Deno Sandbox vs Google Agent Runtime

Deno Sandbox gives code a small Firecracker microVM with root that starts in under 200 ms and lives up to 30 minutes. Google Agent Runtime hosts your agent code on Google Cloud and gives it sandboxes that run as an unprivileged user for up to 7 days.

The companies and products

Deno Sandbox

Deno Sandbox is part of Deno Deploy, the hosting platform from the company behind the Deno runtime, and runs untrusted or AI-generated code in Linux microVMs.

Each sandbox is a Firecracker microVM with 2 vCPU, up to 4 GB of memory and a 10 GB disk that is wiped when it ends, created from a TypeScript or Python SDK. Outbound traffic goes through a proxy that can enforce a host allowlist and swap placeholder secrets for real ones. It is built for short runs of agent or user code, and a sandbox can be promoted to a Deno Deploy app.

Google Agent Runtime

Google Agent Runtime is the agent hosting part of Gemini Enterprise Agent Platform on Google Cloud, renamed from Vertex AI Agent Engine in April 2026.

It is a hyperscaler service, not a sandbox company. Agent Runtime deploys agents you build with ADK, LangGraph, LangChain, AG2 or LlamaIndex and bills them only while they work on a turn. Next to it, Agent Sandbox gives those agents a place to run code: a Code Execution sandbox, a shell or custom container sandbox, and a Computer Use browser. All of them run as secure containers inside your Google Cloud project.

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. “290th cheapest of 367” means 289 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 workloadDeno Sandbox: price rankGoogle Agent Runtime: price rankboat: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
A software factory needs hours, root and Docker: Deno stops after 30 minutes at 2 vCPU, Google has no root or Docker, and boat has all of it on a full VM.
290th of 367, with limits
$5,665 a month
295th of 367, with limits
$5,801 a month
3rd cheapest of 367
$634 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
Google keeps a sandbox for up to 7 days, which beats Deno's 30 minutes for a long-lived assistant, while boat has no session cap on paid plans.
287th of 366, with limits
$5,803 a month
28th cheapest of 366
$7,519 a month
5th cheapest of 366
$657 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
Both fit short code runs: Deno starts in under 200 ms and bills CPU only while code runs, and Google's Code Execution sandbox is a ready Python interpreter.
141st of 345, with limits
$329 a month
254th of 345, with limits
$275 a month
186th of 345, with limits
$2,000 a month
Sandboxes for your users
300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month
Deno gives each exposed port a public HTTPS URL but caps Builder at 20 sandboxes at once per region; Google has no public URL, and boat serves token-protected HTTPS on up to 50 ports.
148th of 361, with limits
$1,863 a month
280th of 361, with limits
$1,916 a month
Cheapest of 361
$209 a month
RL rollouts
1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month
122nd of 360, with limits
$7,963 a month
128th of 360, with limits
$8,240 a month
34th of 360, with limits
$2,000 a month
Agent evals
250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month
296th of 364, with limits
$1,563 a month
79th cheapest of 364
$2,060 a month
10th cheapest of 364
$200 a month
CI runners
40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month
CI jobs that build containers need Docker, which Google's sandboxes do not offer and Deno does not document; boat runs Docker and systemd on every VM.
304th of 377, with limits
$731 a month
98th cheapest of 377
$989 a month
4th cheapest of 377
$86 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
Deno has SSH but a 30-minute lifetime, Google has no SSH at all, and boat gives you SSH, a desktop and a disk that stays.
179th of 368, with limits
$200 a month
311th of 368, with limits
$241 a month
18th cheapest of 368
$26 a month
Computer-use agents
50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month
Google offers a browser with a VNC view and a click and type API, Deno has no desktop, and boat streams a full Linux desktop at 60fps.
280th of 369, with limits
$1,583 a month
283rd of 369, with limits
$1,648 a month
6th cheapest of 369
$200 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
Google ships a real Chromium you can drive over CDP, while on Deno you would install a headless browser yourself in 4 GB of memory.
284th of 367, with limits
$783 a month
102nd of 367, with limits
$824 a month
7th cheapest of 367
$200 a month
Scraping & bots (startup scale)
50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month
286th of 346, with limits
$573 a month
53rd cheapest of 346
$515 a month
6th cheapest of 346
$90 a month
Mass scraping & bots (undetectable)
1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month
247th of 338, with limits
$11,923 a month
210th of 338, with limits
$8,240 a month
71st of 338, with limits
$1,800 a month
Parallel red teaming
200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month
251st of 360, with limits
$367 a month
260th of 360, with limits
$412 a month
2nd cheapest of 360
$200 a month
GPU eval jobs
10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month
Can't run itCan't run itCan't run it
macOS agents
5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month
Can't run itCan't run itCan't run it

Key features

FeatureDeno SandboxGoogle Agent Runtimeboat
IsolationFirecracker microVM with its own kernelSecure container sandboxing inside your Google Cloud projectFull VM with its own kernel on dedicated AMD Ryzen 9 9950X hosts
Root and DockerRoot access; Docker and nested virtualization not documentedRuns as appuser with no sudo; images must not need root; no DockerRoot, systemd, Docker and /dev/kvm on every machine
Machine size2 vCPU fixed, 768 MB to 4 GB memory, 10 GB diskRuntime: 1 to 8 vCPU and up to 32 GiB; sandbox maximum not published4 vCPU / 8 GB by default, up to 16 vCPU / 32 GB from the $100 plan
How long a sandbox livesUp to 30 minutesTTL of up to 7 days for a custom container, 14 days for Code ExecutionNo session cap on paid plans
Files after it endsRoot disk wiped; opt-in volumes of 300 MB to 20 GB, Chicago onlyPause keeps the file system; the sandbox is deleted when its TTL endsWhole disk saved every minute and on stop; a stopped sandbox costs nothing
Snapshots and branchingRead-only volume snapshots that new sandboxes can boot fromSnapshots keep memory and files; start many sandboxes from oneFork a running sandbox with its disk; resume reboots from disk
Shell accessSSH and a browser VS Code editorexecute_bash() API calls, each in a new shell; no SSHSSH, plus an API for commands
Reaching what it servesPublic HTTPS URL per exposed port, without authenticationGoogle's API or a private endpoint in your VPC; no public URLHTTPS on up to 50 ports, token-protected by default; raw TCP
Desktop and browserNo desktop or browserBrowser in a container with a VNC view, click and type API and CDPLinux desktop streamed at 60fps with Chromium and computer use
Outbound rulesHost allowlist per sandbox, enforced by a proxy outside the VMOff unless the template turns it on; route through your VPC firewallOpen internet, no provider-side allowlist
Keeping API keys outPlaceholder secrets swapped for real ones on approved hostsYou provision and scope the credentials the agent usesCredentials proxy for model providers and any HTTPS API; switch accounts when credits run out
Coding agentsRoot image with Claude Code, run through shell commandsManaged Agents API (Preview) runs Google's Antigravity harnessSeven agents preinstalled, with a prompt, steer and events API
Hosting agent codeA sandbox can be promoted to a Deno Deploy appDeploys ADK, LangGraph and other agents; idle time between turns not billedYou run your own agent loop inside a sandbox
Start time and regionsUnder 200 ms; Amsterdam and ChicagoStart time not published; 23 locations worldwideBoots a full VM; EU only (Germany, Finland, France)
Every feature Battleships tracks (43)
FeatureDeno SandboxGoogle Agent Runtimeboat
Full VM (own kernel)YesNoYes
Docker insideNot publishedNoYes
Nested virtualizationNoNoYes
RootYesNoYes
SystemdNot publishedNot publishedYes
Desktop GUINoNoYes
BrowserNoYesYes
Computer-use APINoYesYes
SSHYesNoYes
Persistent diskYesYesYes
Pause/resumeNoYesYes
Automatic snapshotsNoYesYes
Fork/cloneNoYesYes
Live fork (no pause)NoYesYes
Memory forkNoYesNo
Custom image (Docker or snapshot)YesYesYes
≥24 h sessionsNoYesYes
Idle auto-stopNoNoNo
Wake on requestNoNoNo
HTTPS preview URLsYesNoYes
Raw TCP inboundNoNot publishedYes
Public IPv4NoNoNo
Egress allowlistYesYesNo
Secret proxyYesNoYes
Secret proxy for any APIYesNoYes
Preinstalled agentsYesYesYes
Agent harness APINoYesYes
MCP serverNoNoNo
GPUNoNoNo
WindowsNoNoNo
Arm64NoYesNo
Self-hostingNoNoNo
BYOCNoNoNo
Open sourceNoNoNo
Scoped API keysNot publishedNot publishedYes
Spend limitsYesNot publishedYes
Usage per sandbox (API)NoYesYes
Audit logsNot publishedNot publishedNo
SSONoYesNo
SOC 2YesYesNo
HIPAAYesYesNo
EU data residencyYesYesYes
Zero data retentionNot publishedNot publishedYes

Performance

Deno SandboxNot measured
Google Agent RuntimeNot measured
boat26.69 runs/s

There is no comparable measurement for Deno Sandbox or Google Agent Runtime. boat did 26.69 runs per second on StarSling's Node.js build-and-test benchmark at 4 vCPU / 8 GB; neither of the other two was in that run, and Deno's fixed 2 vCPU shape could not run that test as set up.

Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Deno Sandbox and Google Agent Runtime have 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.

ScenarioDeno Sandbox, a monthGoogle Agent Runtime, a monthboat, 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
$53 · Pro · with limits$73 · Pay as you go$20 · $20/mo
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$200 · Builder · with limits$301 · Pay as you go$26 · $20/mo
50 agents in parallel
50 machines of 4 vCPU / 8 GB at once, 8-hour sessions, 1,100 sessions a month, Docker inside
$2,779 · Builder · with limits$3,626 · Pay as you go · with limits$317 · $20/mo × 3 seats
Sandboxes for your users
2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once
$783 · Builder · with limits$1,030 · Pay as you go$200 · $100/mo × 2 seats
1,000 short jobs at once
2 vCPU / 4 GB, 1,000 at once, 3 minutes each, 200,000 jobs a month
$1,563 · Builder · with limits$2,060 · Pay as you go$2,000 · $2000/mo · with limits

Deno Sandbox: Sandboxes need Deno Deploy Pro at $20 a month or Builder at $200 a month, with CPU and memory allowances shared with Deploy apps. Past those, CPU costs $0.10 per hour of active CPU time and memory $0.025 per GiB-hour for the whole life of the sandbox.

Google Agent Runtime: Google bills $0.085 per vCPU-hour and $0.009 per GiB-hour allocated, with the first 50 vCPU-hours and 100 GiB-hours free each month. Sandboxes bill while they exist, Runtime skips idle time between turns, and disk and egress rates are not published.

Google Agent Runtime: Plans start at $20 a month, and the whole fee is usage credit spent per second while a sandbox runs. A stopped sandbox costs nothing and keeps its disk.

Limits in those scenarios

Deno Sandbox fits none of the five scenarios as written: it lives up to 30 minutes at a fixed 2 vCPU, and Builder allows 20 sandboxes at once per region. Google fits four of them but leaves disk and egress unpriced, and it cannot run the 50-agent fleet because its sandboxes have no Docker. boat fits four of five; at 1,000 short jobs at once its starts are capped at 90 a minute and 2,400 a day, so the jobs queue.

ScenarioDeno SandboxGoogle Agent Runtimeboat
One coding agent, workdays
  • Pro: sessions capped at 0.5 h
  • capped at 2 vCPU / 4 GiB
  • largest size is 2 vCPU / 4 GiB
  • disk beyond 10 GiB: price unknown
  • disk beyond 0 GiB: price unknown
  • egress price not published
None that bite
One agent that never stops
  • Builder: sessions capped at 0.5 h
  • capped at 2 vCPU / 4 GiB
  • largest size is 2 vCPU / 4 GiB
  • disk beyond 10 GiB: price unknown
  • disk beyond 0 GiB: price unknown
  • egress price not published
None that bite
50 agents in parallel
  • Builder: max 20 concurrent
  • Builder: sessions capped at 0.5 h
  • capped at 2 vCPU / 4 GiB
  • largest size is 2 vCPU / 4 GiB
  • missing Docker inside
  • disk beyond 0 GiB: price unknown
  • egress price not published
None that bite
Sandboxes for your users
  • Builder: max 20 concurrent
  • disk beyond 0 GiB: price unknown
  • egress price not published
None that bite
1,000 short jobs at once
  • Builder: max 20 concurrent
  • disk beyond 0 GiB: price unknown
  • egress price not published
  • max 2,400 starts/day (needs ~6,667)
  • max 600 starts/hour (needs ~1,000)
  • bursts above 90 starts/min are queued (a burst of 333 takes ~4 min to start)

Which one should you use?

Use Deno Sandbox when

you run short pieces of untrusted code that must start in milliseconds, reach only hosts you allow, and maybe go live on Deno Deploy.

  • Fast starts, billed by active CPU. Good for short bursts of code: a sandbox starts in under 200 ms and CPU bills only while code runs.
  • Root and a public URL. Useful when code needs root or a quick preview: Deno gives root, SSH and a public HTTPS URL per port.
  • Secrets kept out by a proxy. Good for untrusted code: a proxy outside the VM allows only listed hosts and swaps in real secrets there.

Use Google Agent Runtime when

you build agents with ADK or LangGraph on Google Cloud and want them hosted inside your own project, network rules and regions.

  • A managed home for agent code. Useful if you build with ADK or LangGraph: Agent Runtime hosts your agents and skips idle time between turns.
  • Sandboxes that last for days. Fits longer tasks than Deno allows: a sandbox can live up to 7 days and snapshot its memory to branch.
  • Inside your Google Cloud network. Useful if agents must stay in your Google Cloud perimeter, with egress off by default, in 23 locations.

Where boat fits

Deno Sandbox is built for short runs and Google Agent Runtime for agents that live inside Google Cloud. If your agents do real work for hours, with installs, Docker, a desktop and something to show people, boat gives each one a full VM that keeps its disk.

Root, Docker and /dev/kvm

Google sandboxes run without sudo or Docker, and Deno does not document Docker. An agent that runs apt install, docker compose or an Android emulator has to work around that or run elsewhere; boat has root, Docker, systemd and /dev/kvm.

A machine that stays

A Deno sandbox ends after 30 minutes and wipes its root disk, and a Google sandbox is deleted when its 7-day TTL runs out. boat has no session cap on paid plans, saves the disk every minute and can fork a running sandbox.

Agents, desktop and URLs built in

Deno has no desktop and Google has only a browser and no public URL. boat ships Claude Code, Codex and other agents you drive from outside, a 60fps desktop, and HTTPS on 50 ports, so people can open what agents build.

Use boat when your agents need a real computer for hours or weeks: root, Docker, SSH, a desktop and public URLs on a full VM that keeps its disk, with coding agents built in. Compare it one to one: boat vs Deno Sandbox and boat vs Google Agent Runtime.

Questions people ask

Is Deno Sandbox cheaper than Google Agent Runtime?

On paper, yes, in every scenario we priced: one agent on workdays comes to $53 a month on Deno and $73 on Google, and sandboxes for your users $783 against $1,030. But Deno does not fit any of those scenarios as written, because of its 30-minute lifetime, 2 vCPU size and 20-at-once cap. boat comes to $20 and $200 for the same two.

Which is faster?

Neither Deno Sandbox nor Google Agent Runtime has a comparable CPU measurement. Deno says a sandbox starts in under 200 ms. boat did 26.69 runs per second on StarSling's Node.js benchmark at 4 vCPU / 8 GB.

Can Deno Sandbox or Google Agent Runtime run Docker?

Google's sandboxes cannot: commands run as appuser with no sudo, and images must not need root. Deno gives root but its docs do not say Docker or nested virtualization works. boat runs Docker and systemd and exposes /dev/kvm on every VM.

How long can a sandbox run on each?

A Deno sandbox lives up to 30 minutes and wipes its root disk when it ends. A Google sandbox can live up to 7 days for a custom container or 14 days for Code Execution. A boat sandbox has no session cap on paid plans.

Is Google Agent Runtime the same as Vertex AI Agent Engine?

Yes. Google renamed Vertex AI Agent Engine to Agent Runtime in April 2026, as part of Gemini Enterprise Agent Platform. Its code, shell and browser sandboxes are a separate part called Agent Sandbox.

Can either serve a web app to the public?

Deno gives each exposed port a public HTTPS URL, without authentication. Google's docs describe only its API or a private endpoint in your VPC. boat serves HTTPS on up to 50 ports, token-protected by default, and accepts raw TCP.

Try boat

Install the CLI, then run boat new.

7-day free trial

More comparisons