dstack - Provision and orchestrate GPU workloads
Provisions and orchestrates GPU workloads across GPU clouds, Kubernetes clusters, and on-premises clusters.
Tags
Updated: 2026-10-06Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Provision GPU workloads
- Orchestrate compute resources
- Create YAML configurations
- Apply workload configurations
- Manage compute fleets
- Monitor workload runs
- Inspect CLI help
Inputs
- dstack YAML configuration files
- GPU resource requirements
- Backend and fleet configuration
- dstack server settings
- Project selection
- Environment variables and secrets
- User confirmation
Outputs
- Provisioned workloads
- CLI status and log output
- SSH tunnels
- Forwarded ports
- Service endpoints
- Fleet state changes
Requirements
- dstack server
- dstack CLI
- Configured compute backend
- Matching configured fleet
- Access to GPU cloud, Kubernetes, or SSH infrastructure