Guides

Managing GPU Instances

Learn how to create, monitor, and control your GPU rentals on GPUniq.

Creating Rentals

Rent GPU instances through the GPUniq marketplace or API based on your compute needs.

Browse Marketplace

Navigate to gpuniq.com and browse available GPUs.

Select GPU

Click on a GPU card to view full specifications:

GPU model and VRAM
Total RAM and CPU cores
Storage capacity
Location and provider reliability

Choose Pricing

Select billing type: per minute, hour, day, week, or month.

Rent

Click "Rent Now" - your instance starts within seconds.

API-Based Rental

Create rentals programmatically for automation workflows.

curl -X POST https://api.gpuniq.com/v1/marketplace/order \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": 123,
    "gpu_required": 1,
    "pricing_type": "hour"
  }'

const response = await fetch('https://api.gpuniq.com/v1/marketplace/order', {
  method: 'POST',
  headers: { 
    'Authorization': `Bearer ${YOUR_API_KEY}`, 
    'Content-Type': 'application/json' 
  },
  body: JSON.stringify({
    agent_id: 123,
    gpu_required: 1,
    pricing_type: 'hour'
  })
});
const result = await response.json();
console.log(result.data.task_id);

import requests
response = requests.post(
  'https://api.gpuniq.com/v1/marketplace/order',
  headers={'Authorization': 'Bearer YOUR_API_KEY', 'Content-Type': 'application/json'},
  json={
    'agent_id': 123,
    'gpu_required': 1,
    'pricing_type': 'hour'
  }
)
result = response.json()
print(result['data']['task_id'])

Async Order Creation

For better reliability, use the async order endpoint which returns a job ID for polling:

# Create order asynchronously
curl -X POST https://api.gpuniq.com/v1/marketplace/order/async \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"agent_id": 123}'

# Poll for status
curl https://api.gpuniq.com/v1/marketplace/order/status/{job_id} \
  -H "Authorization: Bearer YOUR_API_KEY"

Use async orders for production systems to handle network timeouts gracefully.

Viewing Your Instances

Access your active rentals through the dashboard or API.

Go to My Instances in your dashboard to see:

Active rentals with SSH credentials
Running time and cost
GPU utilization metrics
Quick actions (stop, delete)

List your instances programmatically:

curl https://api.gpuniq.com/v1/instances/my \
  -H "Authorization: Bearer YOUR_API_KEY"

Response:

{
  "data": {
    "items": [
      {
        "task_id": 456,
        "agent_id": 123,
        "status": "running",
        "ssh_host": "203.0.113.50",
        "ssh_port": 22001,
        "ssh_password": "generated_password",
        "started_at": "2026-01-09T10:30:00Z",
        "hourly_cost": 0.45
      }
    ],
    "total": 1
  }
}

Instance Lifecycle

Control your instances efficiently to manage costs.

Start

Resume a stopped instance.

Stop

Pause billing while preserving state.

Delete

Permanently terminate the rental.

Stop Instance

Pause your rental to stop billing. The instance state may or may not be preserved depending on the provider.

curl -X POST https://api.gpuniq.com/v1/instances/{task_id}/stop \
  -H "Authorization: Bearer YOUR_API_KEY"

Start Instance

Resume a stopped instance:

curl -X POST https://api.gpuniq.com/v1/instances/{task_id}/start \
  -H "Authorization: Bearer YOUR_API_KEY"

Delete Instance

Permanently terminate a rental:

curl -X DELETE https://api.gpuniq.com/v1/instances/{task_id} \
  -H "Authorization: Bearer YOUR_API_KEY"

Deleting an instance is irreversible. All data on the instance will be lost.

Connecting to Your Instance

After renting, you receive SSH access credentials.

SSH Connection

ssh root@{ssh_host} -p {ssh_port}
# Enter the password from your dashboard

Using SSH Keys

For passwordless access, add your public key after first login:

# On your local machine
cat ~/.ssh/id_rsa.pub | pbcopy

# On the rented GPU
mkdir -p ~/.ssh
echo "YOUR_PUBLIC_KEY" >> ~/.ssh/authorized_keys
chmod 600 ~/.ssh/authorized_keys

Instance Details

Get detailed information about a specific rental:

curl https://api.gpuniq.com/v1/instances/{task_id} \
  -H "Authorization: Bearer YOUR_API_KEY"

Response includes:

SSH connection details
Container information
Start time and duration
Cost breakdown
GPU utilization (if available)

Task History

View your rental history and statistics:

curl https://api.gpuniq.com/v1/task-history/completed?days=30 \
  -H "Authorization: Bearer YOUR_API_KEY"

Best Practices

Use per-minute billing for short experiments and testing
Commit to longer periods for production workloads (20-40% savings)
Stop instances when not in use instead of keeping them running
Monitor utilization - if GPU usage is low, consider a cheaper option
Use verified providers for critical workloads to ensure reliability

Your balance is deducted in real-time. Ensure sufficient funds before starting long-running jobs.

Was this page helpful?

Built with Documentation.AI

Last updated Jan 9, 2026