Spin up a GPU instance

Instance sizes

• h200-1x — 1 GPU, 80GB VRAM, 128GB RAM ($2.40/hour)
• h200-2x — 2 GPUs, 160GB VRAM, 256GB RAM ($4.60/hour)
• h200-4x — 4 GPUs, 320GB VRAM, 512GB RAM ($8.90/hour)
• h200-8x — 8 GPUs, 640GB VRAM, 1TB RAM, NVLink Fabric ($18.50/hour)

Launch via API

bash
`curl -X POST https://api.xobriq.com/v1/cloud/instances \
  -H "Authorization: Bearer $XOBRIQ_KEY" \
  -d '{
    "size": "h200-4x",
    "image": "pytorch-2.4-cuda-12",
    "region": "nairobi-1",
    "name": "llama-finetune"
  }'`

SSH access

Add your SSH public key from Console → Compute → SSH Keys. New instances embed the key on first boot for immediate access.

Billing model

Per-second billing after the first minute. Stopped instances only charge storage. Idle GPUs auto-hibernate after 15 minutes (configurable).