boat vs PPIO Agent Sandbox

Both give an AI agent its own isolated Linux machine. boat gives it a full VM in Europe that runs Docker and emulators and comes with coding agents installed. PPIO is an E2B-compatible sandbox in mainland China that can pause with its memory and add shared Volumes.

Where boat stands out

A whole machine, not just a runtime

A boat sandbox is a full VM with root, systemd, Docker and /dev/kvm, so an agent can run compose stacks, Android emulators and VMs. Machines go up to 16 vCPU / 32 GB on the $100 plan and above. You can SSH straight in or open raw ports.

Coding agents you drive from outside

Claude Code, Codex and five other agents are installed, and you prompt, steer and follow them through one API. With the optional paid credentials proxy on, the real keys of supported providers stay outside the sandbox. A credential can switch to another account when one runs out of credits.

Machines in Europe

boat runs sandboxes on machines in Germany, Finland and France. You cannot pick a region, and snapshot copies can be stored outside the EU. PPIO's sandboxes run in Shanghai and Beijing.

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.

PPIO Agent Sandbox

PPIO is a Chinese AI cloud that also sells GPUs and model APIs. Agent Sandbox is its sandbox service for AI agents.

Each PPIO sandbox is a Firecracker microVM with its own filesystem and runtime, built from a template that can start from a Docker image or Dockerfile. It speaks the E2B API, so the E2B SDK works with a PPIO key. It adds pause and resume with memory, snapshots with memory, shared Volumes, S3 mounts, a VNC desktop and a browser template. It runs in two regions in mainland China and suits teams whose agents and users are there.

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. “5th cheapest of 365” means 4 of the other 364 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 workloadboat: price rankPPIO Agent Sandbox: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
A software factory needs Docker for databases and services. boat runs Docker and compose in every sandbox. PPIO's docs do not describe Docker inside a sandbox, so a compose stack is not a documented path there.
5th cheapest of 365
$634 a month
42nd of 365, with limits
$1,566 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
An assistant that lives on its machine all month fits boat once you turn auto-stop off (default 1 h); it bills at full rate, busy or idle. On PPIO you create it in long-running mode with a long timeout and watch the idle timeout.
6th cheapest of 361
$657 a month
15th cheapest of 361
$2,021 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
PPIO suits short code runs: sandboxes bill per second and can pause with their memory between requests.
189th of 336, with limits
$2,000 a month
88th of 336, with limits
$106 a month
Sandboxes for your users
300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month
Both can host sandboxes for your users. PPIO allows 50 at once by default and publishes no start-rate limit. boat's $20 plan allows 100 at once and includes 200 starts a day, charges $0.03 to $0.07 for each extra start and refuses starts above 1,200 a day.
3rd cheapest of 349
$209 a month
81st of 349, with limits
$668 a month
RL rollouts
1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month
39th of 349, with limits
$2,000 a month
62nd of 349, with limits
$3,201 a month
Agent evals
250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month
10th cheapest of 359
$200 a month
127th of 359, with limits
$515 a month
CI runners
40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month
CI jobs that build containers or Android apps need Docker and /dev/kvm. boat has both in every sandbox. PPIO's docs describe neither, so it is not a documented path there.
2nd cheapest of 374
$86 a month
282nd of 374, with limits
$464 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
A dev box needs SSH, Docker and a disk that stays. boat gives direct SSH and keeps a stopped sandbox's disk for at least 14 days, or longer as a saved snapshot. On PPIO you set up SSH yourself: OpenSSH in a custom template through a WebSocket tunnel, or Tailscale or OpenVPN. A paused PPIO sandbox keeps its files and memory until you kill it.
19th cheapest of 371
$26 a month
64th cheapest of 371
$81 a month
Computer-use agents
50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month
Both have a Linux desktop for computer use. boat streams it at 1920x1080 and 60 fps. PPIO shows it through a VNC page.
7th cheapest of 367
$200 a month
13th cheapest of 367
$464 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
7th cheapest of 356
$200 a month
67th of 356, with limits
$232 a month
Scraping & bots (startup scale)
50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month
6th cheapest of 343
$90 a month
12th cheapest of 343
$157 a month
Mass scraping & bots (undetectable)
1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month
73rd of 327, with limits
$1,800 a month
118th of 327, with limits
$3,138 a month
Parallel red teaming
200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month
PPIO can snapshot a running sandbox with its memory and start many sandboxes from it (Beijing region), which helps when many attackers start from the same state. boat forks a running sandbox from its last 1-minute disk snapshot, without memory.
3rd cheapest of 351
$200 a month
40th of 351, with limits
$154 a month
GPU eval jobs
10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month
Not ranked: misses a requirementNot ranked: misses a requirement
macOS agents
5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month
Not ranked: misses a requirementNot ranked: misses a requirement

Key features

FeatureboatPPIO Agent Sandbox
IsolationFull VM with its own kernel, on physical hosts dedicated to boat (AMD Ryzen 7 9700X and Ryzen 9 9950X); vCPUs are sharedFirecracker microVM with its own filesystem and runtime
Docker insideYes, Docker with BuildKit and Compose in every sandboxNot described in the docs; Docker images and Dockerfiles build templates
Nested virtualizationYes, /dev/kvm for Android emulators and VMsNot described in the docs
Biggest machine16 vCPU / 32 GB on the $100 plan and above; 8 vCPU / 16 GB on the $20 plan8 vCPU / 8 GiB
Persistent storage51 to 251 GB disk, fixed by size; kept at least 14 days after a stop, longer as a saved snapshot20 GB local disk, plus Volumes shared across sandboxes (Beijing, closed beta) and S3 mounts
Session lengthAuto-stop after 1 hour by default; up to 30 days, or off on a paid plan for no cap5 minutes by default, 1 hour at most, or longer in long-running mode (docs show 7 days; no maximum stated); idle timeout still applies
Pause and resumeDisk is saved every minute and at each stop; memory and running programs are not saved, so a resume is like a rebootPause keeps files and memory with running processes; can wake on the next command
Snapshots and forkFork a running sandbox from its last 1-minute disk snapshot; saved snapshots with no scheduled expirySnapshots with memory of a running sandbox, to start many new sandboxes from (Beijing region)
SSHDirect SSH to every sandbox, no setupYou set it up: OpenSSH in a custom template via a WebSocket tunnel; Tailscale and OpenVPN guides for private access
Inbound accessUp to 50 HTTPS URLs, token-gated by default; public IPv6 per sandbox, raw ports over IPv6 where a sandbox has no IPv4 of its ownHTTPS URL per port, token-protected unless you allow public traffic; no public IP per sandbox
Outbound rulesOpen internet, no provider-side allowlistAllow and deny lists by IP or CIDR, changeable on a running sandbox
Coding agentsSeven agents installed (Claude Code, Codex, pi, OpenCode, Prime Agent, Kimi Code, Mistral Vibe), with a prompt, steer and events APIA Codex template, plus guides to run OpenAI Agents API or Kimi Hosted Agents sessions in PPIO sandboxes
Keeping API keys out of the sandboxOptional paid credentials proxy ($1 a day on, plus $1 per 1,000 requests) for model providers and custom HTTPS APIs; switch a credential to another account when credits run outSecrets swapped into HTTPS request headers by an outside proxy, Beijing region only
DesktopLinux desktop and Chrome streamed at 1920x1080, 60 fps, with computer-use toolsVNC desktop, a Chromium browser template and a computer-use guide
RegionsGermany, Finland and France; no US or Asia region, and you cannot pick oneMainland China (Shanghai, Beijing); you pick the region
Compliance and transparencySOC 2 is in progress. We're open about where we stand: live status and data retention policy.Not published
Every feature Battleships tracks (43)
FeatureboatPPIO Agent Sandbox
Full VM (own kernel)YesYes
Docker insideYesNot published
Nested virtualizationYesNot published
RootYesNot published
SystemdYesNot published
Desktop GUIYesYes
BrowserYesYes
Computer-use APIYesYes
SSHYesYes
Persistent diskYesYes
Pause/resumeYesYes
Automatic snapshotsYesYes
Fork/cloneYesYes
Live fork (no pause)YesYes
Memory forkNo documented wayYes
Custom image (Docker or snapshot)YesYes
≥24 h sessionsYesYes
Idle auto-stopNo documented wayYes
Wake on requestNo documented wayYes
HTTPS preview URLsYesYes
Raw TCP inboundYesNo documented way
Public IPv4No documented wayNo documented way
Egress allowlistNo documented wayYes
Secret proxyYesYes
Secret proxy for any APIYesYes
Preinstalled agentsYesYes
Agent harness APIYesNo documented way
MCP serverNo documented wayYes
GPUNo documented wayNo documented way
WindowsNo documented wayNo documented way
Arm64No documented wayNo documented way
Self-hostingNo documented wayNo documented way
BYOCNo documented wayNo documented way
Open sourceNo documented wayNo documented way
Scoped API keysYesNot published
Spend limitsYesNot published
Usage per sandbox (API)YesNo documented way
Audit logsNo documented wayNot published
SSONo documented wayNo documented way
SOC 2No documented wayNot published
HIPAANo documented wayNo documented way
EU data residencyNo documented wayNo documented way
Zero data retentionYesNot published

Performance

boat28.06 runs/s
PPIO Agent SandboxNot measured

PPIO Agent Sandbox is not in the public benchmarks we use, so we cannot say which is faster. In StarSling's run of 3 October 2026 (run 37073680030), boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, second of 15. 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.

V8 Web Tooling Benchmark at a 4 vCPU / 8 GB target, runs per second, higher is better. Source: StarSling's run of 3 October 2026 (run 37073680030), median of 5 sandboxes. Providers ran on different hardware in different regions. StarSling gives a shared rank when the run cannot separate two providers. 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. PPIO Agent Sandbox is not in the public benchmarks we use, so we cannot say which is faster.

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. Start limits, per-start fees and plan minimums are in the price, boat's included.

Scenarioboat, a monthPPIO Agent Sandbox, 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$20 · Pay as you go
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$26 · $20/mo$85 · Pay as you go
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$1,020 · 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$290 · 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
$8,800 · $2000/mo · with limits$640 · Pay as you go · with limits

boat: Plans cost $20, $100, $500 or $2,000 a month, and the whole plan price comes back as VM time for that month. Included time expires at the end of the period. A default sandbox (4 vCPU / 8 GB) costs $0.036 an hour, billed per second while it runs, busy or idle. Each plan includes a start budget ($20: 12 a minute, 60 an hour, 200 a day); starts above it cost $0.03 to $0.07 each, and starts above 6 times the daily budget (3 times per minute) fail. The $20 plan goes up to the large size (8 vCPU / 16 GB). A stopped sandbox has no compute charge, but extra saved snapshots and the AI credentials proxy still charge. A card is required, also for the 7-day trial.

PPIO Agent Sandbox: PPIO bills per second for the vCPU and RAM you set, in CNY from a prepaid balance, with no monthly fee and no published start-rate limit. The 20 GB local disk is free. Templates, snapshots and paused sandboxes use persistent storage that is free up to 60 GB per account and then billed by size. PPIO publishes no Volume price and no egress price.

Limits in those scenarios

PPIO's long-running mode lifts the 1-hour cap; the docs show a 7-day example and state no maximum. The idle timeout still applies, and long-running mode must be set when you create the sandbox. PPIO allows 50 running sandboxes per user by default, so the fleet and burst scenarios need a quota raise through its sales team. Each sandbox has a 20 GB local disk; more persistent space comes from Volumes (Beijing region, beta) or an S3 mount. PPIO publishes no egress price. boat's auto-stop defaults to 1 hour; turn it off on a paid plan for no session cap. boat keeps a stopped sandbox's disk for at least 14 days. Memory and running programs are not saved, and a machine failure loses what was written after the last snapshot. At 1,000 short jobs at once, boat's $2,000 plan charges for starts above 90 a minute and refuses them above 270.

ScenarioboatPPIO Agent Sandbox
One coding agent, workdaysNone for this scenario
  • disk beyond 20 GiB: price unknown
  • egress price not published
One agent that never stopsNone for this scenario
  • disk beyond 20 GiB: price unknown
  • egress price not published
50 agents in parallelNone for this scenario
  • not confirmed: Docker inside
  • egress price not published
Sandboxes for your usersNone for this scenario
  • Pay as you go: max 50 concurrent
  • egress price not published
1,000 short jobs at once
  • 2,400 starts/day included, paid above that (needs ~6,667)
  • 600 starts/hour included, paid above that (needs ~1,000)
  • starts above 90/min are paid, and starts above 270/min fail with 429
  • Pay as you go: max 50 concurrent
  • egress price not published
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

Which one should you use?

Use PPIO Agent Sandbox when

your agents and users are in mainland China, or when you run short E2B-style jobs that pause with their memory between requests.

  • Memory pause, snapshots and wake on command. Useful for short jobs that sleep between requests: a paused sandbox keeps its running processes and wakes on the next command.
  • Close to users in mainland China. Its sandboxes run in Shanghai and Beijing, close to agents, users and services inside mainland China.
  • E2B API, shared storage and outbound rules. Code written for E2B can move over by changing a domain and key, sandboxes can share a Volume, and outbound traffic can be limited by IP range.

Use boat when

your agents do long work with Docker, emulators, builds and a browser on a full VM, with coding agents you drive by API, on machines in Germany, Finland and France.

  • A whole machine, not just a runtime. PPIO's docs do not describe Docker or /dev/kvm inside a sandbox and cap RAM at 8 GiB, so a compose stack or an emulator is not a documented path there.
  • Coding agents you drive from outside. PPIO has no agent API of its own. You drive its Codex template through the command API, or wire in OpenAI's or Kimi's hosted agent API. Its secrets proxy works only in the Beijing region.
  • Machines in Europe. Teams whose agents, users and services are in Europe get machines close to them. PPIO has no region outside mainland China.

Questions people ask

Is boat cheaper than PPIO Agent Sandbox?

It depends on the workload. For one agent on workdays both cost about $20 a month. boat costs less for an agent that runs all month ($26 vs $85) and for 50 agents in parallel ($317 vs $1,020). For sandboxes for your users, boat's estimate is $200 vs $290 within its included starts, and PPIO needs a quota raise above 50 at once. For 1,000 short jobs at once neither fits as listed: boat's start limits refuse part of the burst, and PPIO costs less ($640 vs $8,800) once it raises your quota.

Can I run Docker in a PPIO sandbox?

PPIO uses Docker images and Dockerfiles to build sandbox templates. Its docs do not describe running Docker inside a sandbox, and we have not tested it. boat runs Docker and compose in every sandbox and also exposes /dev/kvm for emulators and VMs.

How long can a PPIO sandbox run?

A new PPIO sandbox lives 5 minutes by default and at most 1 hour, unless you create it in long-running mode, which lifts that cap. PPIO's docs show a 7-day example and state no maximum, and the idle timeout still applies. boat's auto-stop defaults to 1 hour; on a paid plan you can set up to 30 days or turn it off for no cap.

Does PPIO Agent Sandbox work with the E2B SDK?

Yes. PPIO offers an E2B-compatible API, so the E2B SDK works after you set its domain and use a PPIO key. PPIO recommends its own SDK for every feature.

Where does each one run?

PPIO runs sandboxes in Shanghai and Beijing in mainland China, and you pick the region. boat runs sandboxes on machines in Germany, Finland and France; you cannot pick a region, and snapshot copies can be stored outside the EU.

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

More comparisons