Hardware
GPU hardware.
What every node includes, and GPU specifications and availability side by side.
Every node is NVIDIA SXM hardware in one of our regions, sold as interruptible capacity at the published GPU-hour rate. Reserved capacity is unavailable for new requests. Request capacity when you are ready, or read the uptime SLA first.
- 1×, 2×, 4×, 8× SXM GPUs per node, connected over NVLink.
- Interruptible (preemptible) capacity for all new requests, priced per GPU-hour.
- United States and European Union regions.
- 24/7 support: 1 hour response when reserved capacity is down, backed by SLA credits.
- No platform fee. The GPU-hour rate is the bill.
The GPUs.
NVIDIA’s published figures per GPU; our availability and per-GPU-hour prices.
| Spec | H100Available | H200Available | B300Available | B200Sold out |
|---|---|---|---|---|
| Model | H100 SXM 80GB | H200 SXM 141GB | B300 SXM 288GB | B200 SXM 180GB |
| Architecture | Hopper | Hopper | Blackwell Ultra | Blackwell |
| GPU memory | 80 GB HBM3 | 141 GB HBM3e | 288 GB HBM3e | 180 GB HBM3e |
| Memory bandwidth | 3.35 TB/s | 4.8 TB/s | 8 TB/s | 8 TB/s |
| NVLink, per GPU | 900 GB/s | 900 GB/s | 1.8 TB/s | 1.8 TB/s |
| Form factor | SXM5 | SXM5 | SXM | SXM |
| Launch sizes | 1×, 2×, 4×, 8× | 1×, 2×, 4×, 8× | 1×, 2×, 4×, 8× | 1×, 2×, 4×, 8× |
| Interruptible | $1.38 / GPU-hour | $1.67 / GPU-hour | $3.49 / GPU-hour | — |
| Reserved | Unavailable | Unavailable | Unavailable | — |
Figures from NVIDIA: NVIDIA H100 Tensor Core GPU · NVIDIA H200 Tensor Core GPU · NVIDIA HGX B300 · NVIDIA HGX B200. Availability and prices are ours.
Common questions.
Everything you need to know about running on Compute Cheap. Still have questions? Email us and we’re quick to respond.
Which GPUs are available right now?
H100 SXM 80GB, H200 SXM 141GB, and B300 SXM 288GB are available for interruptible capacity requests. B300 is $3.49 per GPU-hour. Reserved capacity is currently unavailable for new requests. B200 is currently sold out: leave your email on the request page and we'll notify you when capacity opens.
How many GPUs can I run per node?
1, 2, 4, or 8 SXM GPUs per node, connected over NVLink. Need more than eight? Say so in the request notes and we'll confirm what's available before anything is reserved.
Is reserved capacity available?
New compute requests are currently interruptible only. Capacity can be reclaimed if demand spikes, so it suits fault-tolerant training and batch jobs. Reserved capacity is unavailable for new requests; existing agreed terms stay in place.
Which regions is capacity available in?
United States and European Union today. Mention your preferred region in your request's notes, and we'll confirm availability there before anything is reserved.
Is there a minimum reservation?
Yes, 2,000 GPU-hours per request. That's roughly 8 GPUs for 10 days, or fewer GPUs over a longer window, split however your workload needs.
What support and uptime come with it?
Support runs 24/7: a 1-hour response when reserved capacity is down and 4 hours for degraded performance. Uptime below 99.9% earns service credits of 10% to 50% of the affected block; the Uptime SLA has the full schedule.
How is the price calculated?
Hourly rate × GPU-hours, shown before you request. There's no separate platform fee layered on top of the GPU-hour price.