GPU prices
Price per GPU-hour for 48 GPU models, from hyperscalers to GPU clouds and marketplaces.
Checked 2 min agoCheapest H100, per GPU-hour
$2.8900
RunPod, on-demand
Median H100
$6.8800
26 on-demand offers
Cheapest A100 80 GB
$1.7900
RunPod
Cheapest L40S
$1.0900
RunPod
Cheapest L4
$0.2020
AWS
Offers
457
12 providers
Every GPU model
Per GPU-hour, cheapest first
| GPU | Best on-demand | Spot or community | Marketplace low, median | Best hyperscaler | Median on-demand | Offers | Box sizes |
|---|---|---|---|---|---|---|---|
| L4 24 GB, NVIDIA Ada | $0.2020 AWS | $0.0227 AWS | $0.3609, $0.3618 7 listings | $0.2020 AWS | $0.9971 | 57 | 1, 2, 4, 8 |
| RTX-2000-Ada NVIDIA | $0.2400 RunPod | $0.5000 RunPod | — | — | $0.2400 | 2 | 1 |
| A40 48 GB, NVIDIA Ampere | $0.2500 RunPod | $0.1700 RunPod | $0.4678, $0.5344 7 listings | — | $1.0453 | 12 | 1, 2, 4, 8 |
| RTX-A4500 NVIDIA | $0.2500 RunPod | $0.1900 RunPod | — | — | $0.2500 | 2 | 1 |
| RTX-A5000 NVIDIA | $0.2700 RunPod | $0.1600 RunPod | — | — | $0.2700 | 2 | 1 |
| RTX-4000-ADA NVIDIA | $0.2800 RunPod | $0.1800 RunPod | — | — | $0.7050 | 14 | 1, 2, 4 |
| V520 NVIDIA | $0.3785 AWS | $0.1118 AWS | — | $0.3785 AWS | $0.8670 | 10 | 1, 2, 4 |
| T4 16 GB, NVIDIA Turing | $0.4200 AWS | $0.0684 Azure | — | $0.4200 AWS | $0.9780 | 34 | 1, 2, 4, 8 |
| RTX-3090-Ti NVIDIA | $0.4600 RunPod | $0.2700 RunPod | — | — | $0.4600 | 2 | 1 |
| RTX PRO 6000 96 GB, NVIDIA Blackwell | $0.5000 RunPod | $1.6400 RunPod | $1.3363, $1.6022 54 listings | — | $2.0833 | 12 | 1, 2, 4, 8 |
| RTX-3090 NVIDIA | $0.5000 RunPod | $0.2200 RunPod | $0.1891, $0.3841 37 listings | — | $0.5000 | 4 | 1 |
| RTX-4070-Ti NVIDIA | $0.5000 RunPod | $0.1900 RunPod | — | — | $0.5000 | 2 | 1 |
| RTX-4080 NVIDIA | $0.5000 RunPod | $0.2700 RunPod | — | — | $0.5000 | 2 | 1 |
| RTX-4080-SUPER NVIDIA | $0.5000 RunPod | $0.2800 RunPod | — | — | $0.5000 | 2 | 1 |
| RTX-A2000 NVIDIA | $0.5000 RunPod | $0.1200 RunPod | — | — | $0.5000 | 2 | 1 |
| RTX-PRO-4000 NVIDIA | $0.5700 RunPod | $0.5000 RunPod | — | — | $0.5700 | 2 | 1 |
| RTX A6000 48 GB, NVIDIA Ampere | $0.5900 RunPod | $0.3300 RunPod | $0.3911, $0.4006 3 listings | — | $0.5900 | 4 | 1 |
| RTX-5080 NVIDIA | $0.5900 RunPod | $0.3900 RunPod | — | — | $0.5900 | 2 | 1 |
| RTX-PRO-4500 NVIDIA | $0.7200 RunPod | $0.3400 RunPod | — | — | $0.7200 | 2 | 1 |
| RTX-PRO-4500-SE NVIDIA | $0.7200 RunPod | $0.5000 RunPod | — | — | $0.7200 | 2 | 1 |
| M60 NVIDIA | $0.7500 AWS | $0.2107 Azure | — | $0.7500 AWS | $1.1400 | 10 | 1, 2, 4 |
| L40 48 GB, NVIDIA Ada | $0.8200 RunPod | $0.6900 RunPod | — | — | $0.8200 | 2 | 1 |
| V100 16 GB, NVIDIA Volta | $0.8299 Exoscale, 4-GPU box | $0.1900 RunPod | — | $2.7540 Azure, 8-GPU box | $2.9500 | 31 | 1, 2, 3, 4, 8 |
| RTX-5000-Ada NVIDIA | $0.8300 RunPod | $0.4900 RunPod | — | — | $0.8300 | 2 | 1 |
| RTX 4090 24 GB, NVIDIA Ada | $0.8900 RunPod | $0.3400 RunPod | $0.4009, $0.8014 64 listings | — | $0.8900 | 4 | 1 |
| K80 NVIDIA | $0.9000 AWS | — | — | $0.9000 AWS | $1.8000 | 3 | 1, 4, 8 |
| RTX-PRO-5000 NVIDIA | $0.9600 RunPod | $0.8200 RunPod | — | — | $0.9600 | 2 | 1 |
| RTX-6000-ADA NVIDIA | $0.9900 RunPod | $0.7400 RunPod | $0.5971, $0.6700 8 listings | — | $1.2800 | 5 | 1 |
| A10G 24 GB, NVIDIA Ampere | $1.0060 AWS | $0.1408 AWS | — | $1.0060 AWS | $1.8300 | 16 | 1, 4, 8 |
| L40S 48 GB, NVIDIA Ada | $1.0900 RunPod | $0.5000 AWS, 4-GPU box | $0.5341, $0.7680 14 listings | $1.8610 AWS | $1.8000 | 34 | 1, 2, 4, 8 |
| RTX-5090 NVIDIA | $1.1900 RunPod | $0.6900 RunPod | $0.5723, $1.1148 58 listings | — | $1.1900 | 4 | 1 |
| P100 NVIDIA | $1.3681 Scaleway | — | — | — | $1.3681 | 1 | 1 |
| RTX-6000 NVIDIA | $1.5000 Linode | — | — | — | $1.5000 | 4 | 1, 2, 3, 4 |
| A100 80 GB 80 GB, NVIDIA Ampere | $1.7900 RunPod | $0.6788 Azure | $0.9339, $1.1361 29 listings | $3.3996 Azure, 8-GPU box | $3.6730 | 32 | 1, 2, 4, 8 |
| H200 141 GB, NVIDIA Hopper | $1.9900 Vultr | $0.5000 RunPod | $3.5927, $4.2119 27 listings | $7.9120 AWS, 8-GPU box | $6.6010 | 18 | 1, 8 |
| A10 24 GB, NVIDIA Ampere | $2.0000 Oracle, 4-GPU box | $0.4066 Azure | — | $2.0000 Oracle, 4-GPU box | $2.2000 | 13 | 1, 2, 4 |
| MI300X 192 GB, AMD CDNA 3 | $2.3900 RunPod | $0.5000 RunPod | — | $6.0000 Oracle, 8-GPU box | $2.5900 | 7 | 1, 8 |
| MI355X NVIDIA | $2.5900 Vultr, 8-GPU box | $5.5000 DigitalOcean, 8-GPU box | — | $8.6000 Oracle, 8-GPU box | $2.5900 | 4 | 8 |
| A100 40 GB 40 GB, NVIDIA Ampere | $2.7447 AWS, 8-GPU box | $0.4784 AWS, 8-GPU box | $0.4004, $0.5563 18 listings | $2.7447 AWS, 8-GPU box | $3.6734 | 16 | 1, 2, 4, 8, 16 |
| H100 80 GB, NVIDIA Hopper | $2.8900 RunPod | $1.2899 Azure | $1.7878, $3.0746 6 listings | $6.8800 AWS, 8-GPU box | $6.8800 | 46 | 1, 2, 4, 8 |
| MI325X 256 GB, AMD CDNA 3 | $3.8000 DigitalOcean | $2.0000 Vultr, 8-GPU box | — | — | $3.8000 | 4 | 1, 8 |
| MI350X NVIDIA | $5.4900 RunPod | $0.5000 RunPod | — | — | $6.1600 | 5 | 1, 8 |
| B200 180 GB, NVIDIA Blackwell | $7.9900 RunPod | $2.0081 AWS, 8-GPU box | $6.2510, $10.252 10 listings | $14.000 Oracle, 8-GPU box | $8.5000 | 11 | 1, 8 |
| B300 NVIDIA | $8.4038 Scaleway, 8-GPU box | $2.5101 AWS, 8-GPU box | — | $17.802 AWS, 8-GPU box | $11.190 | 8 | 1, 8 |
| GB200 192 GB, NVIDIA Blackwell | $16.000 Oracle, 4-GPU box | — | — | $16.000 Oracle, 4-GPU box | $16.000 | 1 | 4 |
| RTX-3070 NVIDIA | — | $0.1300 RunPod | — | — | — | 1 | 1 |
| RTX-3080 NVIDIA | — | $0.1700 RunPod | — | — | — | 1 | 1 |
| RTX-3080-Ti NVIDIA | — | $0.1800 RunPod | — | — | — | 1 | 1 |
Price against memory
Cheapest on-demand, $ per GPU-hour against GB per GPU
Where H100 is cheapest
Get GPU prices from the API
curl "https://prices.cloudadapt.io/api/v1/gpus"
Per GPU-hour is the box price divided by its GPUs. Virtual machines are priced in their cheapest region; GPU clouds and marketplaces at their listed price. Marketplaces publish the cheapest and median listing, not individual offers. US dollars, before discounts and taxes.