Provider profile
Runpod offers GPU infrastructure for model development and inference through Pods, Serverless endpoints, and multi-node Clusters. Its public GPU and price tables change with product type and current capacity, so a rate-card entry is not a guarantee that a particular GPU is available to every account or region.
Published GPU offers
| GPU | VRAM | System RAM | vCPU | Rate shown | Regions | Provider rate card |
|---|---|---|---|---|---|---|
| NVIDIA H100 PCIe | 80 GB | 188 GB | 16 | $2.89/hour | Check live inventory | Check current offer |
| NVIDIA L4 | 24 GB | 50 GB | 12 | $0.49/hour | Check live inventory | Check current offer |
| NVIDIA RTX A5000 | 24 GB | 25 GB | 9 | $0.27/hour | Check live inventory | Check current offer |
Rates are examples from the provider’s published card; stock, selected cloud tier, region, storage and final checkout price may differ. GPU count is not stated in this listing.
Research notes
The catalog separates Pods for persistent GPU instances, Serverless for usage-based inference endpoints, and Clusters for multi-GPU jobs. Published GPU cards show model, VRAM, system RAM, vCPU count, and product-specific rates where provided. The price page distinguishes Community Cloud and Secure Cloud; verify the exact offering and region in the live catalog before deployment.
Runpod documents per-second billing for compute and storage, with product-specific billing rules. The billing guide states there are no data-transfer fees. Storage charges vary by type; a stopped volume disk can continue to incur storage charges. Published rates are dynamic and should be checked against the selected GPU and cloud tier at checkout.
The provider's price page describes GPU capacity across 30+ regions. A GPU card and cloud tier do not establish availability in every region; confirm live inventory and account eligibility in the deployment selector.
Runpod publishes product documentation and account support channels. Support scope depends on the product and account terms; no response-time guarantee is stated here.
Community and Secure Cloud offerings have different host and service characteristics. GPU stock and prices change, storage is billed separately, and a displayed rate does not include a model-specific throughput or total inference cost. No fixed plan rows are recorded from this dynamic rate card.
No verified records yet
Review the sourced product details and any GPU offers above. Check the provider’s site for current configurations and the final price.
Explore products and pricingCounts describe the records published on HostCritiq, not every offer Runpod sells. Prices can differ by billing term, region, currency and tax. Open a plan for its full terms.
