Volumes - DeployGPU

Persistent volumes let you store data that survives across deployments, on Verda and Hyperstack. Attach a volume when you create a deployment, then format it the first time and mount it yourself on each deployment: see Volumes. Deployment creation validates every volume before reserving credit or calling a provider. All volumes must be owned by the caller, available, on the selected provider, and in one location supported by the offer. Invalid selections fail immediately. Volumes are never silently skipped. All volume endpoints require authentication.

Create Volume

POST /api/v1/volumes

Creates a persistent storage volume at a provider. The volume is billed hourly while it exists.

Request Body

{
  "provider": "verda",
  "name": "my-training-data",
  "size_gb": 100,
  "volume_type": "HDD",
  "region": "FIN-01",
  "description": "Training dataset storage"
}

Field	Type	Required	Description
`provider`	string	Yes	Provider name
`name`	string	Yes	Volume name
`size_gb`	int	Yes	Size in GB (must be positive)
`volume_type`	string	No	From the volume types endpoint
`region`	string	Yes, where the provider has locations	Location code, for example `FIN-01`. Must be one of the locations reported by the provider info endpoint. A volume can only ever be attached to deployments in this location, and deployments that attach it are pinned there. Choose the location where you intend to run.
`description`	string	No	Optional description

Response `201 Created`

{
  "id": "e6a12cc1-5d22-48c6-bda3-01fce11c9d19",
  "provider": "verda",
  "name": "my-training-data",
  "size_gb": 100,
  "volume_type": "HDD",
  "status": "creating",
  "region": "FIN-01",
  "description": "Training dataset storage",
  "attached_to": null,
  "hourly_rate": "0.0076",
  "created_at": "2026-05-10T19:54:15.906789Z",
  "updated_at": "2026-05-10T19:54:15.906789Z"
}

Errors

Code	When
`insufficient_balance`	Not enough credits to cover the volume’s hourly rate
`validation_error`	Invalid parameters: missing name, size outside the provider range, unknown volume type, or a missing or unknown `region`. The message lists the valid locations, for example `A valid region is required for verda volumes — one of: FIN-01, FIN-02, FIN-03`
`provider_maintenance`	The provider is in maintenance and volumes cannot be created there. Returned as `503 Service Unavailable` with `Retry-After`. Nothing is created or charged. See provider maintenance

List Volumes

GET /api/v1/volumes

Returns all your volumes.

Response

{
  "data": [
    {
      "id": "e6a12cc1-5d22-48c6-bda3-01fce11c9d19",
      "provider": "verda",
      "name": "iron-panda-434",
      "size_gb": 50,
      "volume_type": "HDD",
      "status": "available",
      "region": "FIN-01",
      "description": null,
      "attached_to": null,
      "hourly_rate": "0.0038",
      "created_at": "2026-05-10T19:54:15.906789Z",
      "updated_at": "2026-05-10T19:54:17.547158Z"
    }
  ]
}

Volume Status

Status	Description
`creating`	Being provisioned at the provider
`available`	Ready to attach to a deployment
`attaching`	Being attached to an instance
`in_use`	Attached to a running deployment
`detaching`	Being detached
`deleting`	Delete accepted. No longer billed, and removed once the provider confirms
`deleted`	Fully removed
`error`	The provider did not confirm the delete after repeated attempts. Delete it again to retry

Get Volume

GET /api/v1/volumes/{id}

Returns a single volume. Same shape as items in the list response.

Errors

404 not_found: the volume does not exist or is not owned by you.

Delete Volume

DELETE /api/v1/volumes/{id}

Deletes a volume. The volume must not be attached to a running deployment.

Response

204 No Content The 204 means the delete was accepted. The volume is not billed from that moment. It shows deleting until the provider confirms, then deleted. That is usually immediate. If the provider is in maintenance or does not answer, the delete is retried for you every few minutes, and the volume stays deleting in the meantime. You do not need to send the request again, although doing so is harmless.

Errors

Code	When
`not_found`	Volume not found
`validation_error`	Volume is currently attached to a deployment, or is still being created, attached or detached. In the second case, try again once that operation has finished

​Create Volume

​Request Body

​Response 201 Created

​Errors

​List Volumes

​Response

​Volume Status

​Get Volume

​Errors

​Delete Volume

​Response

​Errors

Create Volume

Request Body

Response `201 Created`

Errors

List Volumes

Response

Volume Status

Get Volume

Errors

Delete Volume

Response

Errors