AWS P3 instance types

P3 is a gpu & accelerated Amazon EC2 family with 3 instance types, from 8 – 64 vCPUs and 61 GiB – 488 GiB of memory with NVIDIA V100 accelerators, running on Intel Xeon E5-2686 v4 (Broadwell). Linux on demand starts at $3.06 per hour.

gpu & acceleratedx86-64

sizes: 3
vCPUs: 8 – 64
memory: 61 GiB – 488 GiB
accelerator: V100
on-demand from: $3.06

Compare sizes

P3 sizes and prices

Linux, hourly, US East (N. Virginia) (other regions on each size's page)

Size	vCPU	Memory	Accelerator	Local storage	On-demand	Monthly	Spot
p3.2xlarge	8	61 GiB	1 × NVIDIA V100	–	$3.06	$2,234	$1.381
p3.8xlarge	32	244 GiB	4 × NVIDIA V100	–	$12.24	$8,935	$4.955
p3.16xlarge	64	488 GiB	8 × NVIDIA V100	–	$24.48	$17,870	$8.825

Other gpu & accelerated families on AWS

Dl1 1 Dl2q 1 F1 3 F2 3 G3 3 G3s 1 G4ad 5 G4dn 7 G5 8 G5g 6 G6 8 G6e 8 G6f 4 G7 6 G7e 6 Gr6 2 Gr6f 1 Inf1 4 Inf2 4 P2 3 P3dn 1 P4d 1 P4de 1 P5 2 P5e 1 P5en 1 P6-b200 1 P6-b300 1 Trn1 2 Trn1n 1

P3 FAQ

What sizes does the AWS P3 family have?

P3 has 3 instance types, from p3.2xlarge (8 vCPU, 61 GiB) to p3.16xlarge (64 vCPU, 488 GiB). They run on Intel Xeon E5-2686 v4 (Broadwell).

What is the cheapest P3 instance type?

p3.2xlarge at $3.06 per hour (Linux, on demand) in US East (N. Virginia), about $2,234 per month.

What is P3 best for?

P3 is a gpu & accelerated family. GPUs and AI accelerators for training, inference, rendering and video.