Cost per hour
You pay per second, only while a sandbox runs. A stopped sandbox costs no machine time.
*
xlarge needs the $100 plan or higher. Subscribe from Dashboard > Billing, then pass --type xlarge. You do not need to contact us.
defaultis the size that other providers call medium.- Start with
default. Usesmallfor light scripts. Uselargeorxlargefor big builds or many containers. - No size has a GPU.
- “Disk” is the space for your own files. The OS and the pre-installed tools are not in it.
- One
defaultsandbox that runs a full month (730 h) costs $26. - Every size is a full Linux virtual machine with the same tools, snapshots, hosting and desktop. Only the machine changes.
default costs $0.036 an hour. E2B and Daytona charge $0.331 an hour for the same machine (9x). Modal charges $0.760 (21x). Read Compared to others.
Pick the size when you create a sandbox. To change it, stop the sandbox, then resume it with a new size. A fork can also take a new size.
Included at no cost
- Automatic snapshots of your sandboxes, with no count limit. boat keeps the newest snapshot for at least 14 days after a stop. Automated retention cleanup can delete it later once enabled.
- Saved snapshots included with your plan: 10 for every $20 of plan per seat, never fewer than 10. A saved snapshot has no auto-deletion date. You cannot save one during the free trial.
- Network traffic. boat does not bill egress.
- A public IPv6 address on every sandbox, and outbound IPv4. The IP address can change when a sandbox resumes.
- Public URLs for the ports of a sandbox.
Plans
A plan is a monthly price that your card pays. It is not a fee. Every dollar comes back as VM time at the rates above. Example: the $20 plan includes 555 hours ofdefault and lets you run 100 sandboxes at the same time.
Your first plan starts with a 7-day free trial. A card is required.
Plan limits and included
- Create, fork and resume each count as one start. Above your included starts, a start costs $0.03 to $0.07. Read Extra starts.
- Included VM time expires at the end of the billing period. boat does not carry it over.
- Seats: an organization pays the plan price once per seat. A 5-seat organization on the $20 plan pays $100 a month. Each seat adds the plan’s included amounts. It adds the full limits on the $2000 plan, and 20% of them below. Read Seats, limits and caps.
- boat raises limits only on custom plans above $2000 a month. Contact us for a custom plan.
Overage, paid from the wallet
boat uses the included amounts first. Then the wallet pays.- Add money with Pay or auto top-up. A payment of $200 or more gets a bonus of 2% to 25%. Read Wallet.
- Saved snapshots and the AI proxy also charge while your sandboxes are stopped. Delete a saved snapshot to stop its charge.
- For an empty wallet, read Running out.
boat billing plan 100. A change applies at once. Read Card, invoices and plan changes.
Compared to others
† exe.dev: standalone VM rate, RAM included; Cloudflare: active CPU only; Modal: 1 core = 2 vCPU (Modal pricing); Vercel Sandbox: active CPU only; Blaxel: CPU included in the RAM price; billed only while active; Railway: active CPU and used RAM only; Fly.io Sprites: active CPU and used RAM only; AWS AgentCore: active CPU and used RAM only.
”Active CPU” billing
Some providers bill only the CPU time that your code uses. They bill RAM in addition. Some bill the RAM that you reserve. Others (‡) bill the RAM in use. Their claim is “you will rarely max the box”. This is what it means for you:- You cannot know the price before you run. The same job costs a different amount each time. Builds, test suites, browsers and agents use mostly CPU.
- RAM bills while you are idle. Reserved or in use, you pay for it every second that the sandbox runs.
- They do not publish what their users use. So you cannot check what “rarely max the box” saves you.
‡ bills RAM used, not RAM reserved.
A sandbox that is almost idle can cost less with a provider that bills only busy CPU and used RAM. From the median sandbox up, boat has the lowest price in this table. Below about $20 of use a month, providers with a $0 plan or free credits cost less.
Are your sandboxes idle most of the time? Then stop them, and resume them when work arrives. Use
boat stop / boat resume, or the API. A stopped sandbox has no compute charge and keeps its disk for at least 14 days. Saved snapshots above the included count and the AI credentials proxy still charge while sandboxes are stopped.
Performance
StarSling’s hpc-sandbox-benchmarks ran the V8 Web Tooling Benchmark on the 4 vCPU / 8 GB machine of every provider. This is ourdefault size. Every row below comes from StarSling’s run of 3 October 2026 (run 37073680030). StarSling marks that run partial: 800 of 840 cells. For the full results, read the leaderboard.
† 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.
StarSling’s run of 2026-10-03, one run for every row. Providers ran on different hardware in different regions. Multiples are against boat.
- boat did 28.06 runs per second, second of 15 in that run. StarSling ranks boat, Namespace, Daytona VM, microsandbox and Novita together: the run cannot separate them. Brezel scored highest.
- The measured sandbox ran on an AMD Ryzen 9 9950X host. The fleet also has AMD Ryzen 7 9700X hosts. boat does not promise a CPU model. Sandboxes are full virtual machines with shared vCPUs.
- Values are medians of 5 sandboxes per provider, each asked for 4 vCPU / 8 GB.
Worked examples
For the API
GET /limits gives the plan and the wallet:
Each size uses VM time at its own rate:
boat usage <id>andGET /sandboxes/{id}/usagegive the bill of one sandbox. The size is already applied.creditBalanceHours,packBalanceDollarsand the otherpack…fields are deprecated. ReadwalletandplanIncluded.