Question 1

What AI models can run on H100 NVL?

Accepted Answer

The H100 NVL (94 GB VRAM) can run models including Llama-2-7b, Llama-2-7b-chat, NV-Llama2-13B-RLHF-RM, MediPhi-MedWiki, intent-aware-lfqa-qwen3-4b-baseline and more — 10 compatible models indexed so far. Compatible models are determined by whether the model's memory footprint fits within the GPU's 94 GB VRAM, potentially using quantization (Q8, Q6, or Q4) to reduce memory usage.

Question 2

What is the cheapest cloud option for H100 NVL?

Accepted Answer

The cheapest on-demand option for the H100 NVL is Vast.ai at $1.80/hr (1x-H100_SXM-32-65536). Spot/preemptible pricing is available at $0.00/hr. We track pricing across 83 providers to help you find the best deal.

Question 3

Is H100 NVL good for fine-tuning?

Accepted Answer

Yes, the H100 NVL is excellent for fine-tuning. With 94 GB of VRAM, it can handle full fine-tuning of models up to ~30B parameters and LoRA/QLoRA fine-tuning of much larger models. It's a top-tier choice for training workloads.

Question 4

How much VRAM does H100 NVL have?

Accepted Answer

The H100 NVL has 94 GB of VRAM (HBM3). It also offers 3900 GB/s memory bandwidth, 835.3 TFLOPS of FP16 compute, FP8 support for efficient inference.

Question 5

What quantization formats work on H100 NVL?

Accepted Answer

The H100 NVL supports the following quantization formats: FP16 (full precision), FP8, Q8 (8-bit), Q6 (6-bit), Q4 (4-bit). FP16 works for models up to ~40B parameters. FP8 halves memory vs FP16 with minimal quality loss. Q4 quantization lets models up to ~319B parameters (by VRAM estimate) fit on the H100 NVL.

Provider	On-demand	Spot	Actions
Vast.ai	$1.800/hr	$0.000/hr	Rent
RunPod	$2.890/hr	$2.890/hr	Rent Get $5–$500 free
Lambda	$3.290/hr	—	—
Nebius	$3.850/hr	$2.150/hr	—
Amazon Web Services	$6.880/hr	$2.478/hr	—
Microsoft Azure	$6.980/hr	$6.980/hr	—
Google Cloud	$9.797/hr	$0.971/hr	—

H100 NVL

Specifications

Cloud pricing

Compatible models

Compare this GPU

Frequently asked questions

VRAM	94 GB
Memory type	HBM3
Bandwidth	3900 GB/s
FP16 TFLOPS	835.3
FP8 support	Yes