Google Agent Runtime vs Unikraft Cloud
Google Agent Runtime hosts the agent code you build with ADK or LangGraph. Its sandboxes run as an unprivileged user in gVisor-isolated containers. Unikraft Cloud turns a Dockerfile into a minimal Firecracker microVM that boots in milliseconds, sleeps when traffic stops and is sold as flat pools of running capacity.
The companies and products
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. Each sandbox runs in a gVisor-isolated container in a Google-managed tenant project. Teams on GKE can also run the open-source GKE Agent Sandbox in their own cluster, with gVisor or Kata Containers, and manage it themselves.
Unikraft Cloud
Unikraft Cloud is a microVM cloud from Unikraft Inc. It runs any Dockerfile-based workload as a minimal Linux microVM, with cold starts it states as under 10 ms and scale-to-zero.
Your Dockerfile becomes the VM image at build time. Each instance runs on a customized Firecracker VMM and can go into standby when no traffic arrives. It is a general compute platform for dense, bursty work such as HTTP services, browsers and agent fleets. It sells flat self-serve plans, plus Enterprise dedicated hosts, on-prem and bring your own cloud. You reach instances through services, the CLI and a sandbox plugin for commands and files. For SSH or a desktop, you add them to your image with its official guides.
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. “34th cheapest of 361” means 33 of the other 360 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 workload | Google Agent Runtime: price rank | Unikraft Cloud: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Neither documents Docker inside a sandbox: Google's run as appuser with no sudo, and Unikraft builds your Dockerfile into the VM itself. Unikraft's Team pool has 8 GiB for all running instances. boat has root, Docker and systemd on every VM. | 310th of 365, with limits $5,801 a month | Not ranked: misses a requirement | 5th cheapest of 365 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB A Google sandbox is deleted when its TTL ends, 7 days at most for a custom container; a snapshot (Preview) carries it into a new sandbox. Unikraft's Team pool fits one 4 vCPU / 8 GB agent that runs with no published time limit. A boat VM has no session cap once auto-stop is off (default 1 hour). | 34th cheapest of 361 $7,519 a month | Not ranked: misses a requirement | 6th cheapest of 361 $657 a month |
| Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month Short code runs suit both: Google has a ready Code Execution sandbox, and Unikraft instances sleep between calls and wake in milliseconds, though self-serve pools cap running capacity. | 251st of 336, with limits $275 a month | Not ranked: misses a requirement | 189th of 336, 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 Google reaches a sandbox through its API or a private endpoint in your VPC, with no public URL documented. Unikraft serves HTTP and HTTPS through its load balancer, but hundreds of active users need Enterprise. boat serves token-gated HTTPS on up to 50 ports per sandbox. | 275th of 349, with limits $1,916 a month | Not ranked: misses a requirement | 3rd cheapest of 349 $209 a month |
| RL rollouts 1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month | 132nd of 349, with limits $8,240 a month | Not ranked: misses a requirement | 39th of 349, with limits $2,000 a month |
| Agent evals 250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month | 90th cheapest of 359 $2,060 a month | Not ranked: misses a requirement | 10th cheapest of 359 $200 a month |
| CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month Jobs that build Docker images need a container runtime that neither documents inside a sandbox, and Unikraft offers nested virtualization on Enterprise only. boat has Docker and /dev/kvm on every plan. | 104th cheapest of 374 $989 a month | Not ranked: misses a requirement | 2nd cheapest of 374 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB Google documents no SSH. Unikraft supports SSH if you run sshd in your image and tunnel to it, and its Team pool comes to $39 a month against $241 on Google. boat has SSH built in and a desktop, for $26 a month with auto-stop off. | 317th of 371, with limits $241 a month | 121st cheapest of 371 $39 a month | 19th cheapest of 371 $26 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Google gives agents a browser with a VNC live view and a click and type API. Unikraft has an official VNC/noVNC desktop recipe that you build into your image. boat streams a built-in 60fps desktop with Chrome. | 85th of 367, with limits $1,648 a month | Not ranked: misses a requirement | 7th cheapest of 367 $200 a month |
| Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month Google runs a Chromium you can drive over CDP with Playwright. On Unikraft you package your own headless browser in a Dockerfile, and it sleeps between requests. boat has Chrome on a desktop out of the box. | 132nd of 356, with limits $824 a month | Not ranked: misses a requirement | 7th cheapest of 356 $200 a month |
| Scraping & bots (startup scale) 50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month | 56th cheapest of 343 $515 a month | Not ranked: misses a requirement | 6th cheapest of 343 $90 a month |
| Mass scraping & bots (undetectable) 1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month | 212th of 327, with limits $8,240 a month | Not ranked: misses a requirement | 73rd of 327, with limits $1,800 a month |
| Parallel red teaming 200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month | 281st of 351, with limits $412 a month | Not ranked: misses a requirement | 3rd cheapest of 351 $200 a month |
| GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month | Not ranked: misses a requirement | Not ranked: misses a requirement | Not 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 requirement | Not ranked: misses a requirement | Not ranked: misses a requirement |
Key features
| Feature | Google Agent Runtime | Unikraft Cloud | boat |
|---|---|---|---|
| Isolation | gVisor-isolated container in a Google-managed tenant project | Minimal Linux microVM per instance on a customized Firecracker VMM; full VMs on Enterprise | Full VM with its own kernel, on hosts dedicated to boat (AMD Ryzen 7 9700X and Ryzen 9 9950X); vCPUs are shared |
| Root and Docker inside | Runs as appuser with no sudo; images must not need root; Docker inside is not documented | Runs as root; the Dockerfile becomes the VM image, with no container runtime inside documented | Root, systemd and Docker, like any Ubuntu machine |
| Nested virtualization | /dev/kvm not documented | Per-instance flag and permission, Enterprise only | /dev/kvm for Android emulators and VMs on every plan |
| Shell access | Commands run through API calls; SSH is not documented | With setup: run sshd in your image and reach it through a TLS service port plus a local tunnel; commands and files also through the sandbox plugin and CLI | SSH built into every sandbox, plus an API for commands |
| Desktop and browser | A browser in a container with a VNC live view, a click, type and screenshot API, and CDP | No prebuilt desktop; an official VNC/noVNC desktop recipe you build into your image | Linux desktop and Chrome streamed at 60fps, with computer tools for agents |
| Idle and wake | A paused sandbox keeps its files at a lower, unpublished rate; no wake on request documented | Standby after a 1 second idle cooldown by default, memory kept in stateful mode, woken by the next request in milliseconds | Auto-stop on a timer, not on idle; a stopped sandbox has no compute charge and boots again from its disk when started by API |
| How long a sandbox lives | TTL of up to 7 days for a custom container and 14 days for Code Execution, then deleted; a snapshot (Preview) restores it into a new sandbox | No published maximum; instances run until stopped or go into standby | Auto-stop defaults to 1 hour; set up to 30 days, or turn it off on a paid plan for no session cap |
| Memory snapshots and branching | Snapshots (Preview) keep memory and files; start many sandboxes from one snapshot | Checkpoint memory and volumes of a running microVM with a pause of a few milliseconds, then start new instances from it | Fork a running sandbox from its last 1-minute snapshot; memory is not saved, so a resume is like a reboot |
| Inbound access | Through Google's API, or a private endpoint in your own VPC; a public URL is not documented | Services on its load balancer: HTTP on 80, HTTPS on 443, TLS-wrapped TCP on other ports, custom domains | Up to 50 HTTPS URLs, token-gated by default; raw ports over IPv6 only on sandboxes with no IPv4 of their own |
| Outbound rules | Off unless the template turns it on; route through your VPC with Cloud NAT, a static IP, Cloud NGFW or Secure Web Proxy | Network shield filters traffic, Enterprise only, limited access | Open internet on every plan; no provider-side allowlist |
| Keeping API keys out of the sandbox | You provision and scope the credentials the agent uses | Network shield injects secrets into HTTP traffic, Enterprise only, limited access | Optional paid credentials proxy ($1 a day plus $1 per 1,000 requests) on every plan: real keys of supported model providers stay out; other secrets still enter |
| Coding agents and agent hosting | Agent Runtime hosts ADK, LangGraph, LangChain, AG2 and LlamaIndex agents; Managed Agents API (Preview) runs Google's Antigravity harness | None installed; you add the agent to your own image | Seven harnesses installed, driven from outside with a prompt, steer and events API |
| Machine size | Runtime: 1 to 8 vCPU and up to 32 GiB; sandbox maximum not published | Plan pool shared by all running instances: 2 vCPU / 4 GiB, 8 vCPU / 8 GiB or 24 vCPU / 32 GiB; per-instance maximum not published | Up to 8 vCPU / 16 GB on the $20 plan, 16 vCPU / 32 GB per sandbox from the $100 plan |
| Regions | 23 locations across the US, Europe, the Middle East, Asia, Canada, Australia and Brazil | Dallas, Frankfurt, San Francisco, Singapore and Washington DC, more on request | Germany, Finland and France; you cannot pick a region; no US or Asia region |
Every feature Battleships tracks (43)
| Feature | Google Agent Runtime | Unikraft Cloud | boat |
|---|---|---|---|
| Full VM (own kernel) | No documented way | Yes | Yes |
| Docker inside | No documented way | No documented way | Yes |
| Nested virtualization | No documented way | Yes | Yes |
| Root | No documented way | Yes | Yes |
| Systemd | Not published | Not published | Yes |
| Desktop GUI | Not published | With your own setup | Yes |
| Browser | Yes | With your own setup | Yes |
| Computer-use API | Yes | No documented way | Yes |
| SSH | No documented way | With your own setup | Yes |
| Persistent disk | Yes | Yes | Yes |
| Pause/resume | Yes | Yes | Yes |
| Automatic snapshots | Yes | Yes | Yes |
| Fork/clone | Yes | Yes | Yes |
| Live fork (no pause) | Yes | Yes | Yes |
| Memory fork | Yes | Yes | No documented way |
| Custom image (Docker or snapshot) | Yes | Yes | Yes |
| ≥24 h sessions | Yes | Yes | Yes |
| Idle auto-stop | No documented way | Yes | No documented way |
| Wake on request | No documented way | Yes | No documented way |
| HTTPS preview URLs | No documented way | Yes | Yes |
| Raw TCP inbound | Not published | Not published | Yes |
| Public IPv4 | No documented way | No documented way | No documented way |
| Egress allowlist | Yes | Yes | No documented way |
| Secret proxy | No documented way | Yes | Yes |
| Secret proxy for any API | No documented way | Yes | Yes |
| Preinstalled agents | Yes | No documented way | Yes |
| Agent harness API | Yes | No documented way | Yes |
| MCP server | No documented way | No documented way | No documented way |
| GPU | No documented way | Yes | No documented way |
| Windows | No documented way | No documented way | No documented way |
| Arm64 | Yes | Yes | No documented way |
| Self-hosting | No documented way | Yes | No documented way |
| BYOC | No documented way | Yes | No documented way |
| Open source | No documented way | No documented way | No documented way |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | Not published | Not published | Yes |
| Usage per sandbox (API) | Yes | No documented way | Yes |
| Audit logs | Not published | Yes | No documented way |
| SSO | Yes | No documented way | No documented way |
| SOC 2 | Yes | Yes | No documented way |
| HIPAA | Yes | Yes | No documented way |
| EU data residency | Yes | Yes | No documented way |
| Zero data retention | Not published | Not published | Yes |
Performance
Neither Google Agent Runtime nor Unikraft Cloud is in the public benchmarks we use, so we cannot say which is faster. Unikraft states cold starts under 10 ms; that is its own figure for start time, not CPU speed. For reference, boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, second of 15 in StarSling's run of 3 October 2026 (run 37073680030). 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. Google Agent Runtime and Unikraft Cloud are 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.
| Scenario | Google Agent Runtime, a month | Unikraft Cloud, a month | boat, 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 | $73 · Pay as you go | $39 · Hobby | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $301 · Pay as you go | $39 · Hobby | $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 | $3,626 · Pay as you go · with limits | Not ranked: misses a requirement | $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 | $1,030 · Pay as you go | Not ranked: misses a requirement | $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 | $2,060 · Pay as you go | Not ranked: misses a requirement | $8,800 · $2000/mo · with limits |
Google Agent Runtime: Google bills $0.085 per vCPU-hour and $0.009 per GiB-hour allocated, rounded to the second. The pricing page lists a monthly free amount but does not say it covers sandboxes, so our totals do not subtract it. Sandboxes are billed while they exist. A paused one costs less at an unpublished rate, and disk, snapshot and egress rates are not published.
Unikraft Cloud: Unikraft charges a flat monthly fee for a pool of running capacity: free for 2 vCPU / 4 GiB, $39 for 8 vCPU / 8 GiB, $199 for 24 vCPU / 32 GiB. Standby instances do not count against the pool. It publishes no per-hour, disk or egress price, so anything bigger is an Enterprise quote.
Unikraft Cloud: Plans cost $20, $100, $500 or $2,000 a month. The whole plan price comes back as VM time for that month, and it does not carry over. A 4 vCPU / 8 GB sandbox costs $0.036 an hour, billed per second while it runs, busy or idle. The $20 plan includes 200 starts a day; extra starts cost $0.03 to $0.07. A stopped sandbox has no compute charge; saved snapshots above the plan's count and the AI credentials proxy still charge. A card is required, also for the 7-day trial.
Limits in those scenarios
On Google, a sandbox is deleted when its TTL ends unless you snapshot it, and the 50-agent fleet does not fit because Google documents no Docker inside its sandboxes. Unikraft's limits are shared pools: the Team pool of 8 vCPU and 8 GiB is filled by one 4 vCPU / 8 GB agent, so the fleet, user-sandbox and burst scenarios need an Enterprise contract. Neither publishes disk or egress prices. On boat the 1,000-job burst does not fit: the $2,000 plan includes 90 starts a minute and 2,400 a day, charges for extra starts and refuses starts above 270 a minute. boat's always-on agent needs auto-stop turned off (default 1 hour).
| Scenario | Google Agent Runtime | Unikraft Cloud | boat |
|---|---|---|---|
| One coding agent, workdays |
|
| None for this scenario |
| One agent that never stops |
|
| None for this scenario |
| 50 agents in parallel |
|
| None for this scenario |
| Sandboxes for your users |
|
| None for this scenario |
| 1,000 short jobs at once |
|
|
|
Which one should you use?
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, with short code runs or large bursts in a gVisor sandbox.
- A managed home for your agent code. Useful if you build agents with ADK or LangGraph: Agent Runtime hosts them and skips billing for idle time between turns.
- Inside your Google Cloud network. Useful if agents must stay in a Google Cloud perimeter, with sandbox egress off by default and routed through your own firewall.
- Many regions, ready browser, large bursts. Useful if you need 23 locations worldwide, a browser sandbox with a live view and a click and type API, or 1,000 short jobs at once.
Use Unikraft Cloud when
you run many small services or headless browsers that sleep most of the time and must wake in milliseconds, want a free plan or flat prices, or want GPUs, on-prem or your own cloud under an Enterprise contract.
- Millisecond starts and scale-to-zero. Best for many small services that sleep most of the time and must answer the next request in milliseconds.
- Checkpoints and a flat monthly pool. Useful if you want root in a microVM, checkpoints of a running instance, and a known monthly bill, with a free Hobby plan to start.
- GPUs and your own hardware. Enterprise customers can get NVIDIA GPUs, arm64 hosts, dedicated hosts, on-prem or their own cloud.
Where boat fits
Google Agent Runtime is built to host agent code, and Unikraft Cloud is built for small services that sleep and wake fast. If you run agents that need a whole computer to work in, with Docker, SSH, a desktop and public URLs, look at boat. Each boat sandbox is a full VM that saves its disk every minute, with coding agents installed. boat has no memory snapshots, no wake on request, no US region, and caps starts per plan.
Root, Docker and /dev/kvm
Google runs commands as appuser with no sudo, and Unikraft builds your Dockerfile into the VM with no container runtime inside documented. On both you bake every tool into an image up front, and Unikraft's nested virtualization needs Enterprise. boat has all three on every plan.
SSH and public URLs with no setup
Google documents no SSH or public URL, so you set up a VPC endpoint or tunnel yourself. On Unikraft, SSH needs sshd in your image and a tunnel, and ports go through its load balancer. boat has SSH built in and token-gated HTTPS on up to 50 ports.
Coding agents and keys
Google's Managed Agents API runs its Antigravity harness, in Preview. Unikraft installs no agents, and its secret injection is Enterprise only. boat ships seven harnesses you drive by API, and its optional paid credentials proxy keeps real keys of supported providers out.
Use boat when your agents need a real machine to work in: root, Docker, SSH, a desktop and public HTTPS URLs on a full VM that saves its disk every minute, with coding agents you drive from outside. Compare it one to one: boat vs Google Agent Runtime and boat vs Unikraft Cloud.
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
Questions people ask
Is Google Agent Runtime cheaper than Unikraft Cloud?
For one agent, no: one coding agent on workdays comes to $73 a month on Google and $39 on Unikraft's Team pool, and one agent running all month to $301 against $39. boat comes to $20 and $26 for the same two. For 50 agents, sandboxes for your users or 1,000 short jobs, no self-serve Unikraft plan is big enough, so the price is an Enterprise quote. Google fits 300 user sandboxes at $1,030 and the 1,000-job burst at $2,060.
Which is faster, Google Agent Runtime or Unikraft Cloud?
Neither is in the public benchmarks we use, so we cannot say which computes faster. Unikraft is built to start fast: it states cold boots under 10 ms and wakes on the next request. For reference, boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB in StarSling's run of 3 October 2026, on an AMD Ryzen 9 9950X host.
Can Google Agent Runtime or Unikraft Cloud run Docker?
Neither documents it inside a sandbox. Google's sandboxes run as appuser with no sudo; you build images outside and push them to Artifact Registry. Unikraft uses your Dockerfile to build the VM's filesystem and documents no container runtime inside it. A boat sandbox is a full Ubuntu VM with Docker, systemd and /dev/kvm.
Can I SSH into Google Agent Runtime or Unikraft Cloud?
Google documents no SSH; you run commands through its API. Unikraft supports SSH with setup: you run sshd in your image and reach it through a TLS service port and a local tunnel, as its Debian SSH guide shows. boat has SSH built into every sandbox.
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. The code, shell and browser sandboxes are a separate part of the same platform called Agent Sandbox.
Where do they run?
Google runs in 23 locations across the US, Europe, the Middle East, Asia, Canada, Australia and Brazil. Unikraft runs in Dallas, Frankfurt, San Francisco, Singapore and Washington DC by default. boat runs on machines in Germany, Finland and France; you cannot pick a region, and snapshot copies can be stored outside the EU.
Try boat
Install the CLI, then run boat new.
7-day free trial
More comparisons
- boat vs Google Agent Runtime
- boat vs Unikraft Cloud
- E2B vs Google Agent Runtime
- Daytona vs Google Agent Runtime
- Modal vs Google Agent Runtime
- Vercel Sandbox vs Google Agent Runtime
- Cloudflare Sandbox vs Google Agent Runtime
- Blaxel vs Google Agent Runtime
- Runloop vs Google Agent Runtime
- Tensorlake vs Google Agent Runtime
- Morph Cloud vs Google Agent Runtime
- CodeSandbox SDK vs Google Agent Runtime