A little look ahead.
A lot more possibility.
More ways to put shared compute to work. We’re looking beyond the single workstation, toward jobs that run themselves, machines that work together, and payments that follow your workload.
Explore what’s comingOF COMMUNITY COMPUTE
API access
Your next workstation,
a script away.
Bring CPUBorrow into the tools you already use. Planned API access will let you automate rentals and submit jobs from a script, so a new build or a fresh dataset can kick off remote work without a trip through the dashboard.
Think repeatable compile jobs, scheduled data processing, and pipelines that borrow extra capacity when it’s needed. The aim is to make remote compute a natural step in your workflow, from submitting the job to bringing the results back.
Swarm mode
Fifty machines.
One bigger idea.
Some jobs need more than a faster workstation. Swarm mode is planned to let you rent 50 machines at once as one virtual cluster, bringing independent capacity together for a single distributed workload.
The vision is instant, ad-hoc compute for parallel training tasks, rendering, and other work that can be split across machines. Assemble a swarm for the job in front of you, then release it when you’re done—without setting up an account with a cloud provider or maintaining a permanent cluster.
Pay-as-you-compute
streaming
A running meter.
A fairer fit.
Workloads rarely run at full speed for an entire hour. Planned streaming payments will meter compute by the second, with charges based on live CPU or GPU utilization rather than a fixed hourly rental block.
As a job gets busy, usage rises. When it slows down, the meter follows. The goal is a closer connection between the resources your work actually consumes and what you pay, especially for short bursts, uneven processing, and experiments with unpredictable runtimes.
More on the horizon.
Room to compute today.
These features are planned, with timing and details still to come.
In the meantime, find a workstation for your next big job.