01
FOR DEVELOPERS
Build on the network
One API, served by a network instead of a data center. Run open models across distributed compute with verifiable execution underneath every response.
A permissionless network of user-owned GPUs powering inference and training for open models.
One API, served by a network instead of a data center. Run open models across distributed compute with verifiable execution underneath every response.
Put idle hardware to work on real inference and training jobs. Earn for the compute you contribute, keep control of your machine, and leave whenever you want.
Open models need open infrastructure. Crunch connects the people building intelligence with the people supplying the machines that run it.

Crunch turns scattered machines into a shared intelligence layer—without asking users to surrender their data or contributors to surrender control.
Start building ↗