Pay for what you use: a small base fee for every hour a machine is running — the isolated slot itself, not reserved vCPU/RAM — plus what your workload actually consumes on top. Sit idle and the consumption lines fall to near zero; stop the machine and even the base fee stops. The first $10.00/month is free.
| Resource | Unit | Price |
|---|---|---|
| Base (running machine) | per machine-hour running | $0.04 |
| Compute (active vCPU) | per vCPU-hour consumed | $0.05 |
| Memory (active RSS) | per GB-hour in use | $0.0162 |
| Disk (used) | per GB-hour used (~$0.073/GB-month) | $0.0001 |
| Network egress | per GB out | $0.05 |
beta 10% off the spend above $2,000.00 a month. The threshold and rate may still change.
Every plan bills the rates above — a plan buys limits and support, never a different price per unit — and machines start from warm pools on all of them, free included. Machines go up to 8 vCPU and 16 GB unless you need more.
Start free, or email us to move to startup or enterprise. Self-serve upgrade is coming shortly.
One machine, 1 vCPU + 512 MB, ~2 GB of disk, busy 24/7 for a month (730h):
A stopped machine pays no base fee and no CPU or memory — those meters stop the moment it stops. Its disk still occupies space on the host, so that held storage keeps billing at the disk rate (a few cents per GB-month). Delete the machine to stop that too.
The control plane samples each machine's live CPU, memory, and disk continuously and accumulates a durable per-account total. View it in real-time at /console/usage.
Every machine runs in its own hardware-isolated microVM. Not a container — a real virtual machine with its own kernel. Your code can't see other tenants.
Yes. The dashboard uses the same REST API you can call from your code. Generate an API key at /console/keys.