Frequently Asked Questions

Vast.ai is a P2P marketplace — cheapest prices, but hosts can terminate your instance anytime. RunPod Community Cloud is more stable but still shared infrastructure. SynpixCloud provides dedicated GPU instances with no preemption, no queue, and consistent performance. You pay slightly more than marketplace pricing ($0.39/hr vs $0.29/hr for RTX 4090), but your jobs won't get interrupted.

Under 5 minutes. Sign up, add funds, pick a GPU, and SSH in. Every instance comes with CUDA 12, PyTorch, and TensorFlow pre-installed. No setup, no waiting.

Only if you use it more than ~400 hours per month (13+ hrs/day every day). A local RTX 4090 system costs $2,500–$3,300 to build, plus electricity and maintenance. At $0.39/hr, cloud is cheaper for most users. Use our cost calculator to check your break-even point.

RTX 4090 (24GB) from $0.39/hr, RTX 3090 (24GB) from $0.29/hr, A100 40GB from $0.63/hr, A100 80GB from $1.89/hr, H100 80GB from $2.49/hr, and more. Check the GPU marketplace for live availability and pricing.

If an instance fails during creation, you are automatically refunded. For long-running workloads, enable disk retention — your entire environment persists, so you can resume on a new GPU in seconds.

Visa, Mastercard, American Express, Google Pay, Apple Pay, and other methods via Stripe. Prepaid billing — add funds to your balance, then use compute. No long-term contracts.