bochka vs Jarvislabs

Jarvislabs is a GPU cloud: you rent Nvidia GPUs, GPU containers and plain CPU virtual machines by the minute, mostly in India. bochka is built for agents, with coding agents, a 60fps desktop, minute-by-minute snapshots and free stops on every VM in the EU.

Where bochka stands out

Free stops, snapshots every minute, live fork

A stopped bochka sandbox costs nothing, and its disk is snapshotted every minute and on stop. You can fork a sandbox while it runs, so one set-up machine becomes many. Jarvislabs has no snapshot, clone or fork command: a paused instance keeps its disk and keeps paying for it.

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. A Jarvislabs CPU VM is a clean Linux machine: you install the agent, a browser and any desktop yourself.

HTTPS ports and keys out of the sandbox

Each bochka sandbox can serve HTTPS on up to 50 ports behind a token, with no certificates to set up. bochka's credentials proxy keeps real AI keys out of the sandbox and works with any HTTPS API you register. A Jarvislabs VM gets a public IP with every port open until you add a security group, and your keys live on the machine as environment variables.

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

Jarvislabs

Jarvislabs is an AI cloud from India that rents Nvidia GPUs by the minute for training and serving models.

You can rent a GPU VM with root and your own kernel, a Template (a GPU container with PyTorch, ComfyUI or similar ready, opened in JupyterLab, VS Code or SSH), a plain CPU VM, or a multi-node GPU cluster. It also serves open models through Managed Endpoints, Serverless workers and a token-priced Model API. It is built for people who train, fine-tune and serve AI models, not as a sandbox for agents, though its jl CLI is written so a coding agent on your laptop can drive the GPUs.

Kind of product: GPU cloud: GPU VMs, GPU containers (Templates), CPU VMs, GPU clusters and model serving
Founded: 2021; says it has served 27,000+ AI developers and 50M+ GPU hours
Isolation: GPU VMs are KVM virtual machines with direct GPU passthrough; Templates are containers
Regions: India (IN1, IN2) for new instances; the EU1 region no longer accepts new instances; CPU VMs only in Noida
GPUs: Nvidia H200, H100, RTX Pro 6000 Blackwell, A100 80GB and 40GB, A30 and L4, 1 to 8 per instance; B200 as reserved clusters
CPU machines: 2 to 32 vCPU with 4 GB of RAM per vCPU, on-demand only
Billing: Per minute from a prepaid wallet, $10 minimum top-up, no free trial

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	Jarvislabs: price rank
Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A Jarvislabs CPU VM gives a coding agent root and Docker, but you install the agent and tooling yourself, and there are no snapshots or forks to branch work.	3rd cheapest of 367 $634 a month	154th of 367, with limits $1,406 a month
Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB An always-on assistant fits on a Jarvislabs CPU VM, but the browser, the desktop and the agent are yours to install; bochka ships them ready.	5th cheapest of 366 $657 a month	185th of 366, with limits $1,670 a month
Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month Jarvislabs bills small CPU VMs by the minute with no plan fee, which suits short bursts better than bochka's start limits, though each VM takes about 90 seconds to boot.	186th of 345, with limits $2,000 a month	103rd of 345, with limits $169 a month
Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month Jarvislabs does not publish how many instances an account can run at once, so 300 sandboxes for your users means asking first.	Cheapest of 361 $209 a month	215th of 361, with limits $539 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	43rd of 360, with limits $2,568 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	113th of 364, with limits $411 a month
CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month Both fit: a Jarvislabs CPU VM makes a plain CI runner, and bochka adds per-second billing and stops that cost nothing between jobs.	4th cheapest of 377 $86 a month	20th cheapest of 377 $253 a month
24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB	18th cheapest of 368 $26 a month	52nd cheapest of 368 $68 a month
Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Jarvislabs has no desktop or computer-use API, only JupyterLab, VS Code in the browser and SSH, while bochka streams a 60fps desktop with computer use.	6th cheapest of 369 $200 a month	154th of 369, with limits $377 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	204th of 367, with limits $205 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	244th of 346, with limits $247 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	158th of 338, with limits $4,942 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	156th of 360, with limits $127 a month
GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month Only Jarvislabs fits: it rents Nvidia GPUs from L4 to H200 by the minute, and bochka has no GPUs.	Can't run it	9th cheapest of 98 $876 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	Jarvislabs
Kind of product	Sandboxes for AI agents: VMs with agents, desktop and hosting ready	GPU cloud: GPU VMs, GPU containers, CPU VMs, clusters and model serving
Isolation	Every sandbox is a full VM with its own kernel	GPU VMs and CPU VMs are KVM virtual machines; Templates are containers
Docker and nested virtualization	Docker, root, systemd and /dev/kvm on every machine	Root on VMs, so you can run your own Docker daemon and kernel modules; nested virtualization is not documented
GPUs	None	Nvidia L4 up to H200, 1 to 8 per instance, on-demand, spot (Templates) or reserved, plus clusters
Biggest CPU machine	16 vCPU / 32 GB on the $100 plan and up	CPU VMs up to 32 vCPU / 128 GB
Stopping a machine	A stopped sandbox costs nothing and keeps its disk	Pause stops compute billing; the disk keeps billing at $0.10/GB-month, and VMs need at least 100 GB
Snapshots and fork	Snapshots every minute and on stop; fork a running sandbox	No snapshot, clone or fork command; pause and resume keeps the disk (VMs) or only /home (containers)
Desktop and computer use	60fps desktop and Chromium, computer use and browser control, embeddable stream	No desktop; JupyterLab, VS Code in the browser and SSH
Coding agents	Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API	A CLI and skill file so an agent on your own machine can create GPUs, run jobs and read logs
Reaching the machine	HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4	VMs get a public IP with every port open by default; Templates expose chosen HTTP ports through proxy URLs
Network rules	Open internet on every plan; no provider-side allowlist	Security groups with inbound and outbound rules by port and IP range, VPCs and reserved IPs, on VMs in India
Keeping keys out of the machine	AI credentials proxy for model providers and any HTTPS API you register	Keys are passed in as environment variables or uploaded files
Scale to zero	No wake on request; you start a stopped sandbox through the API	Serverless model workers stop after an idle timeout and start again on the next request
Regions	EU only: Germany, Finland, France	India only for new instances (IN1, IN2)
Compliance and transparency	SOC 2 is in progress. We're open about where we stand: live status and data retention policy.	No SOC 2 report published

Every feature Battleships tracks (43)

Feature	bochka	Jarvislabs
Full VM (own kernel)	Yes	Yes
Docker inside	Yes	Yes
Nested virtualization	Yes	Not published
Root	Yes	Yes
Systemd	Yes	Not published
Desktop GUI	Yes	No
Browser	Yes	No
Computer-use API	Yes	No
SSH	Yes	Yes
Persistent disk	Yes	Yes
Pause/resume	Yes	Yes
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	Yes
Public IPv4	No	Yes
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	No
GPU	No	Yes
Windows	No	No
Arm64	No	No
Self-hosting	No	No
BYOC	No	No
Open source	No	No
Scoped API keys	Yes	Not published
Spend limits	Yes	Yes
Usage per sandbox (API)	Yes	Yes
Audit logs	No	Not published
SSO	No	No
SOC 2	No	No
HIPAA	No	No
EU data residency	Yes	No
Zero data retention	Yes	Not published

Performance

bochka26.69 runs/s

JarvislabsNot measured

There is no comparable measurement: Jarvislabs 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. Jarvislabs publishes GPU benchmarks for its H200 VMs, but no CPU benchmark for its CPU VMs.

Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Jarvislabs 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	Jarvislabs, 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	$21 · On-demand
One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month	$26 · $20/mo	$71 · On-demand
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	$892 · On-demand
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	$231 · On-demand · 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	$514 · On-demand · 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.

Jarvislabs: Jarvislabs bills per minute from a prepaid wallet with no plan fee. CPU VMs cost $0.012 per vCPU-hour plus $0.0032 per GB of RAM, with 4 GB of RAM per vCPU, and the disk is extra at $0.10/GB-month, billed while running and while paused, with at least 100 GB on a VM.

Limits in those scenarios

Jarvislabs does not publish how many instances an account can run at once, so 300 sandboxes for your users or 1,000 short jobs at once are not confirmed until you ask. A paused VM, including a CPU VM, keeps its whole disk; only GPU containers (Templates) lose everything outside /home. Its smallest CPU VM is 2 vCPU / 8 GB, and each VM takes about 90 seconds to boot. 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	Jarvislabs
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	how many machines can run at once: limit not published, probably fine at 50 (the typical published limit is 50)
Sandboxes for your users	None that bite	not confirmed: how many machines can run at once (you need 300; the typical published limit is 50)
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)	not confirmed: how many machines can run at once (you need 1,000; the typical published limit is 50)

Which one should you use?

Use Jarvislabs when

your work needs Nvidia GPUs, model serving that scales to zero, or machines in India.

Nvidia GPUs by the minute. Useful if your work trains, fine-tunes or serves models: Jarvislabs rents Nvidia GPUs from L4 to H200 by the minute.
Model serving that scales to zero. Useful for inference: Jarvislabs serves open models on GPU workers that stop when idle and start on the next request.
Network rules and India regions. Useful for teams in India, or who need firewall rules set by the provider: Jarvislabs has security groups, VPCs and reserved IPs.

Use bochka when

you run agents: VMs that come with coding agents, a desktop, HTTPS ports and a credentials proxy, that snapshot every minute, fork while running and cost nothing while stopped.

Free stops, snapshots every minute, live fork. Agents sit idle between tasks and often need to branch work. On Jarvislabs every paused VM still bills at least 100 GB of disk, and copying a set-up machine means building it again by hand.
Coding agents and a desktop built in. Without these, you write and maintain the setup scripts, the agent runner and the desktop stream for every machine. Jarvislabs' CLI lets an agent on your laptop manage machines, not work inside one.
HTTPS ports and keys out of the sandbox. An agent that holds a real API key can leak it in a log or a commit. On Jarvislabs the key has to sit on the machine, and every preview server needs its own TLS and firewall rules.

Questions people ask

Is bochka cheaper than Jarvislabs?

It depends on the scenario. One coding agent on workdays comes to $20 a month on bochka and $21 on Jarvislabs, one agent running all month is $26 versus $71, and 50 agents in parallel are $317 versus $892. For 1,000 short jobs at once Jarvislabs comes to $514 and bochka to $2,000, though neither fits without higher limits.

Does bochka have GPUs like Jarvislabs?

No. bochka runs Linux VMs on dedicated AMD Ryzen 9 9950X hosts and has no GPUs. If your agents need to train or serve models, rent the GPUs on Jarvislabs.

Does a paused Jarvislabs instance cost money?

Yes, for storage. Pausing stops compute billing, but the disk keeps billing at $0.00014 per GB-hour, and VMs need at least 100 GB. A stopped bochka sandbox costs nothing and keeps its disk.

What happens when a Jarvislabs balance runs out?

Jarvislabs' FAQ says instances pause automatically when the balance reaches zero, and if the balance goes negative all data is permanently deleted and cannot be recovered. Keep enough credit in the wallet to cover storage for paused machines.

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. Jarvislabs creates new instances only in its India regions; its old EU1 region no longer accepts new instances.

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