Pay only for what you use. No subscriptions, no minimums, no hidden fees. Just GPU compute when you need it.
Perfect for development, testing, and smaller workloads
For production workloads and growing teams
Maximum capability for demanding AI workloads
Cost Comparison
MetalCloud offers unique value for memory-intensive AI workloads
| Feature | MetalCloud M3 Ultra | AWS p5 (H100) | Lambda Labs (H100) |
|---|---|---|---|
| GPU Memory | 512GB unified | 80GB per GPU | 80GB per GPU |
| Memory for 512GB capacity | 1 machine | 7+ H100s required | 7+ H100s required |
| Hourly cost (512GB equiv.) | £3.50/hr | ~$25+/hr (multi-GPU) | ~$18+/hr (multi-GPU) |
| Run Llama 70B FP16 | ✓ Single machine | Requires 3+ GPUs | Requires 3+ GPUs |
| Minimum commitment | None (per-second) | On-demand or reserved | Hourly minimum |
| Memory bandwidth | 800GB/s | 3.35TB/s | 3.35TB/s |
FAQ
We bill per second of actual GPU compute time. Your job starts, we start the meter. Your job ends, we stop it. No rounding up to the nearest hour, no minimum charges. You'll receive a detailed invoice at the end of each billing period showing exactly what you used.
No hidden fees. The price you see is the price you pay. Data transfer, storage for job artifacts, and API access are all included. We believe in transparent pricing—you should never be surprised by your bill.
Yes. For teams with consistent high-volume usage (1000+ GPU hours/month), we offer custom pricing with significant discounts. Contact us at enterprise@metalcloud.space to discuss your needs.
We accept all major credit cards (Visa, Mastercard, Amex) and can arrange invoicing for enterprise customers. All payments are processed securely through Stripe.
New users receive £10 in free credits upon signup—enough for several hours of M3 Pro compute or a few hours of M3 Ultra time. This lets you test our platform risk-free before committing to paid usage.
Join the waitlist for early access and £10 in free credits.
Join Waitlist