DallasTexas · US-TX-1
260cards in catalogA dedicated AMD Instinct MI300X with 192 GB HBM3, 28 vCPU, 256 GB system RAM, and 3.84 TB local NVMe per card.
Node pricing
The product model allocates each GPU to one account for the complete term. Adding cards to a node applies the catalog volume discount to every card.
| Node | Total VRAM | 7 days | 30 days | Discount |
|---|---|---|---|---|
| 1× MI300X | 192 GB | $450 | $1,590 | 0% |
| 2× MI300X | 384 GB | $855 | $3,021 | 5% |
| 4× MI300X | 768 GB | $1,656 | $5,851 | 8% |
| 8× MI300X | 1536 GB | $3,240 | $11,448 | 10% |
Regional inventory
Ready-to-run images
Workload fit
The largest single-card VRAM allocation in this catalog, worth evaluating when fitting a large model on one accelerator simplifies serving or training.
Your catalog configuration pairs 192 GB GPU memory with 256 GB host RAM and 3.84 TB local NVMe per card.
Use a ROCm-qualified environment. Custom CUDA extensions and NVIDIA-specific kernels require an explicit compatibility check or replacement.
Use $450 for a 7-day pilot, or $1,590 for a 30-day allocation after validating the workload. Options are additional.