bochka vs MIOSA
Both give each agent its own Linux machine with Docker, a desktop and a built-in agent API. bochka gives it a full VM that can run Android emulators and keep going for weeks in the EU, while MIOSA is strongest at short sessions that pause with their memory and stay inside strict outbound rules.
Where bochka stands out
Nested virtualization in a full VM
A bochka sandbox is a full VM with its own kernel, root, systemd, Docker and /dev/kvm. That means Android emulators, QEMU, Firecracker and other VMs start inside it. MIOSA's docs say KVM and QEMU cannot run inside its sandboxes.
No timeout on long work
A bochka sandbox has no session cap on paid plans: it runs until you stop it or its time to live ends. MIOSA sandboxes stop after their timeout, 1 hour by default and 24 hours at most, and new accounts also pause after 30 idle minutes unless you change it.
EU hosting with zero data retention
bochka runs in the EU, in Germany, Finland and France, and offers zero data retention. MIOSA's three regions are all in the US, and its docs say there is no published zero-retention mode.
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
- Regions
- EU: Germany, Finland, France
MIOSA
MIOSA calls itself infrastructure for agent-built software: sandboxes, cloud desktops, deployments and data services in one platform.
A MIOSA Sandbox is a Firecracker microVM in one of five fixed sizes, with Docker preinstalled and a timeout of up to 24 hours. A Computer is a durable Linux desktop with a live pixel stream for computer-use agents, and the same account also hosts apps, Postgres, Redis and object storage. It runs in its own US data centers and is built for teams that want agents, previews and deploys from one API.
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 | MIOSA: price rank |
|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Both run Docker, but builds that need an Android emulator or a VM need /dev/kvm, which bochka has and MIOSA's docs rule out. | 3rd cheapest of 367 $634 a month | 13th cheapest of 367 $3,761 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB An assistant that lives on its machine all month runs into MIOSA's 24-hour timeout and has to be extended or resumed; bochka can simply keep running. | 5th cheapest of 366 $657 a month | 48th of 366, with limits $4,876 a month |
| Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month MIOSA fits short code runs well: sandboxes start fast and can pause with memory and processes intact for up to 30 minutes. | 186th of 345, with limits $2,000 a month | 43rd cheapest of 345 $500 a month |
| Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month Both give each user a machine with preview URLs; MIOSA adds outbound rules per sandbox, while bochka adds raw TCP, a dedicated IP and no timeout. | Cheapest of 361 $209 a month | 135th of 361, with limits $1,614 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 | 120th of 360, with limits $7,719 a month |
| Agent evals 250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month Both can fork a running sandbox for evals; bochka runs each eval for as long as it needs, and MIOSA can pause a forked one with its memory. | 10th cheapest of 364 $200 a month | 60th cheapest of 364 $1,244 a month |
| CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month CI runners that build Android apps or VM images need nested virtualization, which bochka has and MIOSA does not. | 4th cheapest of 377 $86 a month | 79th cheapest of 377 $755 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB A dev box that stays on all month needs no session timeout and a stable IP, which bochka has and MIOSA does not. | 18th cheapest of 368 $26 a month | 196th of 368, with limits $500 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Both have a Linux desktop for computer-use agents; MIOSA's is one 1920 x 1080 screen without audio, and bochka streams at 60fps with Chromium built in. | 6th cheapest of 369 $200 a month | 25th cheapest of 369 $1,132 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 | 89th of 367, with limits $574 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 | 51st cheapest of 346 $500 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 | 130th of 338, with limits $3,789 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 | 7th cheapest of 360 $500 a month |
| GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month | Can't run it | 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 |
Key features
| Feature | bochka | MIOSA |
|---|---|---|
| Isolation | Full VM with its own kernel on dedicated AMD Ryzen 9 9950X hosts | Firecracker microVM with its own kernel |
| Nested virtualization | Yes, /dev/kvm for Android emulators and VMs | Not supported; KVM and QEMU cannot run inside a sandbox |
| Docker inside | Yes, with root and systemd | Preinstalled in every sandbox; build, buildx and compose work |
| Session length | No cap on paid plans; you set the time to live | Timeout up to 24 hours, 1 hour by default; a persistent sandbox pauses when it ends |
| Pause and resume | Disk is saved every minute and on stop; resume boots the machine again | Pause keeps memory and processes for up to 30 minutes, then falls back to the disk |
| Fork | Fork a running sandbox with its disk | Copy-on-write fork of a running sandbox into a new one |
| Machine sizes | Up to 16 vCPU / 32 GB on the $100 plan and above, disk sized separately | Five fixed sizes up to 16 vCPU / 32 GiB; disk is set by the size (10 GiB on the default) |
| Desktop and computer use | Linux desktop streamed at 60fps, with Chromium and computer use | Computers: a durable Linux desktop with a live pixel stream, one 1920 x 1080 screen, video only |
| Inbound access | HTTPS URLs for up to 50 ports, token-protected by default; raw TCP; dedicated IPv4 or IPv6 | HTTPS preview URL per port, public or private with share tokens; custom domains; no raw TCP documented |
| Outbound rules | Open internet, no provider-side allowlist | Allow and deny rules by host, wildcard or CIDR, with HTTP method and path, plus an audit log |
| Coding agents | Seven agents preinstalled, with a prompt, steer and events API | Agent Runs API sends a prompt to OSA, Codex, Claude Code, Pi, Hermes or a custom runtime, with streamed events |
| Keeping API keys out of the sandbox | Credentials proxy for model providers and any HTTPS API; switch a credential to another account when credits run out | Host proxy swaps a placeholder for the real key on any HTTP API; OAuth tokens refresh, rotation without restart |
| Regions and data retention | EU only (Germany, Finland, France), with zero data retention | Three US regions; no published zero-retention mode |
| Bring your own cloud | No | Yes, over your AWS, GCP or Azure account; sandboxes in private preview |
| 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 | MIOSA |
|---|---|---|
| Full VM (own kernel) | Yes | Yes |
| Docker inside | Yes | Yes |
| Nested virtualization | Yes | No |
| Root | Yes | Yes |
| Systemd | Yes | Yes |
| Desktop GUI | Yes | Yes |
| Browser | Yes | Yes |
| Computer-use API | Yes | Yes |
| SSH | Yes | Yes |
| Persistent disk | Yes | Yes |
| Pause/resume | Yes | Yes |
| Automatic snapshots | Yes | Yes |
| Fork/clone | Yes | Yes |
| Live fork (no pause) | Yes | Yes |
| Memory fork | No | Yes |
| 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 | Yes |
| Secret proxy for any API | Yes | Yes |
| Preinstalled agents | Yes | Yes |
| Agent harness API | Yes | Yes |
| MCP server | No | Yes |
| GPU | No | No |
| Windows | No | No |
| Arm64 | No | No |
| Self-hosting | No | No |
| BYOC | No | Yes |
| Open source | No | No |
| Scoped API keys | Yes | Not published |
| Spend limits | Yes | Not published |
| Usage per sandbox (API) | Yes | Yes |
| Audit logs | No | Yes |
| SSO | No | No |
| SOC 2 | No | Yes |
| HIPAA | No | Yes |
| EU data residency | Yes | No |
| Zero data retention | Yes | No |
Performance
There is no comparable measurement. bochka did 26.69 runs per second on the StarSling hpc-sandbox-benchmarks Node.js loop at 4 vCPU / 8 GB, but MIOSA has no run in that benchmark, so the two cannot be compared on CPU speed.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. MIOSA 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 | MIOSA, 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 | $500 · Business |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $26 · $20/mo | $500 · Business · 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 | $2,463 · Business |
| 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 | $713 · Business · 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 | $1,546 · Business · with limits |
bochka: Plans start at $20 a month, and the whole fee is usage credit. You pay per second while a sandbox runs, and a stopped sandbox costs nothing and keeps its disk.
MIOSA: MIOSA bills per second for the vCPU, RAM and disk of the size you pick, idle time included, while a sandbox runs; a paused one bills its disk at a cold-storage rate. Business has a $500 monthly minimum that becomes usage credit, and the free Developer plan gives a one-time $10 credit.
Limits in those scenarios
MIOSA's 24-hour timeout is why it does not fit the always-on scenario: an agent that never stops has to have its timeout extended over and over. Disk comes with the size, so a medium sandbox has 20 GiB and more disk means a bigger size. MIOSA's docs now say Docker is preinstalled in every sandbox, so the limits that bite are disk and Business's cap of 250 sandboxes at once, which the per-user sandboxes and 1,000 short jobs scenarios go past. bochka has no timeout, but at 1,000 short jobs at once its $2,000 plan queues starts above 90 a minute.
| Scenario | bochka | MIOSA |
|---|---|---|
| One coding agent, workdays | None that bite |
|
| One agent that never stops | None that bite |
|
| 50 agents in parallel | None that bite | None that bite |
| Sandboxes for your users | None that bite |
|
| 1,000 short jobs at once |
|
|
Which one should you use?
Use MIOSA when
your data must stay in the US, you need outbound rules with an audit log, or your sessions are short and benefit from pausing with their memory.
- Memory pause for short sessions. Useful for sessions that sleep between steps: a paused sandbox keeps its memory and processes for up to 30 minutes.
- Outbound rules with an audit log. MIOSA can limit which hosts, methods and paths a sandbox reaches, and log every outbound call for 90 days.
- One platform, US regions or your own cloud. Teams in the US can run sandboxes, desktops, deploys and databases from one account, or over their own AWS, GCP or Azure.
Use bochka when
your agents do real work for hours or days: installs, builds, Docker, Android emulators and a desktop, on a full VM in the EU that never times out.
- Nested virtualization in a full VM. An agent that builds and tests an Android app, or boots a VM image to check it, has nowhere to run that step on MIOSA and needs a second machine somewhere else.
- No timeout on long work. An agent that runs for days has to keep extending its MIOSA timeout from your code, and a missed call pauses it mid-task. On bochka there is nothing to babysit.
- EU hosting with zero data retention. Teams with EU customers or data that must stay in Europe cannot place it in MIOSA today, since its EU region is only on the roadmap.
Questions people ask
Is bochka cheaper than MIOSA?
In four of the five scenarios on this page, yes. For one coding agent on workdays, bochka comes to $20 a month and MIOSA to $500 a month, because MIOSA's Business plan has a $500 minimum. For 1,000 short jobs at once MIOSA comes out lower, though neither fits that scenario fully.
Can I run Docker in a MIOSA sandbox?
Yes. MIOSA's docs say Docker is preinstalled in every sandbox, with build, buildx and compose working. bochka runs Docker too, and also gives you systemd and /dev/kvm.
Can MIOSA run Android emulators or VMs?
No. MIOSA's known limitations say KVM and QEMU cannot run inside a sandbox. bochka exposes /dev/kvm, so Android emulators and nested VMs run inside the sandbox.
Can a MIOSA sandbox run for days?
Not in one session. The timeout is 1 hour by default and 24 hours at most; a persistent sandbox pauses when it ends and you can extend the timeout by API. bochka has no session cap on paid plans.
Does MIOSA have an EU region?
Not yet. Its three live regions are in Los Angeles, New York and Miami, and EU and APAC are on its roadmap. bochka runs in Germany, Finland and France.
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