bochka vs Baseten

Baseten is an inference cloud: you deploy models on its GPUs, call hosted open models, and train, with agent sandboxes in private preview. bochka gives each agent a full VM with root, Docker, a desktop and coding agents built in, and its disk is kept with snapshots every minute.

Where bochka stands out

A full VM with root and Docker

On bochka every sandbox has its own kernel, with root, systemd, Docker and /dev/kvm ready. Baseten runs workloads as Kubernetes containers, and its model containers run as the non-root app user. It does not say what isolates its preview sandboxes.

A disk that stays, with snapshots and fork

A bochka sandbox keeps its disk, snapshotted every minute and on stop, and a stopped sandbox costs nothing. You can fork a running sandbox, so one set-up machine becomes many. Baseten replicas start fresh from the image, and its preview sandboxes are deleted with their files after 7 days by default.

Coding agents and a desktop built in

bochka ships Claude Code, Codex, OpenCode and other agents inside each sandbox, and you drive them from outside with a prompt, steer and events API. Every sandbox also has a 60fps desktop with Chromium and computer use. Baseten's coding-agent tools point an agent on your laptop at its hosted models; nothing runs the agent inside a Baseten machine for you.

The companies and products

bochka

bochka.xyz runs sandboxes for AI agents: each one is a full Linux VM with its own kernel.

You start a machine with one command or one API call, then SSH in, run Docker, open a desktop, or let a built-in agent (Claude Code, Codex, OpenCode and others) work in it. The disk stays between sessions, and a stopped sandbox costs nothing.

Isolation: Full VM per sandbox, own kernel
Hardware: Dedicated AMD Ryzen 9 9950X (Zen 5) hosts
Regions: EU: Germany, Finland, France
Plans: From $20 a month, all of it usage credit, 7-day trial

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.

Kind of product: Inference cloud: dedicated model deployments, Model APIs, training; Sandboxes in private preview
Isolation: Containers on Kubernetes, a dedicated namespace per customer, network isolation by Calico and Cilium; GPUs never shared across users
GPUs: Nvidia T4, L4, A10G, A100, H100, H100 MIG, H200, B200 and RTX PRO 6000, up to 8 per instance, plus multi-node
CPU instances: 1 vCPU / 2 GiB up to 16 vCPU / 64 GiB, on shared CPU instance types that cannot pick a region
Regions: United States and European Union for GPU deployments; self-hosted in your own AWS, GCP or OCI account on Enterprise
Billing: Per minute while a workload runs, including image builds and cold starts; no plan fee on Basic, starting credits for new workspaces
Compliance: Says it is SOC 2 Type II certified and HIPAA compliant

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. “3rd cheapest of 367” means 2 of the other 366 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	bochka: price rank	Baseten: price rank
Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A Baseten container runs as a non-root user with no documented Docker inside, so the installs, databases and test stacks a coding agent sets up fit better on a bochka VM.	3rd cheapest of 367 $634 a month	216th of 367, with limits $2,619 a month
Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB Baseten's preview sandboxes keep files on standby but are deleted after 7 days by default, so an assistant that lives for months needs bochka's kept disk or its own storage outside.	5th cheapest of 366 $657 a month	242nd of 366, with limits $3,258 a month
Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month Baseten bills per minute with no plan fee, but each burst also pays for cold start and, by default, 15 idle minutes before a replica scales down.	186th of 345, with limits $2,000 a month	319th of 345, with limits $1,856 a month
Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month Baseten's sandbox preview runs up to 10 sandboxes per organization at once by default, so 300 sandboxes for your users means asking support first.	Cheapest of 361 $209 a month	277th of 361, with limits $1,868 a month
RL rollouts 1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month	34th of 360, with limits $2,000 a month	322nd of 360, with limits $31,140 a month
Agent evals 250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month	10th cheapest of 364 $200 a month	297th of 364, with limits $1,609 a month
CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month A CI job needs Docker and root for most builds; Baseten's model containers run as the non-root app user, while bochka gives root and Docker on every VM.	4th cheapest of 377 $86 a month	320th of 377, with limits $1,119 a month
24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB	18th cheapest of 368 $26 a month	170th of 368, with limits $130 a month
Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Baseten has no desktop or computer-use API; bochka streams a 60fps desktop with Chromium and computer use from every sandbox.	6th cheapest of 369 $200 a month	294th of 369, with limits $1,799 a month
Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month	7th cheapest of 367 $200 a month	318th of 367, with limits $1,194 a month
Scraping & bots (startup scale) 50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month	6th cheapest of 346 $90 a month	220th of 346, with limits $183 a month
Mass scraping & bots (undetectable) 1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month	71st of 338, with limits $1,800 a month	126th of 338, with limits $3,654 a month
Parallel red teaming 200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month	2nd cheapest of 360 $200 a month	254th of 360, with limits $381 a month
GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month Only Baseten fits: it rents Nvidia GPUs from T4 to B200, up to 8 per instance, and bochka has no GPUs.	Can't run it	24th cheapest of 98 $1,625 a month
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

Key features

Feature	bochka	Baseten
Kind of product	Sandboxes for AI agents: VMs with agents, desktop and hosting ready	Inference cloud for serving and training models; agent sandboxes in private preview
Isolation	Every sandbox is a full VM with its own kernel	Kubernetes containers in a dedicated namespace per customer; sandbox isolation is not documented
Root, Docker and nested virtualization	Docker, root, systemd and /dev/kvm on every machine	Containers run as the non-root app user; Docker inside and /dev/kvm are not documented
Getting a sandbox	On every plan, self-serve, 100 to 2,000 at once by plan	Sandboxes in private preview, access on request, 10 per organization at once by default
Files after a stop	Snapshots every minute and on stop; a stopped sandbox costs nothing and keeps its disk	Model replicas start fresh from the image; preview sandboxes keep files on standby but are deleted with their data after 7 days by default
Fork	Fork a running sandbox into many	No fork or clone; new replicas cold-start from the image
Scale to zero and wake on request	No wake on request; you start a stopped sandbox through the API	Deployments scale to zero and start on the next request; standby sandboxes resume on the next request
GPUs	None	Nvidia T4 up to B200, 1 to 8 per instance, multi-node over InfiniBand
Desktop and computer use	60fps desktop and Chromium, computer use and browser control, embeddable stream	No desktop or computer-use API; sandboxes run commands and file operations
Coding agents	Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API	Baseten Switch routes Claude Code or Codex on your laptop to its Model APIs; an agent drives a sandbox through the CLI or MCP
Reaching the sandbox	HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4	HTTPS API endpoint per deployment; sandbox ports through a /port/ route on its API server; SSH once enabled for your organization
Outbound network	Open internet on every plan; no provider-side allowlist	TCP ports 443 and 80 by default; per-deployment IP and domain allowlists once enabled for your organization
Keeping keys out of the sandbox	AI credentials proxy for model providers and any HTTPS API you register	Secrets are passed into the model as arguments or files under /secrets
Regions and hosting	EU only: Germany, Finland, France	US and EU for GPU deployments; self-hosted or hybrid in your own cloud account on Enterprise
Compliance and transparency	SOC 2 is in progress. We're open about where we stand: live status and data retention policy.	SOC 2

Every feature Battleships tracks (43)

Feature	bochka	Baseten
Full VM (own kernel)	Yes	No
Docker inside	Yes	No
Nested virtualization	Yes	No
Root	Yes	Not published
Systemd	Yes	Not published
Desktop GUI	Yes	No
Browser	Yes	No
Computer-use API	Yes	No
SSH	Yes	Yes
Persistent disk	Yes	No
Pause/resume	Yes	No
Automatic snapshots	Yes	No
Fork/clone	Yes	No
Live fork (no pause)	Yes	No
Memory fork	No	No
Custom image (Docker or snapshot)	Yes	Yes
≥24 h sessions	Yes	Yes
Idle auto-stop	No	Yes
Wake on request	No	Yes
HTTPS preview URLs	Yes	Yes
Raw TCP inbound	Yes	Not published
Public IPv4	No	No
Egress allowlist	No	Yes
Secret proxy	Yes	No
Secret proxy for any API	Yes	No
Preinstalled agents	Yes	No
Agent harness API	Yes	No
MCP server	No	Yes
GPU	No	Yes
Windows	No	No
Arm64	No	No
Self-hosting	No	Yes
BYOC	No	Yes
Open source	No	No
Scoped API keys	Yes	Not published
Spend limits	Yes	No
Usage per sandbox (API)	Yes	Yes
Audit logs	No	Not published
SSO	No	Yes
SOC 2	No	Yes
HIPAA	No	Yes
EU data residency	Yes	Yes
Zero data retention	Yes	Not published

Performance

bochka26.69 runs/s

BasetenNot measured

There is no comparable measurement: Baseten was not 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. Baseten publishes no CPU benchmark for its CPU instances or its sandboxes.

Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Baseten has 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	bochka, a month	Baseten, 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	$20 · $20/mo	$34 · Basic · with limits
One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month	$26 · $20/mo	$136 · Basic · with limits
50 agents in parallel 50 machines of 4 vCPU / 8 GB at once, 8-hour sessions, 1,100 sessions a month, Docker inside	$317 · $20/mo × 3 seats	$1,701 · Basic · with limits
Sandboxes for your users 2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once	$200 · $100/mo × 2 seats	$1,246 · Basic · with limits
1,000 short jobs at once 2 vCPU / 4 GB, 1,000 at once, 3 minutes each, 200,000 jobs a month	$2,000 · $2000/mo · with limits	$6,228 · Basic · with limits

bochka: 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.

Baseten: The scenarios price Baseten's CPU-only instances, billed per minute with partial minutes rounded up, from image build and cold start until the replica stops; a 4 vCPU / 16 GiB instance is $0.00346 a minute. There is no plan fee on Basic, and sandbox pricing for the private preview is not published.

Limits in those scenarios

Baseten's CPU instances are model containers with no persistent disk, so in every scenario files are lost when a replica stops or scales to zero, and the table marks them as not fitting. Its preview sandboxes keep files on standby, but run at most 10 per organization by default and are deleted after 7 days unless you set another policy. A monthly budget only sends alerts for dedicated deployments; it never stops them. On bochka the limit is a sandbox count by plan plus starts per minute, so the 1,000-job burst is queued and needs more starts per day than the $2,000 plan allows.

Scenario	bochka	Baseten
One coding agent, workdays	None that bite	adjacent product, not a like-for-like sandbox missing persistent disk different kind of product: app platform no persistent disk: files lost when it stops or sleeps
One agent that never stops	None that bite	adjacent product, not a like-for-like sandbox missing persistent disk different kind of product: app platform no persistent disk: files lost when it stops or sleeps
50 agents in parallel	None that bite	adjacent product, not a like-for-like sandbox missing persistent disk, Docker inside different kind of product: app platform no persistent disk: files lost when it stops or sleeps
Sandboxes for your users	None that bite	adjacent product, not a like-for-like sandbox different kind of product: app platform disk beyond 0 GiB: price unknown egress price not published
1,000 short jobs at once	max 2,400 starts/day (needs ~6,667) max 600 starts/hour (needs ~1,000) bursts above 90 starts/min are queued (a burst of 333 takes ~4 min to start)	adjacent product, not a like-for-like sandbox different kind of product: app platform disk beyond 0 GiB: price unknown egress price not published

Which one should you use?

Use Baseten when

you serve or train models on GPUs, want deployments that scale to zero and wake on request, or must run in the US or in your own cloud account.

GPUs for serving and training models. Useful if your product serves or trains models: Baseten runs Nvidia GPUs from T4 to B200, up to 8 per instance.
Scale to zero, wake on request. Useful for spiky traffic: Baseten deployments scale to zero when idle and start again on the next request.
US region, your own cloud, egress allowlists. Useful if you need the US, your own cloud account, or per-deployment outbound allowlists set by the provider.

Use bochka when

you run agents that build and test software: full VMs with root and Docker, coding agents and a desktop built in, a disk that is snapshotted every minute, and forks of a running sandbox.

A full VM with root and Docker. Coding agents install packages, start databases with Docker and run emulators. In a non-root container most of that fails, and a shared kernel is a way out for code an agent wrote.
A disk that stays, with snapshots and fork. Agents work across days and branch their work. On Baseten you copy results out before the deadline and rebuild the environment by hand, because nothing clones a running machine.
Coding agents and a desktop built in. Without these you write and run the agent loop, the browser and any desktop stream yourself. bochka's credentials proxy also keeps real AI keys off the machine, where Baseten passes secrets into the container.

Questions people ask

Is bochka cheaper than Baseten?

In every scenario we priced, using Baseten's CPU instances: one coding agent on workdays comes to $20 a month on bochka and $34 on Baseten, and one agent running all month is $26 versus $136. 50 agents in parallel are $317 versus $1,701, and 300 sandboxes for your users are $200 versus $1,246. For 1,000 short jobs at once bochka comes to $2,000 and Baseten to $6,228, and neither fits without higher limits.

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. A sandbox runs commands and file operations from an image, goes on standby when idle, and is deleted with its files after 7 days by default. Pricing for it is not published yet.

Does bochka have GPUs like Baseten?

No. bochka runs Linux VMs on dedicated AMD Ryzen 9 9950X hosts and has no GPUs. If your product serves or trains models, run that part on Baseten and let your agents work on bochka.

Can I run Docker inside Baseten?

Baseten does not document Docker inside its containers, and its model containers run as the non-root app user. You can bring your own Docker image for Baseten to run as a model server. Every bochka sandbox has root and Docker.

Where does my data run?

bochka runs only in the EU (Germany, Finland and France), with EU data residency and a zero data retention option. Baseten lets GPU deployments pick a US or EU region, its shared CPU instances cannot pick one, and Enterprise customers can self-host in their own cloud.

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.

Start a sandbox

Install the CLI, then run bochka new.

Product preview