Skip to main content
Choose a product to get started.
SUNKSlurm on Kubernetes for batch and burst workloads.CoreWeave InferenceServerless and dedicated inference options for serving AI models.

​
Run agent workloads in Sandboxes

CoreWeave Sandboxes run agent tools and evaluations on serverless capacity or your own CoreWeave Kubernetes Service (CKS) clusters. The shared documentation is under Forge > Sandboxes. Use a Forge API key or CoreWeave API access token for serverless capacity. Use a CoreWeave token for CKS. Start with the serverless quickstart, or use your own compute to configure a CKS runner. The same SDK guides, GPU guide, and computer-use tutorial cover both placement modes.
Last modified on October 20, 2026