Full machines your agent can drive.
Full microVMs with their own kernel, a real desktop your agent can see, click and type on. Linux, Windows and macOS. Warm through a working day, paused to $0 between bursts, back in under 200ms with everything exactly where it was.
A whole desktop, as an API.
A container gives your agent a shell. A VM gives it a computer, a real windowing system it can see and operate. The same loop a person runs, driven from your code.
The pixels an agent reasons over.
Full-resolution screenshots on demand, or a live 60fps stream. The screen the agent reads is the screen it just acted on, input and video share one clock.
Move, click, type, at OS level.
Move, click, drag, type, key-combo and scroll against the real windowing system, not through a browser shim. Desktop apps, installers, legacy software, a browser, if it runs on a machine, the agent can drive it.
Hand the machine to a person, too.
Every VM streams its desktop at 60fps over WebRTC. Drop it into your product with one component, or send a signed link to a reviewer who takes the wheel for a moment, human-in-the-loop without an RDP client or a VPN.
RDP and VNC are there too if you would rather connect an existing client.
| WebRTC | 60 fps, low-latency, watch or drive |
| Scoped token | One VM, the scopes you set, dies with the TTL |
| RDP · VNC | Standard protocols for existing clients |
Stays warm. Costs nothing paused.
Pausing freezes exact in-memory state to disk and stops both meters. Resume returns in under 200ms with your windows still open, your cursor where it was, your session still logged in.
Linux, Windows and macOS, one API.
Switch the machine with one field. The SDK, the lifecycle, the billing meter and the computer-use primitives are identical across all three; only the desktop underneath changes.
Linux
Ubuntu 24.04 and friends, with or without a desktop. From 2 to 64 vCPU, its own kernel, per-second billing.
Windows
A real Windows 11 / Server 2025 desktop, the OS agents struggle to get anywhere else. RDP, legacy apps, the lot.
macOS
A macOS desktop on Apple silicon we own, one fixed instance, billed per minute. For Mac-only tools and Apple workflows.
A machine worth
keeping around.
A configured desktop, a logged-in session, an installed toolchain. Set it up once, then pause, resume, snapshot and fork it like a repo.
- 01
Pause to $0
Freeze RAM and disk, release the CPU. A paused VM is a snapshot, not a machine, and a snapshot is not billed.
- 02
Resume <200ms
Come back to the exact desktop, open apps, cursor position, in-memory state, from a warm claim off the pool.
- 03
Snapshot & fork
Branch a configured machine like a repo. Set one up once, fork it per task, and each child gets its own clock and tokens.
- 04
Volumes
Named storage that outlives any VM. Mount a home directory or a build cache and a fresh machine starts where the last left off.
Need short-lived, high-concurrency machines instead? Containers →
One balance. Every machine you run.
Virtual machines draw from the same credit pool as your containers, runners and simulators, billed by the second at one published rate card.
Starter
For a solo build or a first agent in production.
$20 of usage credits
≈ 550 VM-hours or 33,000 CI minutes
- 20 concurrent machines
- Committed in-plan rates on every product
- Linux & Windows, sandboxes, VMs, runners
- Per-second billing, paused is free
- Full API, SDKs, CLI & MCP server
- Unlimited seats
- Community support
Team
Apple silicon unlocked, and a balance that rolls over.
$100 of usage credits
≈ 2,700 VM-hours or 10,000 Mac build minutes
- 100 concurrent machines
- Everything in Starter, plus:
- macOS, iOS/Android simulators & Xcode builds
- Unused credits roll over one month
- Usage exports & per-second line items
- Email support
Business
Higher ceilings for a team leaning on the fleet daily.
$250 of usage credits
≈ 6,900 VM-hours or 25,000 Mac build minutes
- 300 concurrent machines
- Everything in Team, plus:
- Raised concurrency & start rate
- Priority support & shared Slack channel
- Single sign-on (SAML / OIDC)
- Volume credit rate
Enterprise
Dedicated capacity, your compliance surface, one invoice.
Committed volume rate
Committed volume rate, negotiated per account
- Reserved bare-metal capacity
- BYOC or self-host the control plane
- SSO & SCIM, audit logs, data residency
- Annual invoicing, MSA, DPA
- Support SLA & solutions engineering
- Custom regions on request
Every plan spends its credit pool first, then usage keeps flowing and bills per second at your committed in-plan rate, no hard stop mid-run, no forced upgrade. Without a plan you pay the on-demand rate instead. Prices are illustrative pending final list pricing; the per-second billing model is final.
The virtual machine rate card.
Linux
Windows
macOS
One fixed instance, billed at $0.01 / minute.Linux from $0.0065 / vCPU-hr + $0.00125 / GB-hr; Windows is the same machine plus licensing at 2x. macOS ships as one fixed Apple-silicon instance billed per minute. On-demand is 1.5× the in-plan rate. All usage bills by the second; paused machines are not billed.
Give your AI agents somewhere to run today.
Your agents get their own machines across Linux, macOS, Windows, iOS and Xcode in milliseconds, and you stop paying for idle.
