NVIDIA H200 Available

GPU Cloud for AI Builders

Rent NVIDIA H200 GPUs on-demand. Per-minute billing from $0.013/min. Jupyter, SSH, or API access.

Three Steps to GPU Compute

Sign Up

Quick email sign-up. After a brief approval, get $5 free credit to start.

Pick Your GPU

Choose from 10GB to 141GB VRAM slices. MIG-partitioned GPU.

Launch & Build

Jupyter, SSH, or API endpoint. Running in under 3 minutes.

Simple, Transparent Pricing

Pay only for what you use. No commitments, no hidden fees. Start with $5 free credit.

Starter

$0.80/hour

10 GB VRAM

Fine-tuning with LoRA, small inference

Standard

$1.50/hour

20 GB VRAM

Medium model training, batch inference

Pro

$2.80/hour

40 GB VRAM

Large model fine-tuning, multi-task

Full GPU

$5.20/hour

141 GB VRAM

Full GPU -- 70B models, max throughput

Programmatic Access

Launch and manage GPU instances from your CI/CD pipeline, Python scripts, or any HTTP client. Full REST API with API key authentication.

launch_gpu.py

import requests

# Launch a GPU instance on the Standard tier (20 GB VRAM).
# gpu_tier_id is the numeric ID from GET /v1/gpus/tiers
resp = requests.post(
    "https://api.ronai.cloud/v1/instances",
    headers={"Authorization": "Bearer rcloud_sk_..."},
    json={
        "gpu_tier_id": 2,
        "instance_type": "jupyter",
    },
)

instance = resp.json()["data"]
print(f"Jupyter ready at: {instance['access_url']}")