Skip to main content

​
Overview

The get_gpu_status action retrieves the status and SSH connection information for all your currently rented GPU instances on the Hyperbolic platform.

​
Function Signature

def get_gpu_status() -> str

​
Input Schema

This action takes no input parameters.

​
Response

Returns the raw JSON response from the Hyperbolic API containing instance details.
​
response
object
JSON object containing an array of your rented instances with details including:
  • Instance ID
  • Status (starting, running, terminated, etc.)
  • SSH connection information
  • GPU configuration
  • Pricing information
  • Runtime details

​
API Endpoint

GET https://api.hyperbolic.xyz/v1/marketplace/instances

​
Example Usage

from hyperbolic_agentkit_core.actions import get_gpu_status

# Get status of all your rented instances
status = get_gpu_status()
print(status)

​
Example Response

{
  "instances": [
    {
      "id": "respectful-rose-pelican",
      "status": "running",
      "ssh_command": "ssh -p 2222 [email protected]",
      "cluster_name": "us-east-1",
      "node_name": "node-abc123",
      "gpu_count": 2,
      "pricing": {
        "price": {
          "amount": 250
        }
      },
      "started_at": "2024-01-15T10:30:00Z"
    }
  ]
}

​
Important Notes

  • Authorization key (HYPERBOLIC_API_KEY) is required for this operation
  • GPU prices in the response are in cents per hour
  • Status values:
    • starting: GPU is not ready yet. Check status again after ~5 seconds
    • running: GPU is ready and accessible
    • terminated: GPU has been terminated
  • Once status is running, you can:
    • Access the instance through the ssh_access action
    • Run commands through the remote_shell action
  • Use the instance ID to terminate instances with the terminate_compute action

​
Status Monitoring

When an instance shows "status": "starting", it means the GPU is still being provisioned. You should:
  1. Wait approximately 5 seconds
  2. Call get_gpu_status again to check if status has changed to running
  3. Repeat until the instance is ready

Build docs developers (and LLMs) love

Get started for freeTalk to us