Nebius vs Paperspace
Both rent full Linux VMs with Nvidia GPUs, but Nebius is a whole AI cloud with B200 and B300 GPUs, Kubernetes, Slurm and per-second billing. Paperspace, now part of DigitalOcean, is simpler: you rent a machine by the hour and work inside it over SSH.
The companies and products
Nebius
Nebius is an AI cloud company, listed on Nasdaq and based in Amsterdam, that rents GPUs and the infrastructure around them.
You create a Compute VM from a preset (GPU or CPU only), add disks and an SSH key, and get a full Linux VM with root and an optional public IP. Around it Nebius sells Managed Kubernetes, Slurm clusters (Soperator), Serverless AI jobs and endpoints, object storage and networking. It is built for teams that train, fine-tune and serve AI models, not as a sandbox for agents.
Paperspace
Paperspace is a GPU cloud for machine learning, owned by DigitalOcean since 2023.
Paperspace Machines are full Linux VMs with a persistent disk, a public IP and optional Nvidia GPUs from M4000 up to H100, billed by the hour. Next to Machines it sells Gradient Notebooks, Deployments and Workflows for training and serving models. It is built for people who train, test and run models, not as a sandbox API: its API manages machines, and you work inside them over SSH.
- Operating systems
- Linux; Windows templates closed to new users since 1 July 2024
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. “29th cheapest of 364” means 28 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 | Nebius: price rank | Paperspace: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Both give you a real VM, but you install the agent and tooling yourself and fork by hand; boat ships coding agents and one-call fork. | 197th of 367, with limits $2,272 a month | 141st of 367, with limits $1,261 a month | 3rd cheapest of 367 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB | 210th of 366, with limits $2,261 a month | 179th of 366, with limits $1,560 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 Paperspace bills a full hour for every short run, so bursts of short jobs cost far more there than on Nebius; boat's start limits slow big bursts too. | 111th of 345, with limits $202 a month | 179th of 345, with limits $1,451 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 Short user sessions favor Nebius, which bills per second, while Paperspace bills a full hour for each 10-minute session; boat comes with the in-machine API both lack. | 81st of 361, with limits $716 a month | 124th of 361, with limits $1,243 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 Nebius bills per second and fits RL rollouts far better than Paperspace, whose hourly step makes many short rollouts expensive. | 52nd of 360, with limits $3,049 a month | 193rd of 360, with limits $30,570 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 | 29th cheapest of 364 $610 a month | 58th cheapest of 364 $1,241 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 Both can host CI runners on plain VMs; Nebius bills per second while Paperspace rounds each job up to the hour. | 27th cheapest of 377 $297 a month | 101st cheapest of 377 $1,026 a month | 4th cheapest of 377 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB For one machine that never stops, Paperspace's hourly C5 rate comes in under Nebius, and boat costs less than both. | 68th cheapest of 368 $91 a month | 49th cheapest of 368 $63 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 has a Linux desktop: Nebius offers a text serial console and Paperspace streams only Windows machines new users cannot create, while boat ships a 60fps desktop. | 187th of 369, with limits $610 a month | 284th of 369, with limits $1,655 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 | 223rd of 367, with limits $305 a month | 319th of 367, with limits $1,241 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 | 253rd of 346, with limits $302 a month | 192nd of 346, with limits $112 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 | 174th of 338, with limits $6,049 a month | 92nd of 338, with limits $2,242 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 | 176th of 360, with limits $156 a month | 207th of 360, with limits $214 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 Both fit GPU eval jobs: Nebius ranks 18th with newer cards up to B300, Paperspace 30th with cards up to H100, and boat has no GPUs. | 18th cheapest of 98 $1,127 a month | 30th cheapest of 98 $2,982 a month | 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 | Nebius | Paperspace | boat |
|---|---|---|---|
| Kind of product | AI cloud: GPU and CPU VMs, Kubernetes, Slurm clusters and serverless jobs | GPU and CPU VM cloud for machine learning, plus Gradient notebooks | Sandboxes for AI agents: VMs with agents, desktop and hosting ready |
| GPUs | Nvidia L40S, RTX PRO 6000, H100, H200, B200 and B300, on-demand or preemptible | Nvidia M4000 up to A100 and H100 (H100 needs approval), up to 8 per machine | None |
| Billing step | Per second | Per hour | Per second, from a $20 monthly plan that becomes usage credit |
| Smallest CPU machine | 2 vCPU / 8 GB | C4: 2 vCPU / 4 GB | 2 vCPU / 4 GB |
| Biggest CPU machine | 256 vCPU / 1,024 GB | C10: 32 vCPU / 244 GB | 16 vCPU / 32 GB on the $100 plan and up |
| Docker and nested virtualization | Root on your VM and container VMs from Docker images; nested virtualization not documented | Docker runs inside; nested virtualization not documented | Docker, root, systemd and /dev/kvm on every machine |
| Stopping a machine | Compute stops billing; disks keep billing on their full size | Disk and public IP keep billing until you delete the machine | A stopped sandbox costs nothing and keeps its disk |
| Snapshots and copies | Manual disk snapshots of a running VM; a copy is a disk from the snapshot, then a VM | Auto-snapshots at most hourly, up to 10 kept; copies need a template from a stopped machine | Snapshots every minute and on stop; fork a running sandbox in one call |
| Working inside from code | No sandbox API; you work inside over SSH | The API and CLI manage machines but cannot connect to them; you use SSH | API, SDK and CLI to run commands, move files and fork |
| Desktop and computer use | No desktop or browser product; a text serial console | Desktop streaming only on Windows machines, which new users cannot create | 60fps desktop and Chromium, computer use and browser control, embeddable stream |
| Coding agents | None built in; you install agents over SSH | None built in; ML templates such as ML-in-a-Box | Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API |
| Network rules | VPC security groups allow or deny ingress and egress by IP range, port and protocol | Public IPv4 on every Linux machine; no provider egress allowlist | Open internet on every plan; HTTPS URLs for up to 50 ports, raw TCP |
| Keeping keys out of the machine | SecretStash stores secrets that the VM fetches itself | No credentials proxy; Gradient secrets are injected as environment variables | AI credentials proxy for model providers and any HTTPS API you register |
| Regions | Finland, France, Spain, United Kingdom, Israel and the US | New York, Amsterdam and Canada | EU only: Germany, Finland, France |
Every feature Battleships tracks (43)
| Feature | Nebius | Paperspace | boat |
|---|---|---|---|
| Full VM (own kernel) | Yes | Yes | Yes |
| Docker inside | Yes | Yes | Yes |
| Nested virtualization | Not published | Not published | Yes |
| Root | Yes | Not published | Yes |
| Systemd | Not published | Not published | 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 | No | 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 | Yes | No |
| Wake on request | No | No | No |
| HTTPS preview URLs | Yes | No | Yes |
| Raw TCP inbound | Not published | Not published | Yes |
| Public IPv4 | Yes | Yes | No |
| Egress allowlist | Yes | No | 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 | No | No |
| GPU | Yes | Yes | No |
| Windows | No | No | No |
| Arm64 | No | No | 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 | Not published | Yes | Yes |
| Usage per sandbox (API) | Yes | No | Yes |
| Audit logs | Not published | 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 Nebius or Paperspace: neither was part of the public StarSling benchmark run. boat did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB in that run, on dedicated AMD Ryzen 9 9950X (Zen 5) hosts. Nebius CPU-only VMs run on AMD EPYC Genoa or Intel Ice Lake.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Nebius and Paperspace 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 | Nebius, a month | Paperspace, 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 | $25 · Pay as you go | $19 · Core usage | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $91 · Pay as you go | $63 · Core usage | $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 | $1,127 · Pay as you go | $804 · Core usage | $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 | $305 · Pay as you go | $1,241 · Core usage | $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 | $610 · Pay as you go | $8,274 · Core usage | $2,000 · $2000/mo · with limits |
Nebius: Nebius bills running VMs per second for vCPU and RAM, with no plan fee, and bills disks by size even while the VM is stopped. From 1 October 2026, AMD Genoa costs $0.015 per vCPU-hour plus $0.0045 per GiB-hour.
Paperspace: Paperspace bills each machine per hour while it is on, at a fixed rate per size (C5 is $0.08 an hour). The boot disk (from $5 a month for 50 GB) and public IP keep billing until you delete the machine.
Paperspace: 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
Nebius starts each region at 12 VMs and, in most regions, 200 vCPU across CPU-only VMs, so 50 agents in parallel, 300 user sandboxes and 1,000 short jobs all need a quota raise first. Paperspace does not publish a machine count limit; your team's limits sit on the billing tab and are raised by support ticket. Paperspace's hourly step is what bites for short work: 1,000 short jobs at once come to $8,274 a month there and $610 on Nebius. boat queues that burst and needs more starts per day than its $2,000 plan allows.
| Scenario | Nebius | Paperspace | boat |
|---|---|---|---|
| One coding agent, workdays | None that bite | None that bite | None that bite |
| One agent that never stops | None that bite | None that bite | None that bite |
| 50 agents in parallel | None that bite | None that bite | None that bite |
| Sandboxes for your users | None that bite | None that bite | None that bite |
| 1,000 short jobs at once | None that bite | None that bite |
|
Which one should you use?
Use Nebius when
you need the newest Nvidia GPUs, Slurm or Kubernetes clusters, very large machines, per-second billing for short jobs, or regions in the UK, Israel or the US.
- Newest GPUs and real clusters. For teams training or serving big models: Nebius rents B200 and B300 GPUs and runs Slurm and Kubernetes clusters.
- Per-second billing. For short or bursty jobs: Nebius bills per second, so a 10-minute session is not charged as a full hour.
- Huge machines and network rules. For one very large job or locked-down networks: CPU VMs reach 256 vCPU, and security groups control egress by IP range.
Use Paperspace when
you want a simple GPU or CPU machine that a person uses over SSH for hours at a time, notebooks with free GPUs, or machines in New York or Canada.
- Simple, low hourly rates. For one machine that runs for hours or all month: a C5 with 4 vCPU and 8 GB is $0.08 an hour.
- Small machines and auto-shutdown. For light work: CPU machines start at 2 vCPU and 4 GB, and auto-shutdown can stop a forgotten machine.
- Notebooks with free GPUs. For people who experiment in notebooks: Gradient Pro unlocks free notebook GPUs like the A4000, when available, for up to 6 hours.
Where boat fits
Nebius and Paperspace are GPU clouds, and boat has no GPUs, so model training belongs on one of them. If what you are really running is agents, both hand you a bare Linux VM reached over SSH, and you build the agent setup, desktop, snapshots and in-machine API yourself. boat ships those on every sandbox.
Agents you drive from outside
Neither has an API to run commands inside a machine, so on both you write your own SSH layer and install each agent by hand. boat preinstalls Claude Code, Codex and OpenCode with a prompt, steer and events API.
Saved every minute, fork while running
Nebius snapshots only when you ask, and Paperspace at most hourly, so a broken agent can lose real work. Copying means a new disk and VM on Nebius, or stopping the machine for a template on Paperspace. boat forks a running sandbox in one call.
A desktop for computer use
Nebius gives a text serial console and Paperspace streams only Windows machines new users cannot create, so a computer-use agent needs a desktop and VNC server you set up on each VM. boat streams a 60fps desktop with Chromium.
Use boat when you run agents rather than train models: VMs that come with coding agents, a desktop and browser, an API to work inside them, snapshots every minute and a fork of a running sandbox. Compare it one to one: boat vs Nebius and boat vs Paperspace.
Questions people ask
Is Nebius cheaper than Paperspace?
For long-running machines, Paperspace is cheaper: one agent that never stops is $63 a month there and $91 on Nebius, and 50 agents in parallel are $804 against $1,127. For short sessions Nebius wins because it bills per second: sandboxes for your users are $305 on Nebius and $1,241 on Paperspace. boat comes to $26, $317 and $200 for those three.
Which has better GPUs, Nebius or Paperspace?
Nebius rents newer cards: L40S, RTX PRO 6000, H100, H200, B200 and B300. Paperspace ranges from older cards like the M4000 up to A100 and H100, and H100s need approval first. boat has no GPUs.
Which is faster?
We have no comparable CPU measurement for either: neither was in the public StarSling benchmark run. boat did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB in that run.
Can Nebius and Paperspace run Docker?
Yes. Both give you a full VM where Docker runs, and Nebius can also start container VMs from Docker images. Neither documents nested virtualization, while boat has Docker, systemd and /dev/kvm on every sandbox.
Does a stopped machine still cost money?
On both, yes. A stopped Nebius VM keeps paying for its disks, and a Paperspace machine that is off keeps paying for its disk and public IP until you delete it. A stopped boat sandbox costs nothing and keeps its disk.
Where do Nebius and Paperspace run?
Nebius runs in Finland, France, Spain, the United Kingdom, Israel and the US. Paperspace runs in New York, Amsterdam and Canada. 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