Google Compute Engine vs Oracle Cloud
Google Compute Engine and Oracle Cloud both rent plain VMs that you set up yourself, so the choice comes down to details. Compute Engine has full-speed shapes in many families and can suspend a VM with its memory, while Oracle Cloud has a free Arm allowance, Mac servers and very low prices on burstable VMs that are throttled under steady load.
The companies and products
Google Compute Engine
Google Compute Engine is Google Cloud's general-purpose VM service, not a sandbox product: you rent virtual machines by the second and set them up yourself.
You pick a machine type (E2, N2D, C3D, Arm T2A, GPU families and more), a boot image and a network, and Compute Engine boots a KVM-based VM with its own kernel and root access. Disks are Persistent Disk or Hyperdisk volumes, traffic goes through VPC firewall rules, and anything agent-specific (agents, desktops, HTTPS URLs, key handling) is yours to build. It is built for every kind of server workload, from websites to GPU training, across Google Cloud's worldwide regions.
Oracle Cloud
Oracle Cloud Infrastructure (OCI) is Oracle's general-purpose cloud: compute, storage, networking, databases and AI services in regions around the world.
OCI Compute rents virtual machines and bare metal servers on x86 (AMD, Intel) and Arm (Ampere), plus GPU machines and Mac servers. You pick a shape, an OS image, a network and a disk, and then install and run everything inside the VM yourself. It is not a sandbox product, so this page compares the part that overlaps: running agent code on a VM.
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. “16th cheapest of 364” means 15 of the other 363 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 Compute Engine: price rank | Oracle Cloud: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month | 120th of 367, with limits $1,106 a month | 58th of 367, with limits $413 a month | 3rd cheapest of 367 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB | 177th of 366, with limits $1,433 a month | 100th of 366, with limits $376 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 Short code runs suit both clouds better than boat, whose start limits push it to 186th; Oracle Cloud ranks 51st and Compute Engine 92nd, each paying a full VM boot per run. | 92nd of 345, with limits $133 a month | 51st of 345, with limits $16 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 boat ranks 1st for sandboxes for your users because each one gets an HTTPS URL by API; on Compute Engine or Oracle Cloud you build that layer yourself. | 92nd of 361, with limits $836 a month | 31st of 361, with limits $115 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 Oracle Cloud ranks 7th for RL rollouts on price, while boat is 34th and Compute Engine 36th; both clouds need raised quotas for a big fleet. | 36th of 360, with limits $2,085 a month | 7th of 360, with limits $679 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 Oracle Cloud ranks 1st for agent evals and Compute Engine 16th; both fit, and boat sits between them at 10th. | 16th cheapest of 364 $263 a month | Cheapest of 364 $136 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 boat ranks 4th for CI runners with /dev/kvm on every machine, Oracle Cloud ranks 6th and Compute Engine 41st. | 41st cheapest of 377 $359 a month | 6th cheapest of 377 $102 a month | 4th cheapest of 377 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB Oracle Cloud ranks 2nd for a 24/7 dev box on list price, ahead of boat at 18th and Compute Engine at 82nd, though its cheap VMs are burstable. | 82nd cheapest of 368 $109 a month | 2nd cheapest of 368 $12 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 Neither cloud has a desktop stream on its VMs, so a computer-use agent needs one installed and exposed by hand; boat ranks 6th with a 60fps desktop built in. | 129th of 369, with limits $293 a month | 75th of 369, with limits $97 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 | 212th of 367, with limits $244 a month | 154th of 367, with limits $49 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 | 218th of 346, with limits $179 a month | 150th of 346, with limits $32 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 | 108th of 338, with limits $2,978 a month | 35th of 338, with limits $642 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 | 152nd of 360, with limits $120 a month | 102nd of 360, with limits $26 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 it | Can't run it | Can't run it |
| macOS agents 5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month | Can't run it | Can't run it | Can't run it |
Key features
| Feature | Google Compute Engine | Oracle Cloud | boat |
|---|---|---|---|
| What you get | A general cloud VM; you install and wire everything else | A plain cloud VM or bare metal server that you set up yourself | An agent sandbox: a full VM with agents, desktop, URLs and snapshots set up |
| Isolation | Every instance is a full KVM-based VM with its own kernel | Every VM has its own kernel on a KVM hypervisor; bare metal also available | Every sandbox is a full VM with its own kernel |
| Docker and nested virtualization | Docker, root and systemd on any VM; nested KVM only on some Intel families and N4D, not on E2, Arm or other AMD VMs | Root on every VM, Docker is yours to install; nested KVM documented on E5 and Standard3 Flex VMs, not on Ampere Arm VMs | Docker, root, systemd and /dev/kvm ready on every machine |
| Coding agents | None preinstalled; you install agents and drive them over SSH or startup scripts | None preinstalled; Oracle's AI agent service builds business agents, not coding agents in a VM | Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API |
| Keeping keys out of the machine | No credential proxy; service account tokens and Secret Manager values are read from inside the VM | Secrets live in OCI Secret Management and the VM fetches them, so the key ends up inside the machine | AI credentials proxy for model providers and any HTTPS API; switch a credential to another account when credits run out |
| Desktop and computer use | No managed desktop, browser or computer-use API on Linux VMs | No desktop on Compute VMs; OCI Secure Desktops is a separate desktop-as-a-service product | 60fps desktop and Chromium, computer use and browser control, embeddable stream |
| Files after a stop | Persistent disks are kept and billed while stopped; snapshots on demand or on a schedule you set | Stopping pauses compute billing; the boot volume stays and is billed until you delete it; backups by hand or on a schedule | Snapshots every minute and on stop; a stopped sandbox costs nothing |
| Pause and fork | Suspend keeps memory for up to 60 days; cloning means making a machine image and booting a new VM | Make a custom image of the boot disk; imaging a running VM shuts it down for several minutes | Fork a running sandbox while it keeps running; resume reboots from disk (no memory snapshots) |
| Reaching the machine | External IPv4 and any port through firewall rules; HTTPS needs your own certificate and a load balancer or web server | Public IPv4 per VM and any port you open in security rules; HTTPS through your own server or a load balancer | HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4 |
| Outbound network rules | VPC egress firewall rules by IP and port; domain rules on Cloud NGFW Standard, a paid tier | Security lists and network security groups by IP range, protocol and port; domain filtering through OCI Network Firewall | Open internet on every plan; no provider-side allowlist |
| Running many at once | Per-region CPU quotas that vary by project, plus a global CPU cap on some new projects; free trial accounts cannot request more | Default or trial service limits per tenancy; more needs a limit increase request, which may not be approved | 100 to 2,000 sandboxes at once by plan, self-serve |
| GPUs | Nvidia GPUs from T4 and L4 up to H100, H200, B200 and GB200 | Nvidia B200, H200, H100 and A100 and AMD MI300X, in VMs and bare metal | None |
| Operating systems and CPUs | Linux and Windows Server on Intel, AMD and Arm; no macOS | Linux and Windows Server on AMD, Intel and Ampere Arm, plus Mac servers | Linux on AMD Ryzen 9 9950X (Zen 5); Wine for Windows apps |
| Regions | 43 regions and 130 zones worldwide, including many in the EU | Regions across the Americas, Europe, the UK, Asia, the Middle East, Africa and Australia, plus government and sovereign clouds | EU only: Germany, Finland, France |
Every feature Battleships tracks (43)
| Feature | Google Compute Engine | Oracle Cloud | boat |
|---|---|---|---|
| Full VM (own kernel) | Yes | Yes | Yes |
| Docker inside | Yes | Yes | Yes |
| Nested virtualization | No | Yes | Yes |
| Root | Yes | Yes | Yes |
| Systemd | Yes | Yes | Yes |
| Desktop GUI | No | No | Yes |
| Browser | No | No | Yes |
| Computer-use API | No | No | Yes |
| SSH | Yes | Yes | Yes |
| Persistent disk | Yes | Yes | Yes |
| Pause/resume | Yes | Yes | Yes |
| Automatic snapshots | No | No | Yes |
| Fork/clone | Yes | Yes | Yes |
| Live fork (no pause) | Yes | No | Yes |
| Memory fork | No | No | No |
| Custom image (Docker or snapshot) | Yes | Yes | Yes |
| ≥24 h sessions | Yes | Yes | Yes |
| Idle auto-stop | No | No | No |
| Wake on request | No | No | No |
| HTTPS preview URLs | No | No | Yes |
| Raw TCP inbound | Yes | Yes | Yes |
| Public IPv4 | Yes | Yes | No |
| Egress allowlist | Yes | Yes | No |
| Secret proxy | No | No | Yes |
| Secret proxy for any API | No | No | Yes |
| Preinstalled agents | No | No | Yes |
| Agent harness API | No | No | Yes |
| MCP server | No | Yes | No |
| GPU | Yes | Yes | No |
| Windows | No | Yes | No |
| Arm64 | Yes | Yes | No |
| Self-hosting | No | No | No |
| BYOC | No | No | No |
| Open source | No | No | No |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | No | Not published | Yes |
| Usage per sandbox (API) | Yes | Yes | Yes |
| Audit logs | Yes | Not published | No |
| SSO | Yes | Yes | No |
| SOC 2 | Yes | Yes | No |
| HIPAA | Yes | Yes | No |
| EU data residency | Yes | Yes | Yes |
| Zero data retention | Not published | Not published | Yes |
Performance
There is no comparable measurement for either cloud: neither Compute Engine nor Oracle Cloud was part of StarSling's public benchmark run of 2026-09-29, where boat did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB on dedicated AMD Ryzen 9 9950X (Zen 5) hosts. Speed on both clouds depends on the machine family you pick, and Oracle's burstable shapes guarantee only 12.5% of each vCPU.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Google Compute Engine and Oracle Cloud 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.
| Scenario | Google Compute Engine, a month | Oracle 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 | $31 · Pay-as-you-go billing account (no plan fee) | $5 · Usage account · with limits | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $64 · Pay-as-you-go billing account (no plan fee) · with limits | $15 · Usage account · with limits | $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 | $1,291 · Pay-as-you-go billing account (no plan fee) | $203 · Usage account · 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 | $378 · Pay-as-you-go billing account (no plan fee) | $49 · Usage account · with limits | $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 | $690 · Pay-as-you-go billing account (no plan fee) | $97 · Usage account · with limits | $2,000 · $2000/mo · with limits |
Google Compute Engine: Compute Engine bills each VM per second after a 1 minute minimum, here E2 sizes on demand in us-central1, with no platform fee. Disks and external IPs bill on top, also while stopped, and the lowest always-on rate needs a 3-year commitment.
Oracle Cloud: Oracle bills per OCPU hour (one x86 OCPU is two vCPUs) and per GB hour of memory while a VM runs, plus the boot disk until you delete it. The prices here use burstable E4 VMs billed at a 12.5% CPU baseline, so they sit far below a full-speed VM.
Oracle Cloud: You pick a monthly plan from $20, and all of it becomes usage credit spent per second on running sandboxes. A stopped sandbox costs nothing and keeps its disk.
Limits in those scenarios
Every Oracle row is marked as not fitting because its prices come from burstable VMs that guarantee only an eighth of each vCPU, so an agent that builds and tests all day gets throttled. Compute Engine fits most rows, but one agent that never stops reaches its price only with a 3-year commitment. Both clouds cap how many VMs a new account can run, with per-region CPU quotas on Compute Engine and tenancy service limits on Oracle, so 50 agents or 300 user sandboxes may need an approved increase first. On boat the 1,000-job burst does not fit: the $2,000 plan allows 90 starts a minute, 600 an hour and 2,400 a day.
| Scenario | Google Compute Engine | Oracle Cloud | boat |
|---|---|---|---|
| One coding agent, workdays | None that bite |
| None that bite |
| One agent that never stops |
|
| None that bite |
| 50 agents in parallel | None that bite |
| None that bite |
| Sandboxes for your users | None that bite |
| None that bite |
| 1,000 short jobs at once | None that bite |
|
|
Which one should you use?
Use Google Compute Engine when
you want full-speed VMs from many families, memory suspend, Nvidia GPUs up to GB200, or 43 regions, and you have people to build the agent tooling on top.
- Full-speed shapes in the priced scenarios. Useful for busy agents: the e2-standard VMs priced here fit four of five scenarios, while Oracle's priced VMs are burstable and throttled.
- Suspend with memory. Useful if a VM must resume with its processes running: suspend keeps memory for up to 60 days.
- Many regions and zones. Useful if you spread VMs across zones: Google Cloud runs 43 regions and 130 zones.
Use Oracle Cloud when
you need Mac servers, AMD GPUs, a free Arm box, or very cheap burstable VMs for light work, and you are ready to build your own sandbox layer.
- A free Arm box every month. Useful for one small personal box: Always Free covers 1,500 Arm OCPU hours and 9,000 GB hours a month.
- Mac servers and AMD GPUs. Useful for iOS builds or AMD GPU work: Oracle Cloud rents Mac servers and AMD MI300X next to Nvidia GPUs.
- Very low prices for light work. Useful for mostly idle VMs: burstable shapes bill at a 12.5% CPU baseline, so list prices are far below full-speed VMs.
Where boat fits
Both are general clouds, not agent sandboxes, so a team comparing them for agents is really choosing which VMs to build a sandbox layer on. If what you are really running is coding or computer-use agents, boat gives every agent a full VM that already has Docker, /dev/kvm, a desktop, coding agents, HTTPS URLs and snapshots every minute. Keep the cloud for GPUs, Macs, Windows or regions outside the EU.
/dev/kvm on every machine
Compute Engine has no nested virtualization on E2, Arm or most AMD VMs, and Oracle documents it only on E5 and Standard3 Flex, not on its Arm shapes. Android emulators mean picking and enabling the right family for each VM.
Saved every minute, forked while running
On both clouds, snapshots run by hand or on a schedule and disks bill while stopped. A copy is a machine image plus a fresh boot on Compute Engine, and on Oracle imaging a running VM shuts it down for minutes.
Agents, desktop and keys handled
Neither cloud ships coding agents, a way to prompt them, or a desktop stream, so you build all three. Keys come from Secret Manager or OCI Secret Management into the VM, where code the agent runs can read them.
Use boat when your agents need full VMs that already have Docker, /dev/kvm, a desktop, coding agents, HTTPS URLs, snapshots every minute and forks, without building that layer on either cloud. Compare it one to one: boat vs Google Compute Engine and boat vs Oracle Cloud.
Questions people ask
Is Google Compute Engine cheaper than Oracle Cloud?
On list price, Oracle Cloud is cheaper in every scenario we priced: one coding agent on workdays is $5 a month against $31, and 50 agents in parallel come to $203 against $1,291 a month. Oracle's numbers use burstable VMs throttled above 12.5% of each vCPU, so they do not fit busy agents, while Compute Engine's e2-standard VMs run at full speed.
Which is faster, Compute Engine or Oracle Cloud?
Neither was in the public benchmark run we use, so there is no fair speed comparison. Both depend on the machine family you choose. Oracle's cheapest burstable shapes guarantee only 12.5% of each vCPU, so a steady build will be slower there than on a full-speed shape.
Can Compute Engine and Oracle Cloud run Docker and nested virtualization?
Both give you root, so Docker runs on either. Nested KVM is partial on both: Compute Engine supports it on some Intel families and N4D but not on E2, Arm or other AMD VMs, and Oracle documents it on E5 and Standard3 Flex but not on Ampere Arm. boat has /dev/kvm on every machine.
Does Oracle Cloud's Always Free tier beat a paid VM?
For one small box, it can. Every tenancy gets 1,500 Arm OCPU hours and 9,000 GB hours of memory a month free, but the shape is Arm, Oracle may reclaim VMs it judges idle, and it is not enough for a fleet.
Which has GPUs, Windows or macOS?
Both rent Nvidia GPUs and run Windows Server. Oracle Cloud also rents AMD MI300X GPUs and Mac servers; Compute Engine has no macOS. boat runs only Linux on x86, with no GPUs, and has Wine docs for Windows apps.
Where does my data run?
Compute Engine runs in 43 regions and 130 zones, and Oracle Cloud has regions across the Americas, Europe, Asia, the Middle East, Africa and Australia, plus government clouds. boat runs only in the EU (Germany, Finland and France), with EU data residency.
Try boat
Install the CLI, then run boat new.
7-day free trial
More comparisons
- boat vs Google Compute Engine
- boat vs Oracle Cloud
- AWS EC2 vs Google Compute Engine
- Google Compute Engine vs Azure Virtual Machines
- Google Compute Engine vs DigitalOcean
- Google Compute Engine vs Hetzner Cloud
- Google Compute Engine vs Vultr
- Google Compute Engine vs Linode
- Google Compute Engine vs Scaleway
- Google Compute Engine vs OVHcloud
- Google Compute Engine vs IONOS Cloud
- Google Compute Engine vs Contabo