How much does H100 PCIe cost per hour?

Runcrate H100 PCIe rental starts at $1.25/hr on-demand (avg $1.40/hr). RunPod lists $1.79, Lambda $2.29, AWS $3.78. Runcrate is the lowest published H100 PCIe cloud rate.

Should I rent H100 PCIe or H100 SXM?

Pick PCIe if you only need 1-4 GPUs and your workload doesn't saturate memory bandwidth (most inference under 70B). Pick SXM if you need 8-GPU clusters with NVLink for distributed training. SXM has 67% more memory bandwidth.

Is the H100 PCIe powerful enough for fine-tuning?

Yes for QLoRA on models up to 70B and full fine-tuning on models up to 7B. For full FT of larger models or for distributed training, you'll want H100 SXM with NVLink.

runcrate

Contact Sales Console

NVIDIA · Hopper · 2022

NVIDIA H100 PCIe.

Name: NVIDIA H100 PCIe GPU Cloud Instance
Brand: NVIDIA
Price: 1.40 USD
Availability: InStock

Hopper performance in a PCIe form factor — drops into standard servers, no SXM rack required.

80 GB

HBM3

2.0 TB/s

Bandwidth

102

TFLOPS FP16

4× GPU

PCIe cluster

Available now · 8 regions

Deploy in 60 seconds Reserved quote

Cloud rental

$1.40/hr

On-demand · per-second billing

$1.25–$1.55/hr range

Or $0.98/hr reserved (30% off)

Pricing across clouds

NVIDIA H100 PCIe cloud rental price comparison.

Same GPU, cheapest-first. Prices reflect publicly listed hourly rates for NVIDIA H100 PCIe on each provider. Runcrate is the lowest published rate.

RuncrateCheapest

$1.40/hr

RunPod

$1.79/hr

Oracle

$2.20/hr

Lambda

$2.29/hr

GCP

$3.50/hr

Azure

$3.50/hr

AWS

$3.78/hr

Workloads

What you'll actually use this for.

Real workloads sized for the NVIDIA H100 PCIe, with concrete performance numbers. Click to deploy preconfigured.

Llama 70B Inference

~310 tok/s in bf16

Deploy

Fine-tune Llama 7B

QLoRA in 4 hrs on 1 GPU

Deploy

Whisper Transcription

Real-time on Whisper Large V3

Deploy

Stable Diffusion XL

1024x1024 in 1.5s

Deploy

One command. NVIDIA H100 PCIe in 60 seconds.

Skip the dashboard if you don't need it. SDK, Python, or cURL — copy the snippet, paste your API key, ship.

import Runcrate from "@runcrate/sdk";

const rc = new Runcrate({ apiKey: "rc_live_••••••••••••••••" });

const instance = await rc.instances.create({
  gpu: "h100-pcie",
  region: "auto",
  image: "runcrate/vllm:latest",
});

console.log(`SSH: ssh root@${instance.host}`);