Get started

The GPU cloud that costs up to 90% less.

Run GPU workloads on SaladCloud's distributed network with published hourly rates, per-second billing, and no minimum commitment.

No contracts. No minimums. Pay only for what you use.

How it works

Deploy in three steps.

Pick from RTX 3060 to RTX 5090 in the portal. Select a container image, set the replica count, choose a priority tier, and configure networking.

Push any Docker image. Your workload is scheduled across available nodes with built-in health checks and automatic reallocation.

Add replicas as demand grows, with no upfront cost. You pay only for the seconds each instance runs.

Resources

Full API reference, tutorials, and how-to guides.

Deploy, manage, and scale containers on the network.

Every GPU class and priority tier, plus live availability.

Image generation, LLMs, molecular dynamics, batch, and more.

Need help?

Ask the team and other developers.

Articles and tickets for SaladCloud customers.

Discuss volume pricing, architecture, reserved capacity, or private deployment options.

Pick a GPU, deploy your container, and run at a fraction of the cost.