Regions

Every Daytona sandbox runs in a region: a geographic or logical grouping of compute infrastructure. When creating a sandbox, you can target a specific region, and Daytona schedules the workload on available capacity within that region.

Shared regions

Regions managed by Daytona and available to all organizations:

from daytona import Daytona, DaytonaConfig

# Configure Daytona to use the US region
config = DaytonaConfig(
    target="us"
)

# Initialize the Daytona client with the specified configuration
daytona = Daytona(config)

# Create a sandbox in the US region
sandbox = daytona.create()

List regions managed by Daytona and available to all organizations:

curl 'https://app.daytona.io/api/shared-regions' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Earth region

GPU sandboxes and GPU snapshots on shared regions belong to the Earth region: a global region that spans the shared GPU fleet. Daytona selects the shared region a GPU workload runs in and ignores the region preference.

Earth is not a physical region. It is not returned by the regions or shared regions endpoints, and it is a reserved region ID that cannot be assigned to a custom region. GPU sandboxes and snapshots on dedicated and custom regions keep their real region ID and remain targetable.

The earth region ID appears in:

• target of a GPU sandbox on a shared region
• regionIds of a GPU snapshot on a shared region
• regionUsage entries of the usage overview: one earth entry per sandbox class aggregates the shared GPU quota and usage; shared region entries report no GPU quota
• Available sandbox classes: one earth entry per sandbox class with gpuAvailable and allowedGpuTypes; shared region entries report gpuAvailable: false
• Region quota of a GPU sandbox: GPU quota fields aggregated under earth
• region.id attribute of the GPU organization metrics
• Error messages that reference the region of a GPU workload

Dedicated regions

Dedicated regions are managed by Daytona and provisioned exclusively for an organization. The infrastructure is not shared with other organizations, and Daytona operates it as a managed service.

Custom regions

Custom regions run on compute that your organization provides and manages. Attach your own machines through bring your own compute (BYOC) to control data locality, compliance, and infrastructure configuration, and scale capacity independently within each region.

Custom regions have no limits on concurrent resource usage: capacity is bounded only by the compute you attach.