One job model
Submit work through the console or API, then collect status, logs, and artifacts.
Distributed compute, two ways
Put idle computers to work.
Connect computers you trust, or access complete machines from the open network.
Private clustersMarketplace
01 / Cluster
Start with computers you trust.
Turn a classroom, lab, studio, office, or collection of personal computers into a private pool. PlanetNodes sends separate tasks to the right machines and gathers their results.
Your laptop
Lab workstation
Office server
It does not combine their RAM into one larger computer. The workload must be divisible into independent jobs.
See private clusters02 / Marketplace
Add outside capacity when you need it.
Select a provider computer by hardware, region, reliability, and price. The complete job runs there; your local machine sends inputs and receives outputs.
Your computer16 GB
Rented provider32 GB
Your laptop remains a 16 GB machine. The task runs on the rented 32 GB computer; this is remote execution, not borrowed RAM.
Browse live capacityOne platform
Private first. Marketplace when useful.
01
Private pool
Use machines whose owners and environment you know.
02
Scheduler
Match each task to a capable, available machine.
03
Optional overflow
Use public providers only when policy and workload allow it.
The same control plane for both products.
Hardware-aware scheduling
Route each independent task to a machine with the CPU, GPU, memory, and availability it needs.
Visible controls
Set time, cost, concurrency, and machine boundaries before a workload starts.
Desktop agent
Connect macOS, Windows, and Linux computers without turning PlanetNodes into a remote desktop.