Hyperstack vs Baseten
Hyperstack is a GPU cloud: you rent full virtual machines, with or without Nvidia GPUs, and run whatever you like on them. Baseten is an inference cloud: you hand it a model or a Docker image, and it runs it as containers that scale with traffic, down to zero when idle.
The companies and products
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.
Baseten
Baseten is an inference cloud: it runs AI models on GPUs for production apps, and sells hosted open models and training.
You package a model with Truss, its open-source format, or bring a Docker image, and Baseten runs it as replicas on Kubernetes that scale up with traffic and down to zero. It also sells Model APIs (open models priced per token) and training jobs on the same GPUs. Sandboxes, isolated environments where an agent runs commands and works with files, are in private preview; this page compares bochka with both the CPU instances and the sandbox preview.
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. “122nd cheapest of 377” means 121 of the other 376 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 | Hyperstack: price rank | Baseten: price rank | bochka: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A coding agent needs root, Docker and a disk that stays; Hyperstack has the first two but bills a stopped VM in full, and Baseten's containers have neither root nor a kept disk. | 343rd of 367, with limits $11,458 a month | 216th of 367, with limits $2,619 a month | 3rd cheapest of 367 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB An assistant that lives for months needs a kept disk: Hyperstack keeps it but bills a stopped VM in full, and Baseten's preview sandboxes are deleted after 7 days by default. | 337th of 366, with limits $12,775 a month | 242nd of 366, with limits $3,258 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 No one fits short code bursts cleanly here: Hyperstack bills by the minute with no idle stop, and each Baseten burst also pays for cold start and, by default, 15 idle minutes. | 172nd of 345, with limits $1,167 a month | 319th of 345, with limits $1,856 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 Hyperstack does not publish how many VMs can run at once, and Baseten's sandbox preview allows 10 per organization by default, so 300 user sandboxes is not confirmed on either. | 173rd of 361, with limits $4,207 a month | 277th of 361, with limits $1,868 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 | 173rd of 360, with limits $17,500 a month | 322nd of 360, with limits $31,140 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 | 204th of 364, with limits $3,250 a month | 297th of 364, with limits $1,609 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 Hyperstack fits CI on full VMs with root and Docker images; Baseten's containers run as a non-root user with no documented Docker inside, and bochka gives root and Docker on every VM. | 122nd cheapest of 377 $1,560 a month | 320th of 377, with limits $1,119 a month | 4th cheapest of 377 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB Hyperstack fits a 24/7 dev box because its VM keeps its whole disk, while Baseten's CPU instances lose their files when they stop; bochka keeps the disk and charges nothing while stopped. | 115th cheapest of 368 $475 a month | 170th of 368, with limits $130 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 graphical desktop: Hyperstack gives SSH and text console logs, Baseten runs commands and file operations, and bochka streams a 60fps desktop with Chromium and computer use. | 332nd of 369, with limits $3,250 a month | 294th of 369, with limits $1,799 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 | 335th of 367, with limits $1,750 a month | 318th of 367, with limits $1,194 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 | 327th of 346, with limits $1,750 a month | 220th of 346, with limits $183 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 | 310th of 338, with limits $35,000 a month | 126th of 338, with limits $3,654 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 | 309th of 360, with limits $876 a month | 254th of 360, with limits $381 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, and here Hyperstack ranks 2nd: it rents whole GPU VMs by the minute, while Baseten bills GPU instances per minute including image builds and cold starts; bochka has no GPUs. | 2nd cheapest of 98 $625 a month | 24th cheapest of 98 $1,625 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 | Hyperstack | Baseten | bochka |
|---|---|---|---|
| Kind of product | GPU cloud: GPU and CPU VMs, managed Kubernetes, volumes, object storage | Inference cloud for serving and training models; agent sandboxes in private preview | Sandboxes for AI agents: VMs with agents, desktop and hosting ready |
| Isolation | Full VMs with root, on flavors mapped 1:1 to physical resources | Kubernetes containers in a dedicated namespace per customer; sandbox isolation is not documented | Every sandbox is a full VM with its own kernel |
| Root and Docker | Root and SSH; Ubuntu 22.04 and 24.04 images with or without Docker | Containers run as the non-root app user; Docker inside is not documented | Docker, Docker Compose, root, systemd and /dev/kvm on every machine |
| Files after a stop | The whole disk stays; a SHUTOFF VM bills at the full rate, a hibernated one bills only disk and a kept IP | Replicas start fresh from the image; preview sandboxes keep files on standby but are deleted after 7 days by default | A stopped sandbox costs nothing and keeps its whole disk |
| Snapshots | On request only, with the VM shut off, one at a time, up to 100 GB | No snapshots of model replicas | The disk is snapshotted every minute and on stop, while the sandbox keeps running |
| Fork | No fork; you snapshot, make a custom image and deploy a new VM | No fork or clone; new replicas cold-start from the image | Fork a running sandbox into a new one with the same disk |
| Scale to zero and wake on request | VMs never stop on their own when idle | Deployments scale to zero and start on the next request; standby sandboxes resume on the next request | No wake on request; you start a stopped sandbox through the API |
| GPUs | Nvidia RTX A4000 up to H100, H200, B200 and B300, on demand, spot or reserved, up to 8-GPU nodes | Nvidia T4 up to B200 and RTX PRO 6000, 1 to 8 per instance, multi-node | None |
| Desktop and computer use | SSH and text console logs; no graphical desktop | No desktop or computer-use API | 60fps desktop and Chromium, computer use and browser control, embeddable stream |
| Coding agents | A hosted MCP server lets your own agent manage Hyperstack; nothing runs an agent inside a VM | Baseten Switch points Claude Code or Codex on your laptop at its Model APIs | Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API |
| Reaching the machine | A public IPv4 you can keep static; you open ports with firewall rules and run your own TLS | An HTTPS API endpoint per deployment; sandbox ports through a /port/ route on its API server | HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4 |
| Network rules | Per-VM firewall: inbound closed by default, outbound rules by IP, protocol and port | Outbound on TCP 443 and 80 by default; IP and domain allowlists once enabled for your organization | Open internet out; inbound through HTTPS URLs and TCP ports |
| Keeping keys out of the machine | No credentials proxy; keys go in with cloud-init, files or environment variables | Secrets are passed into the model as arguments or files under /secrets | AI credentials proxy for model providers and any HTTPS API you register |
| Regions | Norway, Canada and the US | US and EU for GPU deployments; self-hosted in your own AWS, GCP or OCI account on Enterprise | EU only: Germany, Finland, France |
Every feature Battleships tracks (43)
| Feature | Hyperstack | Baseten | bochka |
|---|---|---|---|
| Full VM (own kernel) | Yes | No | Yes |
| Docker inside | Yes | No | Yes |
| Nested virtualization | Not published | No | 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 | No | Yes |
| Pause/resume | Yes | No | Yes |
| Automatic snapshots | No | No | Yes |
| Fork/clone | No | No | Yes |
| Live fork (no pause) | No | 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 | Yes | No |
| HTTPS preview URLs | No | Yes | Yes |
| Raw TCP inbound | No | Not published | Yes |
| Public IPv4 | Yes | No | 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 | Yes | Yes | No |
| GPU | Yes | Yes | No |
| Windows | No | No | No |
| Arm64 | No | No | No |
| Self-hosting | No | Yes | No |
| BYOC | No | Yes | No |
| Open source | No | No | No |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | Not published | No | Yes |
| Usage per sandbox (API) | No | Yes | Yes |
| Audit logs | Not published | Not published | No |
| SSO | No | Yes | No |
| SOC 2 | Yes | Yes | No |
| HIPAA | No | Yes | No |
| EU data residency | Yes | Yes | Yes |
| Zero data retention | Not published | Not published | Yes |
Performance
There is no comparable CPU measurement for Hyperstack or Baseten: neither is in the public StarSling sandbox benchmark run we use. bochka did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB in that run of 2026-09-29, on dedicated AMD Ryzen 9 9950X (Zen 5) hosts.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Hyperstack and Baseten 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 | Hyperstack, a month | Baseten, 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 | $114 · Prepaid on-demand | $34 · Basic · with limits | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $475 · Prepaid on-demand | $136 · Basic · 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 | $5,720 · Prepaid on-demand | $1,701 · Basic · 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 | $1,750 · Prepaid on-demand · with limits | $1,246 · Basic · 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 | $3,500 · Prepaid on-demand · with limits | $6,228 · Basic · with limits | $2,000 · $2000/mo · with limits |
Hyperstack: Hyperstack bills a running VM per minute from a prepaid balance, at a fixed hourly price per flavor, with no charge for bandwidth. The closest CPU flavor in the scenarios is 8 vCPU / 8 GB at $0.65 an hour, and a stopped VM keeps billing in full unless you hibernate it.
Baseten: Baseten bills per minute, partial minutes rounded up, from image build and cold start until the replica stops; a 4 vCPU / 16 GiB CPU instance is $0.00346 a minute. There is no plan fee on Basic, and sandbox pricing for the private preview is not published.
Baseten: 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 whole disk.
Limits in those scenarios
Hyperstack does not publish how many VMs can run at once, so 300 user sandboxes and 1,000 short jobs at once are not confirmed there; its API allows 500 requests a minute per IP. Baseten's CPU instances have no persistent disk, so files are lost when a replica stops and the table marks them as not fitting; its preview sandboxes run at most 10 per organization by default. A Baseten monthly budget only sends alerts for dedicated deployments, and a Hyperstack VM never stops on its own, so both keep billing until you act. On bochka the 1,000-job burst is queued and needs more starts per day than the $2,000 plan allows.
| Scenario | Hyperstack | Baseten | bochka |
|---|---|---|---|
| One coding agent, workdays | None that bite |
| None that bite |
| One agent that never stops | None that bite |
| None that bite |
| 50 agents in parallel |
|
| None that bite |
| Sandboxes for your users |
|
| None that bite |
| 1,000 short jobs at once |
|
|
|
Which one should you use?
Use Hyperstack when
you want plain GPU or CPU VMs with root and a kept disk, for training, fine-tuning or long jobs, in Norway, Canada or the US.
- A plain VM you fully control. Useful if you want a normal machine: Hyperstack gives full VMs with root, SSH, Docker images and a disk that stays.
- GPUs by the minute, up to 8-GPU nodes. Useful for training and fine-tuning: Hyperstack rents Nvidia GPUs up to 8-GPU B300 nodes, with spot and reserved prices too.
- Canada, firewalls and Kubernetes. Useful if you need a region in Canada, outbound rules by IP and port, or managed Kubernetes clusters.
Use Baseten when
you serve models in production and want deployments that scale to zero, hosted open models per token, or Baseten running in your own cloud.
- Scale to zero, wake on request. Useful for spiky traffic: Baseten deployments scale to zero when idle and start again on the next request.
- Model serving and hosted models. Useful if you ship a model: package it with Truss or a Docker image, or call hosted open models priced per token.
- Run it in your own cloud. Useful if your data must stay in your account: on Enterprise, Baseten can run in your own AWS, GCP or OCI account.
Where bochka fits
Hyperstack and Baseten are both built for GPU work: training, fine-tuning and serving models. If what you are really running is agents that write and run code for hours, neither gives each agent a ready machine. bochka does, and it pairs well with either one for the GPU part.
A full VM whose disk is saved every minute
Baseten's containers run as a non-root user and lose their files when they stop. Hyperstack VMs have root, but snapshots are taken by hand with the VM shut off, and a stopped VM bills in full unless you hibernate it.
Coding agents, a desktop and computer use built in
Neither has a desktop: Hyperstack gives SSH and a text console, Baseten runs commands and file operations. On both you install the agent, a browser and a screen stream yourself, then write the code that prompts and watches it.
Fork a running machine, keep keys off it
To copy a set-up machine on Hyperstack you shut it off, snapshot it, make an image and deploy a new VM; Baseten replicas cold-start from the image. Both pass keys onto the machine, where bochka's credentials proxy keeps them outside.
Use bochka when your agents write and run code: full VMs with root and Docker, a disk saved every minute and free stops, forks of a running sandbox, and coding agents and a desktop built in. Compare it one to one: bochka vs Hyperstack and bochka vs Baseten.
Questions people ask
Is Hyperstack cheaper than Baseten?
For CPU work Baseten looks lower in our scenarios: one agent running all month is $136 on Baseten and $475 on Hyperstack, but Baseten's instances lose their files when they stop, so they do not fit. For GPU eval jobs Hyperstack comes to $625 a month against $1,625 on Baseten. The same always-on agent is $26 a month on bochka.
Which is faster, Hyperstack or Baseten?
There is no comparable CPU measurement for either one, so we cannot say. bochka did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB in StarSling's public benchmark run.
Can I run Docker on Hyperstack or Baseten?
On Hyperstack, yes: VMs are full machines with root, and its Ubuntu images come with or without Docker. Baseten runs your Docker image as a model server, but its containers run as a non-root user and Docker inside is not documented. bochka sandboxes come with Docker, Docker Compose, systemd and /dev/kvm.
Do idle machines cost money?
On Hyperstack a stopped (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. Baseten deployments can scale to zero, but by default a replica stays billed for 15 idle minutes, and cold starts are billed. A stopped bochka sandbox costs nothing and keeps its disk.
Does Baseten have sandboxes for AI agents?
Yes, in private preview: you ask Baseten for access, and an organization can run up to 10 at once by default. They go on standby when idle and are deleted with their files after 7 days by default. Hyperstack has no sandbox product; you use its VMs.
Does bochka have GPUs like Hyperstack and Baseten?
No. bochka runs Linux VMs on dedicated AMD Ryzen 9 9950X hosts and has no GPUs. Run training or model serving on Hyperstack or Baseten, and let your agents work on bochka.
Try bochka
Install the CLI, then run bochka new.
Product preview