Nebius vs Hyperstack
Both rent Nvidia GPUs and plain Linux VMs, but Nebius is a wider AI cloud with Slurm, Kubernetes, serverless jobs and CPU VMs up to 256 vCPU, while Hyperstack rents a fixed list of GPU and CPU flavors by the minute. The biggest day-to-day difference: a stopped Nebius VM pays only for its disks, and a stopped Hyperstack VM keeps billing in full until you hibernate it.
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.
Hyperstack
Hyperstack is the GPU cloud of NexGen Cloud, a UK company founded in 2020: you rent GPU and CPU virtual machines, Kubernetes clusters and storage by the minute.
A Hyperstack VM is a full virtual machine with root and SSH, booted from an Ubuntu image (with or without Docker) and sized from a fixed list of flavors, from one RTX A4000 to 8-GPU B300 nodes, plus CPU-only flavors from 4 to 32 vCPUs. It also runs managed Kubernetes, volumes, object storage and an AI Studio inference API. It is built for teams that train, fine-tune and serve models, not as a sandbox API for agents, so this page compares the parts that overlap: VMs as machines for agent work.
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 | Hyperstack: price rank | bochka: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Both give a coding agent a full VM with root and Docker, but you install the agent and tooling yourself, and bochka ranks 3rd with all of that ready. | 197th of 367, with limits $2,272 a month | 343rd of 367, with limits $11,458 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 | 337th of 366, with limits $12,775 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 | 111th of 345, with limits $202 a month | 172nd of 345, with limits $1,167 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 Nebius needs a quota raise past 12 VMs per region, and Hyperstack does not publish how many VMs can run at once, so 300 user sandboxes is not confirmed there. | 81st of 361, with limits $716 a month | 173rd of 361, with limits $4,207 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 None of the three fully fits RL rollouts; Nebius ranks 52nd, ahead of Hyperstack at 173rd, and bochka ranks 34th. | 52nd of 360, with limits $3,049 a month | 173rd of 360, with limits $17,500 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 | 204th of 364, with limits $3,250 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 fit as CI runners on full VMs; Nebius bills per second and Hyperstack per minute, and bochka adds snapshots every minute and free stops. | 27th cheapest of 377 $297 a month | 122nd cheapest of 377 $1,560 a month | 4th cheapest of 377 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB Both keep the whole disk, but a stopped Nebius VM pays only for disks while a stopped Hyperstack VM bills in full unless you hibernate it. | 68th cheapest of 368 $91 a month | 115th cheapest of 368 $475 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 desktop or computer-use product, only SSH and a text console; bochka streams a 60fps desktop with Chromium on every sandbox. | 187th of 369, with limits $610 a month | 332nd of 369, with limits $3,250 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 On both you install and run the browser yourself on a plain VM; bochka ships Chromium and browser control in every sandbox. | 223rd of 367, with limits $305 a month | 335th of 367, with limits $1,750 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 | 327th of 346, with limits $1,750 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 | 310th of 338, with limits $35,000 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 | 309th of 360, with limits $876 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, and Hyperstack ranks 2nd of 98 against Nebius at 18th; bochka has no GPUs, so this work belongs on one of them. | 18th cheapest of 98 $1,127 a month | 2nd cheapest of 98 $625 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 | Hyperstack | bochka |
|---|---|---|---|
| Kind of product | AI cloud: GPU and CPU VMs, Managed Kubernetes, Slurm clusters, Serverless AI and storage | GPU cloud: GPU and CPU VMs, Kubernetes clusters, volumes, object storage and AI Studio inference | Sandboxes for AI agents: VMs with agents, desktop and hosting ready |
| Isolation | Every Compute VM is a full VM behind a hypervisor | Full VMs with root, on flavors mapped 1:1 to physical resources | Every sandbox is a full VM with its own kernel |
| Docker and root | Root on your own VM and container VMs from Docker images | Root and SSH; Ubuntu 22.04 and 24.04 images with or without Docker | Docker, Docker Compose, root, systemd and /dev/kvm on every machine |
| GPUs | Nvidia L40S, RTX PRO 6000, H100, H200, B200 and B300, on-demand or preemptible | Nvidia RTX A4000 up to H100, H200, B200 and B300, as single cards or 8-GPU nodes, on demand, spot or reserved | None |
| CPU machines | AMD EPYC Genoa from 2 vCPU / 8 GB up to 256 vCPU / 1,024 GB | CPU-only flavors from 4 vCPU / 4 GB up to 32 vCPU / 64 GB (Canada) | Up to 16 vCPU / 32 GB on the $100 plan and up |
| Stopping a machine | Compute stops billing; disks keep billing on their full provisioned size | A SHUTOFF VM bills at the full rate; hibernation bills only disk and a kept IP, and a restore needs the flavor in stock | A stopped sandbox costs nothing and keeps its whole disk |
| Snapshots | Manual disk snapshots, taken while the VM runs | Snapshots on request only; the VM is shut off while it is taken, one at a time, up to 100 GB | The disk is snapshotted every minute and on stop, while the sandbox keeps running |
| Fork | Make a disk from a snapshot, then a VM from the disk | No clone or fork; you snapshot, make a custom image and deploy a new VM | Fork a running sandbox into a new one in one call |
| Desktop and computer use | No desktop or browser product; a text serial console | SSH and text console logs; no graphical desktop | 60fps desktop and Chromium, computer use and browser control, embeddable stream |
| Coding agents | None built in; you install and run agents yourself over SSH | A hosted MCP server lets your own agent manage Hyperstack; nothing runs an agent inside a VM for you | Claude Code, Codex, OpenCode and more preinstalled, driven from outside with a prompt, steer and events API |
| Reaching the machine | Optional public IPv4 per VM, SSH and Nebius Tunnels; you set up TLS yourself | A public IPv4 you can keep static; open ports with firewall rules and run your own TLS | HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4 |
| Network rules | VPC security groups allow or deny ingress and egress by IP range, port and protocol | Per-VM firewall: inbound closed by default, outbound rules by IP, protocol and port | Open internet out; inbound through HTTPS URLs and TCP ports |
| Keeping keys out of the machine | SecretStash stores secrets that a VM fetches with its service account | No credentials proxy; you pass keys in with cloud-init, files or environment variables | AI credentials proxy for model providers and any HTTPS API you register |
| Regions | Finland, France, Spain, United Kingdom, Israel and two US regions | Norway, Canada (Quebec) and the US (Texas) | EU only: Germany, Finland, France |
Every feature Battleships tracks (43)
| Feature | Nebius | Hyperstack | bochka |
|---|---|---|---|
| Full VM (own kernel) | Yes | Yes | Yes |
| Docker inside | Yes | Yes | Yes |
| Nested virtualization | Not published | Not published | Yes |
| Root | Yes | Yes | 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 | No | No |
| Wake on request | No | No | No |
| HTTPS preview URLs | Yes | No | Yes |
| Raw TCP inbound | Not published | No | 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 | 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 | Not published | Yes |
| Usage per sandbox (API) | Yes | No | Yes |
| Audit logs | Not published | Not published | No |
| SSO | Yes | No | No |
| SOC 2 | Yes | Yes | No |
| HIPAA | Yes | No | No |
| EU data residency | Yes | Yes | Yes |
| Zero data retention | Not published | Not published | Yes |
Performance
There is no comparable CPU measurement for Nebius or Hyperstack: neither was part of the public StarSling benchmark run. bochka 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 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 Hyperstack 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 | Hyperstack, a month | bochka, 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 | $114 · Prepaid on-demand | $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 | $475 · Prepaid on-demand | $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 | $5,720 · Prepaid on-demand | $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 | $1,750 · Prepaid on-demand · 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 | $610 · Pay as you go | $3,500 · Prepaid on-demand · with limits | $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 provisioned size even while the VM is stopped. The table uses the rates from 1 October 2026: AMD Genoa costs $0.015 per vCPU-hour plus $0.0045 per GiB-hour.
Hyperstack: Hyperstack bills a running VM per minute from a prepaid balance, at a fixed hourly price per flavor, with no charge for bandwidth. CPU-only flavors start at 4 vCPU / 4 GB, so the table uses the closest flavor, and a stopped VM keeps billing until you hibernate it.
Hyperstack: 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 at once all need a quota raise first. Hyperstack publishes no VM-count limit, so 300 sandboxes and 1,000 jobs at once are not confirmed there, and its API allows 500 requests a minute per IP. Nebius's smallest CPU VM is 2 vCPU / 8 GB, and Hyperstack's is 4 vCPU / 4 GB, so neither matches small sandboxes exactly. On bochka the 1,000-job burst is queued and needs more starts per day than the $2,000 plan allows.
| Scenario | Nebius | Hyperstack | bochka |
|---|---|---|---|
| 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 |
| 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 Nebius when
you want a full AI cloud: Slurm or Kubernetes clusters, very large CPU machines, regions in the UK, Israel or Spain, and VMs that pay only for disks while stopped.
- Stopped VMs pay only for disks. Useful if machines sit idle between runs: a stopped Nebius VM stops billing for vCPU, RAM and GPUs, and keeps paying only for disks.
- A wider cloud around the VMs. Useful if you train at scale: Nebius runs Slurm and Managed Kubernetes clusters, serverless jobs and CPU VMs up to 256 vCPU.
- More regions in Europe and beyond. Useful if you need Finland, France, Spain, the UK or Israel, as well as two regions in the US.
Use Hyperstack when
you want simple GPU VMs by the minute, a wide range of cards from an RTX A4000 to 8-GPU B300 nodes, or regions in Canada and Texas.
- More GPU models, down to one small card. Useful if you want a cheap single GPU or 8-GPU nodes: Hyperstack rents from an RTX A4000 up to B300, with spot and reserved prices.
- GPU eval jobs for less. Useful for GPU evals: in our GPU eval scenario Hyperstack ranks 2nd of 98, ahead of Nebius at 18th.
- Regions in Canada and Texas. Useful if your users or data are in North America: Hyperstack runs in Quebec and Texas as well as Norway.
Where bochka fits
Nebius and Hyperstack are GPU clouds, and bochka has no GPUs, so model training and GPU evals belong on one of them. If what you are really running is agents that write code, use a browser or host apps on CPU machines, both give you a bare Ubuntu VM and leave the rest to you. bochka gives every agent a full VM with that work already done.
A disk saved every minute, fork and free stops
On Nebius you snapshot by hand and a stopped VM still pays for its disks. On Hyperstack a stopped VM bills in full, and a snapshot shuts the VM off. bochka saves the disk every minute and can fork a running sandbox.
Coding agents driven from outside, keys kept out
Neither runs an agent inside a VM for you, so you write the runner that starts, prompts and watches each one, and the real key sits on the machine. bochka ships the agents with an API and a credentials proxy.
A desktop, a browser and HTTPS on every machine
Both give you SSH and a text console only, so computer-use agents need a desktop, VNC and a browser installed per image, plus your own TLS. bochka has a 60fps desktop, Chromium and HTTPS URLs for 50 ports.
Use bochka when your agents run on CPU: a full VM with Docker, a disk saved every minute, free stops, fork, a 60fps desktop, HTTPS ports and coding agents built in. Compare it one to one: bochka vs Nebius and bochka vs Hyperstack.
Questions people ask
Is Nebius cheaper than Hyperstack?
For CPU work in our scenarios, yes: one agent running all month is $91 a month on Nebius against $475 on Hyperstack, and 50 agents in parallel are $1,127 against $5,720. Hyperstack's closest CPU flavor has 8 vCPU / 8 GB, so part of the gap is size. For GPU eval jobs Hyperstack is lower, at $625 a month against $1,127.
Which is faster, Nebius or Hyperstack?
We have no comparable CPU measurement for either: neither was in the public StarSling benchmark run. bochka did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB in that run.
Do stopped VMs cost money on Nebius and Hyperstack?
On Nebius a stopped VM stops paying for vCPU, RAM and GPUs, but its disks keep billing. On Hyperstack a SHUTOFF VM bills at the full rate; a hibernated one bills only its disk and a kept IP, and restoring it needs the flavor in stock. A stopped bochka sandbox costs nothing.
Can Nebius and Hyperstack run Docker?
Yes. Both give you full VMs with root: Nebius also runs container VMs from Docker images, and Hyperstack has Ubuntu images with Docker installed. bochka sandboxes come with Docker, Docker Compose, systemd and /dev/kvm ready.
Which has more GPU choice?
Hyperstack lists more models, from an RTX A4000 up to B300, with spot and reserved prices. Nebius rents L40S, RTX PRO 6000, H100, H200, B200 and B300, on demand or preemptible. bochka has no GPUs.
Is bochka SOC 2 compliant?
SOC 2 is in progress. We're open about where we stand: live status at status.bochka.xyz and our data retention policy at docs.bochka.xyz/data-retention.
Try bochka
Install the CLI, then run bochka new.
Product preview