gpu comparison
find the right gpu.
self-host your models on 25 gpu types across 19+ cloud providers. real-time pricing, sorted by performance.
| gpu | vram | fp16 tflops | from |
|---|---|---|---|
 NVIDIA B200 | 192gb | 450 | $5.30/hr~$3816/mo |
| NVIDIA B300 | 288gb | 450 | $7.40/hr~$5328/mo |
| NVIDIA H100 | 80gb | 267.6 | $1.66/hr~$1195/mo |
| NVIDIA GH200 | 96gb | 267.6 | $2.29/hr~$1649/mo |
| NVIDIA H200 | 141gb | 267.6 | $2.45/hr~$1764/mo |
| NVIDIA GeForce RTX 5090 | 32gb | 209.5 | $0.65/hr~$468/mo |
| NVIDIA RTX PRO 6000 | 96gb | 125 | $1.25/hr~$900/mo |
| NVIDIA L40S | 48gb | 91.6 | $0.74/hr~$533/mo |
| NVIDIA RTX 6000 Ada | 48gb | 91.1 | $0.97/hr~$698/mo |
| NVIDIA L40 | 48gb | 90.5 | $0.99/hr~$713/mo |
| NVIDIA GeForce RTX 4090 | 24gb | 82.58 | $0.60/hr~$432/mo |
| NVIDIA A100 80GB | 80gb | 77.97 | $1.24/hr~$893/mo |
| NVIDIA A100 40GB | 40gb | 77.97 | $1.36/hr~$979/mo |
| NVIDIA RTX A6000 | 48gb | 38.7 | $0.49/hr~$353/mo |
| NVIDIA RTX 6000 | 24gb | 38.7 | $0.69/hr~$497/mo |
| NVIDIA A40 | 48gb | 37.4 | $1.10/hr~$792/mo |
| NVIDIA A10 | 24gb | 31.2 | $1.29/hr~$929/mo |
| NVIDIA L4 | 24gb | 30.3 | $0.95/hr~$684/mo |
| NVIDIA V100 16GB | 16gb | 28.26 | $0.39/hr~$281/mo |
| NVIDIA V100 32GB | 32gb | 28.26 | $2.34/hr~$1685/mo |
| NVIDIA RTX A5000 | 24gb | 27.8 | $0.44/hr~$317/mo |
| NVIDIA RTX 4000 Ada | 20gb | 26.7 | $0.79/hr~$569/mo |
| NVIDIA RTX A4000 | 16gb | 19.2 | $0.15/hr~$108/mo |
| NVIDIA A30 | 24gb | 10.3 | $0.35/hr~$252/mo |
| NVIDIA A16 | 16gb | 4.5 | $0.51/hr~$367/mo |
prices are live from 19 cloud providers. updated hourly.
19 cloud providers
verda
11 gpus · 2 regions
massedcompute
10 gpus · 4 regions
lambdalabs
9 gpus · 2 regions
paperspace
8 gpus · 1 regions
hyperstack
6 gpus · 1 regions
excesssupply
6 gpus · 1 regions
boostrun
6 gpus · 1 regions
digitalocean
5 gpus
latitude
4 gpus · 1 regions
scaleway
4 gpus · 3 regions
crusoe
4 gpus · 2 regions
imwt
3 gpus
vultr
3 gpus · 8 regions
denvr
3 gpus · 2 regions
nebius
2 gpus
phyntec
2 gpus · 1 regions
horizon
2 gpus
voltagepark
1 gpus
amaya
1 gpus
self-host your models.
spin up a gpu, deploy your model, own the stack. pay by the hour.
we use cookies
we use cookies to ensure you get the best experience on our website. for more information on how we use cookies, please see our cookie policy.
by clicking "accept", you agree to our use of cookies.
learn more.