Skip to main content
SGL Processors are coming soon. This page previews the direction; details may change before launch.
SGL Processors bring serverless functions to confidential compute: deploy a function once, invoke it on demand, and have it run inside a hardware TEE — so your code and its data stay private from the host.

​
The idea

  • Deploy a function to the network (no servers to manage).
  • Invoke it on demand — it cold-starts in an enclave, runs, returns.
  • Pay per call — credits or x402, like the rest of the network.
  • Confidential — execution happens in a TEE; inputs/outputs can be end-to-end encrypted.

​
How it complements the rest of the network

  • Grid = managed model inference.
  • Machines = full instances you control.
  • Processors = lightweight, event-driven functions for glue logic, pre/post-processing, private data transforms, and agent tools — without running a whole machine.

​
Status

Processors are in development. Follow the dashboard (Processors) for availability. In the meantime, use Grid for inference and Machines for general compute.