No subscription. Pay what runs.

Q: Is there a monthly plan?

No. There's no subscription to pick, no seat count, and no monthly minimum. You pay only for the VMs that run — stop a VM and the meter stops.

Q: What does it cost to get started?

Top up to start — pay-as-you-go from there. A static site runs around $1.93 a month, a small API about $5.10. The meter stops the moment a VM stops.

Q: Can I resize a running VM?

Yes — one call changes a VM's CPU and memory in place, with zero downtime, zero broken connections, and zero agent retries. No reboot, no new IP, no rebuild. Start small and grow when you actually need to.

Q: Will I overpay for an oversized VM?

Only if you ask for one. You choose the CPU and memory, and you're billed per second for exactly that — not for a tier rounded up around it. If it's too big, resize it down live; if you're not using it at all, stop it and it costs nothing.

Q: How is usage metered?

Every VM is metered to the second from boot to stop. Compute bills at flat unit rates — $5.00 per vCPU-month and $2.50 per GB-month — prorated to the second, so there's no rounding to the hour and you never pay for stopped or destroyed VMs.

Q: Are backups included?

Daily snapshots are included on every VM with no extra charge. You can take ad-hoc snapshots at any time and save them as base templates for future VMs.

Designed for how agents actually work — bursty, parallel, experimental. No plan tier, no seats, no monthly minimum. Stop a VM and it costs nothing. Daily backups included.

$0^/MO

MONTHLY MINIMUM
NO SUBSCRIPTION TO BE TIED INTO

0^MS

DOWNTIME WHEN SCALING
LIVE CPU + MEMORY RESIZE

<3^SEC

TO A RUNNING VM
FIRECRACKER MICROVM, NOT A CONTAINER

// unit rates live rates

Resource	Rate	Notes
vCPU	$5.00 / vCPU · mo	Shared / burstable. Billed per second of actual use.
Memory	$2.50 / GB · mo	You choose the allocation. Resize it live, no reboot.
Disk	included	Rootfs is not separately billed. Add volumes for more.
Egress	$0.01 / GiB	Outbound only — inbound is free.

No tiers to buy. You pick the CPU and memory you want, resize them live whenever you want, and pay for exactly that — per second. Stop a VM and it costs nothing until you start it again. Set a monthly spend cap on your account and you can't be surprised. Daily backups included.

Egress is metered at $0.01 / GiB of outbound traffic — inbound is free.

Your VMs run in the EU.

What real workloads cost.

No plan to guess at. Here's a sensible starting size for a few common shapes and what each costs to run — billed by the second, and resizable live the moment the workload changes.

// what real workloads cost live rates

Workload	Size	Est. / mo
Static site A static site or blog on an always-on micro VM — the cheapest way to stay online.	0.25 vCPU · 256 MiB	~$1.93
Small API A small always-on API or webhook backend serving steady traffic.	0.5 vCPU · 1 GiB	~$5.10
CI runner A high-RAM build runner during business hours.	4 vCPU · 8 GiB	~$13.53
Next.js app A server-rendered Next.js app with a build step and moderate traffic.	2 vCPU · 2 GiB	~$15.20
Dev box A cloud machine you code in all day — VS Code Remote, always on.	2 vCPU · 4 GiB	~$20.05

Each row is a starting size and what it costs to run — compute plus egress, billed per second. Pick a size at create, change it live with one call, and stop the VM when you're not using it — a stopped VM costs nothing. Hobby projects land around $1.93–$5.10/mo — no plan to pick, no minimum.

Every project can call a model.

No provider keys, no separate bill — usage bills per token against your Superjolt balance at published, per-model rates. A cheap default like gpt-4o-mini stretches furthest; reach for a frontier model only where quality matters.

// built-in models & rates live rates

Model	Provider	$ / M input tok	$ / M output tok	Context
gemini-2.5-flash-lite	google	$0.12	$0.48	1M
gpt-4o-mini	openai	$0.18	$0.72	128K
gemini-2.5-flash	google	$0.36	$3.00	1M
claude-haiku-4-5	anthropic	$0.96	$4.80	200K
gemini-2.5-pro	google	$1.50	$12.00	2M
gpt-4o	openai	$3.00	$12.00	128K
claude-sonnet-4-6	anthropic	$3.60	$18.00	200K
claude-opus-4-7	anthropic	$18.00	$90.00	200K

Billed per token against your balance — rates include our margin, so there's no separate provider bill and no keys to manage. Point any OpenAI-compatible library at the model URL injected into every VM. Exact live figures are always at llm_list_models / GET /v1/llm/models.

The inbox is free within caps.

Every project gets a zero-config sandbox inbox — free, with a small storage cap and sending rate-limited to 20/hour, 100/day. Connect your own domain for production and outbound bills through your own provider account (SendGrid, SES, Mailgun), not us. Larger inboxes and higher sending rates are a planned paid tier.

What will it cost to run your app?

Pick a workload, tweak the runtime, and watch the monthly figure. You'll see the size we'd start it at and what that size costs. Every number comes straight from our live rates.

// what will it cost to run your app? live rates

Pick a workload

Hours per day 24h

Billed per second — a stopped VM costs nothing.

Egress (GiB / month)

A static site or blog on an always-on micro VM — the cheapest way to stay online.

Size0.25 vCPU · 256 MiB

Compute$1.88

Egress$0.05

Daily backupsincluded

Est. monthly $1.93

That's a sensible starting size for this workload — you pick it at create and change it live with one call, no reboot. Billing is per-second against the size you're on, so a smaller VM simply costs less. Egress is metered on top.

Compare the total, not the box.

Most clouds sell you a fixed box — a set number of vCPUs and gigabytes, rounded up to the next tier, running whether you use it or not. Here the size is yours: pick the exact CPU and memory you want, pay per second for exactly that, resize it live when the workload changes, and stop it when you're done and pay nothing at all.

Cheaper here isn't a spec-sheet trick — it's no tier to round up into, no idle box you forgot to turn off, and a unit rate a fraction of the big clouds'. We quote the cost before you deploy, and run it on compute we own and isolate.

The fine print.

Is there a monthly plan?

No. There's no subscription to pick, no seat count, and no monthly minimum. You pay only for the VMs that run — stop a VM and the meter stops.

What does it cost to get started?

Top up to start — pay-as-you-go from there. A static site runs around $1.93 a month, a small API about $5.10. The meter stops the moment a VM stops.

Can I resize a running VM?

Yes — one call changes a VM's CPU and memory in place, with zero downtime, zero broken connections, and zero agent retries. No reboot, no new IP, no rebuild. Start small and grow when you actually need to.

Will I overpay for an oversized VM?

Only if you ask for one. You choose the CPU and memory, and you're billed per second for exactly that — not for a tier rounded up around it. If it's too big, resize it down live; if you're not using it at all, stop it and it costs nothing.

How is usage metered?

Every VM is metered to the second from boot to stop. Compute bills at flat unit rates — $5.00 per vCPU-month and $2.50 per GB-month — prorated to the second, so there's no rounding to the hour and you never pay for stopped or destroyed VMs.

Are backups included?

Daily snapshots are included on every VM with no extra charge. You can take ad-hoc snapshots at any time and save them as base templates for future VMs.

Spin up your first VM.

Say “Add Superjolt MCP at https://mcp.superjolt.com”, sign in when prompted, then “hello superjolt” — your agent boots a live demo VM you can edit, then top up and ship.

Per-client setup →