I've compiled a list of GPU instances for various providers:
Table of Contents
Please note that these cost estimates are based on the on-demand pricing for the region closest to N. Virginia (USA) or Frankfurt (Germany). Prices are in either USD or EUR, are subject to change, and may vary by region and other factors not listed here.
Replicate
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| Nvidia T4 | 1x T4 | 16 GB | 4 | 8 GB | $0.81 |
| Nvidia A40 | 1x A40 | 48 GB | 4 | 16 GB | $2.07 |
| Nvidia A40 (Large) | 1x A40 | 48 GB | 10 | 72 GB | $2.61 |
| Nvidia A100 (40 GB) | 1x A100 | 40 GB | 10 | 72 GB | $4.14 |
| Nvidia A100 (80 GB) | 1x A100 | 80 GB | 10 | 144 GB | $5.04 |
| 8x Nvidia A40 (Large) | 8x A40 | 48 GB | 48 | 680 GB | $20.88 |
You can find more on Replicate GPU instances here.
Amazon Web Services
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| g5g.xlarge | 1x T4G | 16 GiB | 4 | 16 GiB | $0.42 |
| g4dn.xlarge | 1x T4G | 16 GiB | 4 | 16 GiB | $0.526 |
| g5g.2xlarge | 1x T4G | 16 GiB | 8 | 16 GiB | $0.556 |
| g3s.xlarge | 1x M60 | 8 GiB | 4 | 30.5 GiB | $0.75 |
| g4dn.2xlarge | 1x T4G | 16 GiB | 8 | 32 GiB | $0.752 |
| g5g.4xlarge | 1x T4G | 16 GiB | 16 | 16 GiB | $0.828 |
| p2.xlarge | 1x K80 | 12 GiB | 4 | 61 GiB | $0.90 |
| g5.xlarge | 1x A10G | 24 GiB | 4 | 16 GiB | $1.006 |
| g4dn.4xlarge | 1x T4G | 16 GiB | 16 | 64 GiB | $1.204 |
| g3.4xlarge | 1x M60 | 8 GiB | 16 | 122 GiB | $1.14 |
| g5.2xlarge | 1x A10G | 24 GiB | 8 | 32 GiB | $1.212 |
| g5g.8xlarge | 1x T4G | 16 GiB | 32 | 16 GiB | $1.372 |
| g5.4xlarge | 1x A10G | 24 GiB | 16 | 64 GiB | $1.624 |
| g4dn.8xlarge | 1x T4G | 16 GiB | 32 | 128 GiB | $2.176 |
| g3.8xlarge | 2x M60 | 16 GiB | 32 | 244 GiB | $2.28 |
| g5.8xlarge | 1x A10G | 24 GiB | 32 | 128 GiB | $2.448 |
| g5g.16xlarge | 2x T4G | 32 GiB | 64 | 32 GiB | $2.744 |
| p3.2xlarge | 1x V100 | 16 GiB | 8 | 61 GiB | $3.06 |
| g4dn.12xlarge | 4x T4G | 64 GiB | 48 | 192 GiB | $3.912 |
| g5.12xlarge | 4x A10G | 96 GiB | 48 | 192 GiB | $4.096 |
| g4dn.16xlarge | 1x T4G | 16 GiB | 64 | 256 GiB | $4.352 |
| g3.16xlarge | 4x M60 | 32 GiB | 64 | 488 GiB | $4.56 |
| g5.16xlarge | 2x A10G | 24 GiB | 64 | 256 GiB | $5.672 |
| p2.8xlarge | 8x K80 | 96 GiB | 32 | 488 GiB | $7.20 |
| g4dn.metal | 8x T4G | 128 GiB | 96 | 384 GiB | $7.824 |
| g5.24xlarge | 4x A10G | 96 GiB | 96 | 384 GiB | $8.144 |
| p3.8xlarge | 4x V100 | 64 GiB | 32 | 244 GiB | $12.24 |
| p2.16xlarge | 16x K80 | 192 GiB | 64 | 732 GiB | $14.40 |
| g5.48xlarge | 8x A10G | 192 GiB | 192 | 768 GiB | $16.288 |
| p3.16xlarge | 8x V100 | 128 GiB | 64 | 488 GiB | $24.48 |
| p5.48xlarge | 8x H100 | 640 GB | 192 | 2048 GiB | $98.32 |
| p4d.24xlarge | 8x A100 | 320 GB | 96 | 1152 GiB | $32.7726 |
You can find more on AWS GPU instances here.
Google Cloud Platform
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| g2-standard-4 | 1x L4 | 24 GB | 4 | 16 GB | $0.706832 |
| g2-standard-8 | 1x L4 | 24 GB | 8 | 32 GB | $0.853624 |
| g2-standard-12 | 1x L4 | 24 GB | 12 | 48 GB | $1.000416 |
| g2-standard-16 | 1x L4 | 24 GB | 16 | 64 GB | $1.147208 |
| g2-standard-32 | 1x L4 | 24 GB | 32 | 128 GB | $1.734376 |
| g2-standard-24 | 2x L4 | 48 GB | 24 | 96 GB | $2.000833 |
| a2-highgpu-1g | 1x A100 | 40 GB | 12 | 85 GB | $3.673477 |
| g2-standard-48 | 4x L4 | 96 GB | 48 | 192 GB | $4.001665 |
| a2-highgpu-2g | 2x A100 | 80 GB | 24 | 170 GB | $7.346954 |
| g2-standard-96 | 8x L4 | 192 GB | 96 | 384 GB | $8.003331 |
| a2-highgpu-4g | 4x A100 | 160 GB | 48 | 340 GB | $14.693908 |
| a2-highgpu-8g | 8x A100 | 320 GB | 96 | 680 GB | $29.387816 |
| a2-megagpu-16g | 16x A100 | 640 GB | 96 | 1360 GB | $55.740976 |
You can find more on GCP GPU instances here.
Microsoft Azure
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| NC4as T4 v3 | 1x T4 | 16 GiB | 4 | 28 GiB | $0.526 |
| NC8as T4 v3 | 1x T4 | 16 GiB | 8 | 56 GiB | $0.752 |
| NC6 | 1x K80 | 12 GiB | 6 | 56 GiB | $0.90 |
| NV6 | 1x M60 | 8 GiB | 6 | 56 GiB | $1.14 |
| NV12s v3 | 1x M60 | 8 GiB | 12 | 112 GiB | $1.14 |
| NC16as T4 v3 | 1x T4 | 16 GiB | 16 | 110 GiB | $1.204 |
| NC12 | 2x K80 | 24 GiB | 12 | 112 GiB | $1.80 |
| NC6s v2 | 1x P100 | 16 GiB | 6 | 112 GiB | $2.07 |
| ND6s | 1x P40 | 24 GiB | 6 | 112 GiB | $2.07 |
| NV12 | 2x M60 | 16 GiB | 12 | 112 GiB | $2.28 |
| NV24s v3 | 2x M60 | 16 GiB | 24 | 224 GiB | $2.28 |
| NC6s v3 | 1x V100 | 16 GiB | 6 | 112 GiB | $3.06 |
| NC24 | 4x K80 | 48 GiB | 24 | 224 GiB | $3.60 |
| NC24r | 4x K80 | 48 GiB | 24 | 224 GiB | $3.96 |
| NC12s v2 | 2x P100 | 32 GiB | 12 | 224 GiB | $4.14 |
| ND12s | 2x P40 | 48 GiB | 12 | 224 GiB | $4.14 |
| NC64as T4 v3 | 4x T4 | 64 GiB | 64 | 440 GiB | $4.352 |
| NV48s v3 | 4x M60 | 32 GiB | 48 | 448 GiB | $4.56 |
| NV24 | 4x M60 | 32 GiB | 24 | 224 GiB | $4.56 |
| NC12s v3 | 2x V100 | 32 GiB | 12 | 224 GiB | $6.12 |
| ND24s | 4x P40 | 96 GiB | 24 | 448 GiB | $8.28 |
| NC24s v2 | 4x P100 | 64 GiB | 24 | 448 GiB | $8.28 |
| ND24rs | 4x P40 | 96 GiB | 24 | 448 GiB | $9.108 |
| NC24rs v2 | 4x P100 | 64 GiB | 24 | 448 GiB | $9.108 |
| NC24s v3 | 4x V100 | 64 GiB | 24 | 448 GiB | $12.24 |
| NC24rs v3 | 4x V100 | 64 GiB | 24 | 448 GiB | $13.46 |
| ND40rs v2 | 8x V100 | 256 GiB | 40 | 672 GiB | $22.032 |
| ND96asr A100 v4 | 8x A100 | 320 GiB | 96 | 900 GiB | $27.197 |
Linode
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| GPU1 | 1x RTX 6000 | 24 GB | 8 | 32 GB | $1.50 |
| GPU2 | 2x RTX 6000 | 48 GB | 16 | 64 GB | $3.00 |
| GPU3 | 3x RTX 6000 | 72 GB | 20 | 96 GB | $4.50 |
| GPU4 | 4x RTX 6000 | 96 GB | 24 | 128 GB | $6.00 |
You can find more on Linode GPU instances here.
Paperspace
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| GPU+ (M4000) | 1x M4000 | 8 GB | 8 | 30 GB | $0.45 |
| P4000 | 1x P4000 | 8 GB | 8 | 30 GB | $0.51 |
| RTX4000 | 1x RTX4000 | 8 GB | 8 | 30 GB | $0.56 |
| A4000 | 1x A4000 | 16 GB | 8 | 45 GB | $0.76 |
| P5000 | 1x P5000 | 16 GB | 8 | 30 GB | $0.78 |
| RTX5000 | 1x RTX5000 | 16 GB | 8 | 30 GB | $0.82 |
| P6000 | 1x P6000 | 24 GB | 8 | 30 GB | $1.10 |
| A5000 | 1x A5000 | 24 GB | 8 | 45 GB | $1.38 |
| A6000 | 1x A6000 | 48 GB | 8 | 45 GB | $1.89 |
| V100 | 1x V100 | 16 GB | 8 | 30 GB | $2.30 |
| V100-32G | 1x V100 | 32 GB | 8 | 30 GB | $2.30 |
| A100 | 1x A100 | 40 GB | 12 | 90 GB | $3.09 |
| A100-80G | 1x A100 | 80 GB | 12 | 90 GB | $3.19 |
| H100 | 1x H100 | 80 GB | 20 | 250 GB | $8.96 |
You can find more on Paperspace GPU instances here.
Scaleway
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| GPU-3070 | 1x RTX 3070 | 8 GB | 8 | 16 GB | €0.98 |
| Render-S | 1x P100 | 16 GB | 10 | 42 GB | €1.24 |
| H100-1-80G | 1x H100 | 80 GB | 24 | 240 GB | Starts at €1.90 |
| H100-2-80G | 2x H100 | 160 GB | 48 | 480 GB | Starts at €3.80 |
| L4-1-24G | 1x L4 | 24 GB | 8 | 48 GB | Coming soon |
| L4-2-24G | 2x L4 | 48 GB | 16 | 96 GB | Coming soon |
| L4-4-24G | 4x L4 | 96 GB | 32 | 192 GB | Coming soon |
| L4-8-24G | 8x L4 | 192 GB | 64 | 384 GB | Coming soon |
You can find more on Scaleway GPU instances here.
Lambda Labs
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| 1x RTX 6000 | 1x RTX 6000 | 24 GB | 14 | 46 GiB | $0.50 |
| 1x A10 | 1x A10G | 24 GB | 30 | 200 GiB | $0.75 |
| 1x A6000 | 1x A6000 | 48 GB | 14 | 100 GiB | $0.80 |
| 1x A100 | 1x A100 | 40 GB | 30 | 200 GiB | $1.29 |
| 2x A6000 | 2x A6000 | 96 GB | 28 | 200 GiB | $1.60 |
| 1x H100 | 1x H100 | 80 GB | 26 | 200 GiB | $2.49 |
| 2x A100 | 2x A100 | 80 GB | 60 | 400 GiB | $2.58 |
| 4x A6000 | 4x A6000 | 192 GB | 56 | 400 GiB | $3.20 |
| 8x V100 | 8x V100 | 128 GB | 92 | 448 GiB | $4.40 |
| 4x A100 | 4x A100 | 160 GB | 120 | 800 GiB | $5.16 |
| 8x A100 | 8x A100 | 320 GB | 124 | 1800 GiB | $10.32 |
| 8x A100 | 8x A100 | 640 GB | 240 | 1800 GiB | $14.32 |
| 8x H100 | 8x H100 | 640 GB | 208 | 1800 GiB | $27.92 |
You can find more on Lambda Labs GPU instances here.
OVHcloud
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| t2-le-45 | 1x V100S | 32 GB | 15 | 45 GB | $0.88 |
| t2-le-90 | 2x V100S | 64 GB | 30 | 90 GB | $1.76 |
| t2-45 | 1x V100S | 32 GB | 15 | 45 GB | $2.191 |
| t2-le-180 | 4x V100S | 128 GB | 60 | 180 GB | $3.53 |
| t2-90 | 2x V100S | 64 GB | 30 | 90 GB | $4.38 |
| t2-180 | 4x V100S | 128 GB | 60 | 180 GB | $8.763 |
You can find more on OVHcloud GPU instances here.
Fly.io
Fly.io charges for GPU usage in addition to the regular compute costs. Here's a breakdown of their on-demand GPU pricing:
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| L40S | 1x L40S | 48 GB | -- | -- | $2.50 |
| A100 40G PCIe | 1x A100 40G PCIe | 40 GB | -- | -- | $2.50 |
| A100 80G SXM | 1x A100 80G SXM | 80 GB | -- | -- | $3.50 |
You can find more on Fly.io GPU instances here.
Civo Cloud
Here's a breakdown of Civo Cloud's GPU pricing:
| Name | GPUs | GPU Memory | vCPUs | Memory | Price (hour) |
|---|---|---|---|---|---|
| Medium | 1 x L40S | 48 GB | 8 | 64 GB | $1.785714 |
| Medium | 2 x L40S | 96 GB | 16 | 128 GB | $3.571429 |
| Large | 4 x L40S | 192 GB | 32 | 255 GB | $7.142857 |
| Extra Large | 8 x L40S | 384 GB | 64 | 512 GB | $14.285714 |
You can find more on Civo Cloud GPU instances here.
Tip: You can use this website to compare cloud providers to find which one is the best fit for your use case.