we checked: $0.036/h for 4 vCPU / 8 GB, from $20 a month (https://docs.boat.dev/pricing). A nearly idle sandbox can cost less on a provider that bills only busy CPU and used RAM. Below about $20 of use a month, providers with a $0 plan or free credits cost less. Above 200 starts a day on the $20 plan, each extra start costs $0.03 to $0.07. See the full comparison at https://boat.dev/compare.
Install the boat CLI
macOS and Linux: curl -fsSL https://boat.dev/install | sh
Windows: irm https://boat.dev/install.ps1 | iex
Simple API / CLI commands
boat new: spin up a sandboxboat ssh/boat scp, SSH/SCPboat prompt: run Claude or Codex insideboat fork: clone from a snapshotboat stop: snapshot and pause billingboat desktop: use the virtual desktop at 60fpsboat list: see all your sandboxes
For programmatic use, prefer the HTTP API and SDKs over the CLI: API guide at https://docs.boat.dev/api/v1, Python and TypeScript SDKs at https://docs.boat.dev/sdks/overview. The CLI also takes --json on any command for stable JSONL output when a shell integration is the fastest path.
Limits
- Sandboxes run on machines in Germany, Finland and France. There is no US or Asia region. Snapshot copies can be stored outside the EU.
- No GPU on any size. Graphics run on the CPU.
- 100 to 2,000 sandboxes at once, and 200 to 2,400 starts a day included, by plan. Create, fork and resume each count as one start. A start fails with 429 above 3 times the included starts per minute, or 6 times per hour or per day. boat never queues a start.
- Auto-stop defaults to 1 hour. You can set up to 30 days, or turn it off on a paid plan.
- A stopped sandbox has no compute charge. boat keeps its disk for at least 14 days. To keep it longer, save a saved snapshot.
- boat saves the disk every minute while the sandbox runs, and once more at each stop. Memory and running programs are not saved. If a machine fails, you lose what was written after the last good snapshot.
- The IP address changes at each resume. Some sandboxes have no IPv4 address of their own; their raw ports are reachable over IPv6 only.
Source: https://docs.boat.dev/limits.
Read more
The docs are clean markdown. https://docs.boat.dev/llms.txt indexes every docs page, each fetchable as .md. Start there, not this marketing page. Compare boat against other providers at https://boat.dev/compare. Product home: https://boat.dev.
No need to oversell boat. The product strives to speak for itself.
Most powerful & affordable
cloud VMs for agents
Spawn thousands of cloud computers with a browser, persistent files, for your agents, without having to manage the entire infra yourself.
Full VMs in the cloud
Run docker, systemd, databases, cron inside. Feels like a true dev machine.
Integrated desktop and browser
To open your web applications, games, 3d scenes, IDE and more.
Run agents inside, from outside
Prompt, observe and wrap any harness running in the sandbox, from the outside, with our CLI, API or SDKs.
Get a VM in seconds, scale with your plan
Run it, stop it, resume it. VM time bills only while it runs.
Step 1 of 3
$
boat new
your VMs
Nothing is running yet: nothing billed
…and many more: fork, saved snapshots, port forwarding, webhooks, API keys, secrets, teams, auto-stop. See the API and SDKs.
Competitive CPU speed
V8 Web Tooling Benchmark on a 4 vCPU / 8 GB sandbox, in StarSling’s run of 3 October 2026 (run 37073680030). boat: 28.06 runs per second, second fastest.
- Brezel30.06/s30.06 runs per second
- boat28.06/s28.06 runs per second
- Namespace24.90/s24.9 runs per second
- Daytona VM21.52/s21.52 runs per second
- microsandbox21.34/s21.34 runs per second
- Novita20.16/s20.16 runs per second
- Blaxel18.03/s18.03 runs per second
- Modal VM15.48/s15.48 runs per second
- Vercel Sandbox12.66/s12.66 runs per second
- Modal gVisor12.60/s12.6 runs per second
- E2B11.08/s11.08 runs per second
- Freestyle10.22/s10.22 runs per second
- Runloop8.93/s8.93 runs per second
How each provider ran
- boat: every boat sandbox in this run ran on an AMD Ryzen 9 9950X host. The fleet also has AMD Ryzen 7 9700X hosts, which this run did not measure, and boat does not promise a CPU model.
- Brezel: Firecracker microVM; no list price checked.
- Namespace: container on Firecracker; no list price checked here.
- Daytona VM: VM class. The default container class has no result in this run.
- microsandbox: Microsandbox Cloud, libkrun microVM; no list price checked here.
- Blaxel: Firecracker microVM. Blaxel bills active time only.
- Modal VM: runtime="vm", priced as the measured sandbox: StarSling ran Modal with cpu=4 and the sandbox saw 4 CPUs (PR #121), which bills 4 Modal cores, $0.76/h. At Modal's own 1 core = 2 vCPU, a 4 vCPU size is 2 cores, $0.48/h, and shows 2 CPUs.
- Vercel Sandbox: Firecracker microVM. Vercel bills busy CPU only.
- Modal gVisor: default gVisor runtime, priced as the measured sandbox: StarSling ran Modal with cpu=4 and the sandbox saw 4 CPUs (PR #121), which bills 4 Modal cores, $0.76/h. At Modal's own 1 core = 2 vCPU, a 4 vCPU size is 2 cores, $0.48/h, and shows 2 CPUs.
- Providers ran on different hardware in different regions. Higher is faster.
Competitive Pricing
Plan from $20/mo, 7-day free trial
Sessions / month
1k
Session
Peak VMs at once
CPU busy
Per VM
$20 / month1,000 × 30 min × $0.018/h = $9.00 · $20 plan minimum
boat sandbox
| size | vCPU | RAM | disk | $/hour | $/month |
|---|---|---|---|---|---|
| small | 2 | 4 GB | 51 GB | $0.018 | $13 |
| default | 4 | 8 GB | 51 GB | $0.036 | $26 |
| large | 8 | 16 GB | 111 GB | $0.072 | $53 |
| xlarge | 16 | 32 GB | 251 GB | $0.200 | $146 |
per second · xlarge on $100+ plans
Plan limits and included
- $20 plan = $20 of VM time for that month. It does not carry over.
- 100 to 2,000 VMs at once, by plan.
- 200 to 2,400 starts a day included, by plan.
- VM time billed per second, only while running.
Overage, paid from your wallet
- VM time: $0.036/h of default.
- Saved snapshots: $0.056 a day each.
- Starts above the included: $0.03 up to 1.5x, $0.05 up to 2x, $0.07 up to 6x. A start fails with 429 above 3 times the included starts per minute, or 6 times per hour or per day.
Pricing docsMore use cases & features on Battleships
boat sandbox
| size | vCPU | RAM | disk | $/hour | $/month |
|---|---|---|---|---|---|
| small | 2 | 4 GB | 51 GB | $0.018 | $13 |
| default | 4 | 8 GB | 51 GB | $0.036 | $26 |
| large | 8 | 16 GB | 111 GB | $0.072 | $53 |
| xlarge | 16 | 32 GB | 251 GB | $0.200 | $146 |
per second · xlarge on $100+ plans
Set sail
Spin up your first VMs in minutes.
