p3 Instance Family

The P3 instance family, launched on October 25, 2017, features NVIDIA Tesla V100 GPUs and custom Intel Xeon E5 processors. Networking bandwidth goes up to 100 Gbps depending on the instance size. P3 instances do not offer instance storage; they are EBS-only. These instances are designed for machine learning, high-performance computing, and other computationally intensive workloads.

Source

vCPUs
Up to 64
Memory
Up to 488 GiB
Architecture
x86_64
Processor
Intel
GPUs
Up to 8
Ranked by CPU performance (Passmark)
#1
m8a, c8a, r8a
3846 Up to 192 vCPU
#3
m7a, c7a, r7a
2887 Up to 192 vCPU
#4
2595 Up to 192 vCPU
#5
2303 Up to 128 vCPU
#6
1853 Up to 192 vCPU
#7
m5dn, c4, r5dn, r5n, t2
1764 Up to 96 vCPU
#8
m4, m5, m5n, m7g, m7gd, c3, c7g, c7gd, c7gn, i3, i3en, r3, r4, r5, r5b, r5d, r7g, r7gd, t3
1648 Up to 96 vCPU
#9
1419 Up to 96 vCPU
#10
1098 Up to 64 vCPU
🇯🇵 ap-northeast-1
🇰🇷 ap-northeast-2
🇸🇬 ap-southeast-1
🇦🇺 ap-southeast-2
🇨🇦 ca-central-1
🇩🇪 eu-central-1
🇮🇪 eu-west-1
🇬🇧 eu-west-2
🇺🇸 us-east-1
🇺🇸 us-east-2
🇺🇸 us-west-2
p3.2xlarge 8 vCPU, 61 GiB

On-Demand Price

Spot Price

p3.8xlarge 32 vCPU, 244 GiB

On-Demand Price

Spot Price

p3.16xlarge 64 vCPU, 488 GiB

On-Demand Price

Spot Price