boat vs other providers
A full Linux VM with SSH access. Disk-level fork, public IPv6 and outbound IPv4, 60 fps desktop, Docker inside. One price per hour per size, from $20 a month. Every competitor fact on this page links to the provider's own docs, and boat's own limits are listed below.
Optimized for building powerful & highly profitable agent platforms
Some examples:
Agent factory
Many agents in parallel, each on its own VM. Fork to branch. Cross-VM comms.
24/7 background services
Public IPv6 and outbound IPv4, Docker inside. Auto-stop defaults to 1 hour; turn it off on a paid plan.
Personal dev sandbox
SSH from anywhere, desktop when you need it. Files and installed packages come back after a stop.
Price
One hourly rate per machine size, billed per second while it runs. A stopped sandbox has no compute charge, and boat keeps its disk at least 14 days. The same machine on every other provider, at their list prices.
Per hour, one machine
Each provider is priced on its own published unit. Modal sells CPU by the physical core, and 1 core = 2 vCPU, so a 4 vCPU machine is 2 Modal cores. Providers marked “active CPU” bill only the CPU time your code uses; this view shows them at full use, their highest price. A nearly idle sandbox can cost less on those providers than on boat. boat bills the whole running time, busy or idle.
Per month, a fleet
boat: $20 plan + $412 extra VM time. Other providers: time only, with “active CPU” providers at 30% CPU use (50% RAM use where RAM is billed by use), the Battleships engine default. Their plan fees, start limits and per-start fees are not in these bars; the scenario prices on the one-by-one pages include them.
Plans: what you run at once
| Plan | Run at once | Time included | Starts included / min · hour · day | Most starts a day, with paid extras |
|---|---|---|---|---|
| $20/mo | 100 | $20 worth · 555 h of default | 12 · 60 · 200 | 1,200 |
| $100/mo | 300 | $100 worth · 2,777 h of default | 30 · 210 · 840 | 5,040 |
| $500/mo | 1,000 | $500 worth · 13,888 h of default | 65 · 420 · 1,680 | 10,080 |
| $2,000/mo | 2,000 | $2,000 worth · 55,555 h of default | 90 · 600 · 2,400 | 14,400 |
The plan price comes back as sandbox time at the rates above for that month, plus the right to run that many at once. Unused time does not carry over. Past it, your wallet pays the same rates, and money in the wallet does not expire. Each extra start above the included number costs $0.03 to $0.07, paid from the wallet. A start fails with 429 above 3 times the included starts per minute, or 6 times per hour or per day. A stopped sandbox has no compute charge, and boat keeps its disk at least 14 days. The 7-day trial needs a card and includes 25 hours. Details in the pricing and limits docs.
boat's own limits
- Plans start at $20 a month. The plan price comes back as VM time for that month. Unused time does not carry over. Source
- Starts (create, fork, resume): the $20 plan includes 12 a minute, 60 an hour and 200 a day. The $2,000 plan includes 90 a minute, 600 an hour and 2,400 a day. Extra starts cost $0.03 to $0.07. A start fails with 429 above 3 times the included starts per minute, or 6 times per hour or per day. Source
- A stopped sandbox has no compute charge. boat keeps its disk for at least 14 days. A saved snapshot keeps it longer; extra saved snapshots and the AI credentials proxy are paid. Source
- Auto-stop defaults to 1 hour. You can turn it off on a paid plan. The trial allows 2 hours at most. Source
- boat saves the disk every minute and at each stop. Memory, running programs and /tmp are not saved: a resume is like a reboot. A machine failure loses writes after the last good snapshot. Source
- Machines run in Germany, Finland and France only. Snapshot copies can be stored outside the EU. Source
- Public IPv6 and outbound IPv4. No dedicated IPv4. The IP changes at each resume. Source
- Hosts are AMD Ryzen 7 9700X and Ryzen 9 9950X. vCPUs are shared. No GPU. You cannot boot your own Docker image. Source
Performance
Price per hour is half the story. The other half is how much work an hour does. Each provider appears on the runtimes it was measured on; Modal appears twice, on gVisor and on its VM runtime.
StarSling's run of 3 October 2026 (run 37073680030): the 4 vCPU / 8 GB machine of every provider (our default) running the V8 Web Tooling Benchmark. Seconds per 1,000 runs, computed from runs per second, lower is better:
Speed and price side by side
| 4 vCPU / 8 GB | Runs per second | StarSling rank (of 15) | $ / hour | $ per million runs |
|---|---|---|---|---|
| Brezel † | 30.06 | 1 | – | – |
| boat † | 28.06 | 2 (shared) | $0.036 | $0.36 |
| Namespace † | 24.90 | 2 (shared) | – | – |
| Daytona † | 21.52 | 2 (shared) | $0.331 | $4.28 (12x) |
| microsandbox † | 21.34 | 2 (shared) | – | – |
| Novita | 20.16 | 2 (shared) | $0.233 | $3.21 (9x) |
| Blaxel † | 18.03 | 7 (shared) | $0.331 | $5.10 (14x) |
| Modal (VM runtime) † | 15.48 | 7 (shared) | $0.760 | $13.63 (38x) |
| Vercel Sandbox † | 12.66 | 7 (shared) | $0.682 | $14.96 (42x) |
| Modal (gVisor) † | 12.60 | 7 (shared) | $0.760 | $16.75 (47x) |
| E2B | 11.08 | 7 (shared) | $0.331 | $8.30 (23x) |
| Freestyle | 10.22 | 14 | $0.264 | $7.19 (20x) |
| Runloop | 8.93 | 15 | $0.634 | $19.71 (55x) |
† Brezel: Firecracker microVM; no list price checked.
† boat: every boat sandbox in this run ran on an AMD Ryzen 9 9950X host. boat also places some sandboxes on AMD Ryzen 7 9700X hosts, which this run did not measure.
† Namespace: container on Firecracker; no list price checked here.
† Daytona: 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): 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.
One run for every row. StarSling marks this run partial (800 of 840 cells). Providers ran on different hardware in different regions. A shared rank means the run cannot separate those providers, so read close numbers as a tie. On the three real-repository tests of the same run (Better-Auth, Mastra, OpenClaw), Brezel was fastest and boat second.
Cost per run is the list price per hour divided by the speed, with multiples against boat. boat has the lowest cost per run in this table. Vercel Sandbox bills busy CPU only and Blaxel bills active time only; this table prices both at full use. “–” = no list price checked.
The landscape
01
Isolation
Whether each sandbox runs its own kernel (a VM or microVM) or shares the host kernel (a container or gVisor).
Capabilities
Docker inside
Built in
No documented way, or not published
SSH
Built in
With your own setup
No documented way, or not published
≥24 h sessions
Built in
boat: auto-stop defaults to 1 h. You can turn it off on a paid plan.
Persistent disk
Built in
With your own setup
boat: a stopped disk is kept at least 14 days. A saved snapshot keeps it longer.
Fork/clone
Built in
With your own setup
No documented way, or not published
boat: a fork of a running sandbox uses its last 1-minute snapshot.
Memory fork
Built in
No documented way, or not published
Desktop GUI
Built in
With your own setup
No documented way, or not published
Browser
Built in
With your own setup
No documented way, or not published
Preinstalled agents
Built in
With your own setup
HTTPS preview URLs
Built in
Public IPv4
No documented way, or not published
boat: public IPv6 and outbound IPv4. SSH and HTTPS URLs work over IPv4 through shared hosts. No dedicated IPv4.
Secret proxy
Built in
With your own setup
No documented way, or not published
boat: optional, $1 a day plus $1 per 1,000 requests.
GPU
Built in
No documented way, or not published
Self-hosting
Built in
No documented way, or not published
From the provider cards on battleships.dev, each checked against the provider's own docs. “With your own setup” means the provider documents a way, for example a snapshot API, a mount, an install recipe or another runtime. Open a provider's one-by-one page for the sources.
Compare one by one
Company, use cases, features, performance, prices and limits, side by side.
Agent sandboxes (88)
Dev environments (13)
Cloud browsers (21)
Clouds and VPS hosts (45)
App platforms (22)
CI and build services (11)
GPU clouds (12)
Not listed? Ask us directly.
Head to head
Two other providers side by side, with boat as the third column.
Agent sandboxes and dev environments (666)
Cloud browsers (45)
CI services (55)
Clouds and VPS hosts (66)
GPU clouds (45)
7-day free trial