
NVIDIA / Hopper
H100 NVL
| Provider / plan | Price / GPU-hour | Rental type | Terms & configuration | Last checked | Provider pricing link |
|---|---|---|---|---|---|
| Runpodbest-on-demand | $2.59Updated 2026-11-22 · 15:43 UTCLowest flexible rate | On-demandNo reservation | 1 GPU minimumSee providerView detailsCommunity Cloud: lowest public on-demand GPU-hour list rate among enabled clouds. Not a capacity guarantee; storage/networking extra.
|
2026-11-22Auto-collected | View pricing ↗ |
| Vast.aiBest on-demand | $3.3289Updated 2026-11-22 · 15:43 UTC | On-demandNo reservation | 1 GPU minimumJapan, JPView detailsObserved marketplace offer, not a provider-wide list rate. Total hourly quote for one GPU with the API default disk allocation; bandwidth and additional storage can cost extra. One cheapest verified available on-demand offer per GPU is selected; availability can change.
|
2026-11-22Auto-collected | View pricing ↗ |
Prices are dated listings, not live availability. Compare minimum GPU counts and commitments. Storage, networking and taxes may add charges; CPU/RAM inclusion depends on the provider and is shown beside each serverless rate.
H100 NVL combines a hardware profile with dated rental offers on xpu live. H100 NVL is a Hopper PCIe configuration with 94 GB of memory per GPU. A two-GPU NVLink-connected pair has 188 GB in total, but using that capacity requires software that distributes the model. This profile and rental price unit are per GPU.
H100 NVL specifications and configuration
The catalog records Hopper architecture, 94 GB of VRAM, HBM3 memory, PCIe / NVL form factor, 3900 GB/s memory bandwidth, 400 W maximum board power. These values describe the documented hardware edition. They do not describe the CPU, system RAM or storage of every cloud machine. Open the Specifications tab for the full field-by-field comparison.
94 GB per GPU, not 188 GB per card. NVIDIA lists 350–400 W configurable power. Confirm whether the offer supplies one card or a bridged pair. Consult the manufacturer specification source when checking the exact edition.
Choosing H100 NVL for your workload
H100 NVL is worth comparing for Memory-intensive inference; Models distributed across an NVLink-connected pair; Hopper deployments requiring a PCIe form factor. Its documented strengths include 94 GB HBM3 per GPU; 3.9 TB/s memory bandwidth; NVLink bridge support. Match those features to the application you intend to run rather than treating a model name as a performance guarantee.
Before deployment, consider A two-card memory total is not a single automatic memory pool; Per-GPU pricing can differ from the minimum pair cost; NVL peak rates differ from H100 SXM. 70B 8-bit weights alone are about 70 GB; a full runtime needs additional memory. Compare quantized single-GPU serving with supported multi-GPU execution. Full fine-tuning requires substantially more than inference weights.
H100 NVL rental prices and total cost
The rental table separates provider, plan, hourly rate and rental type. Compare on-demand capacity with reserved, spot and serverless terms where those options are listed. The lowest displayed number is not necessarily the lowest total bill for your workload.
Check the last-checked date, minimum GPU count, included CPU and RAM, disk charges and network fees. Confirm availability directly with the provider before paying. Marketplace quotes describe individual machines and can disappear; they are not a permanent provider-wide price.
Compare alternatives before renting
Use the GPU explorer to compare memory and architecture, or browse GPU cloud providers. For H100 NVL, test the same model, precision, batch size and context length on the proposed instance. Record latency, throughput and total cost so your comparison reflects the actual task.
CHOOSING THIS GPU
Is it right for your workload?
94 GB per GPU, not 188 GB per card. NVIDIA lists 350–400 W configurable power. Confirm whether the offer supplies one card or a bridged pair.
Best suited to
- Memory-intensive inference
- Models distributed across an NVLink-connected pair
- Hopper deployments requiring a PCIe form factor
Strengths
- 94 GB HBM3 per GPU
- 3.9 TB/s memory bandwidth
- NVLink bridge support
Things to consider
- A two-card memory total is not a single automatic memory pool
- Per-GPU pricing can differ from the minimum pair cost
- NVL peak rates differ from H100 SXM
HARDWARE DETAILS
Inside the GPU
94 GB per GPU, not 188 GB per card. NVIDIA lists 350–400 W configurable power. Confirm whether the offer supplies one card or a bridged pair.
Image: Reference hardware; rental provider cooling configuration may differ.
- Manufacturer
- NVIDIA
- Architecture
- Hopper
- VRAM (GB)
- 94
- Memory type
- HBM3
- Form factor
- PCIe / NVL
- Memory bandwidth (GB/s)
- 3900
- Maximum board power (W)
- 400
- Introduction year
- 2023
- CUDA cores
- 16896
- Tensor cores
- 4th generation
- Ray tracing cores
- Not present; data-center compute accelerator
- CUDA capability
- 9.0
- Memory ECC
- Supported; driver and allocation settings determine enablement
- Host interface
- PCIe Gen5
- GPU interconnect
- 600 GB/s aggregate bidirectional; bridge required
- Hardware partitioning (MIG)
- Up to 7 hardware instances; provider must enable it
- Cooling
- Passive; server airflow required
Hardware capabilities are not a guarantee of access in a cloud instance. Check the exact edition, GPU allocation and server topology in the rental offer.
MODEL MEMORY
Plan your model size
70B 8-bit weights alone are about 70 GB; a full runtime needs additional memory. Compare quantized single-GPU serving with supported multi-GPU execution. Full fine-tuning requires substantially more than inference weights.
| Example model class | Weight format | Raw weights ≈ | Minimum GPUs for weights only |
|---|---|---|---|
| Llama 3.1 8B ↗ | 4-bit | 4 GB | 1Runtime needs more memory |
| Llama 3.1 8B ↗ | FP16 | 16 GB | 1Runtime needs more memory |
| Qwen2.5 14B ↗ | 4-bit | 7 GB | 1Runtime needs more memory |
| Qwen2.5 32B ↗ | 4-bit | 16 GB | 1Runtime needs more memory |
| Llama 3.1 70B ↗ | 4-bit | 35 GB | 1Runtime needs more memory |
| Llama 3.1 70B ↗ | 8-bit | 70 GB | 1Runtime needs more memory |
| Llama 3.1 70B ↗ | FP16 | 140 GB | 2Runtime needs more memory |
Arithmetic lower bound: rounded parameter count × bits per weight ÷ 8, in decimal GB. Excludes quantization metadata, KV cache, activations, CUDA workspaces and training states. Actual model sizes differ from their rounded names. No context length, batch size or concurrency is guaranteed. GPU counts assume supported model sharding and can be higher in practice. Quantized checkpoints and compatible kernels are required for 4-bit/8-bit execution.
SOFTWARE & PERFORMANCE
Before you deploy
Software compatibility
Use an NVIDIA CUDA-enabled framework/container compatible with this GPU and the host driver. Check PyTorch build and inference-engine kernel requirements before deployment. vLLM documents NVIDIA compute capability 7.5+ as a baseline; support for each model and quantization method still needs checking.
Measured benchmarks
A reproducible, comparable application benchmark has not yet been recorded for this edition. Peak TFLOPS above are not measured tokens per second or image-generation speed.
Compare tests using the same model, precision, GPU count, engine version, input/output length and batch or concurrency. A provider’s server configuration can change the result.
COMPARE YOUR OPTIONS
Alternatives to consider
BUDGET YOUR RUN
GPU cost estimator
GPU compute only: hourly rate × billable hours × GPU count. CPU, RAM, storage, egress, taxes, billing increments and minimum terms may add charges. Serverless hours mean active GPU task time; spot capacity can be interrupted. Reserved commitments and quote-only offers are excluded from this simple hourly estimator.

