Instance comparison

p5.48xlarge vs. p4d.24xlarge

This page compares recorded catalogue facts for the exact instances p5.48xlarge and p4d.24xlarge.

$98.32/hr
AWS p5.48xlarge
$32.77/hr
AWS p4d.24xlarge

Side-by-side pricing

p5.48xlargep4d.24xlarge
CloudAWSAWS
GPU8× H100 SXM8× A100 40GB
$/hr$98.32$32.77
$/GPU/hr$12.29$4.10
vCPU / RAM192 / 2048GB96 / 1152GB

Total instance price measures the hourly cost of the entire machine, while per-GPU price normalizes that cost across its GPUs. Spot price is a separate observation for interruptible capacity and should not be treated as equivalent to the standard instance price.

p5.48xlarge contains 8 H100 SXM GPUs, while p4d.24xlarge contains 8 A100 40GB GPUs.

Region, source, and observation date define the scope and timing of the catalogue evidence. They do not establish current capacity or benchmark performance.

Limitations

This comparison covers exact recorded configurations and catalogue prices. It does not establish live capacity or measured performance.

Frequently asked questions

Are total instance and per-GPU prices interchangeable for these two exact instances?

No. Each total price applies to the exact full instance configuration, while the per-GPU cost is calculated separately by distributing that total across the configuration's GPUs.

Do the recorded regions prove that either exact instance is available now?

No. Catalogue regions and observation dates describe recorded evidence, but they do not establish live capacity.

View all GPU prices

Stop comparing. Start running.

Marlin matches your workload to the lowest-priced suitable GPU across supported providers.

Get started with Marlin