Vast.ai
PartnerRent cloud GPUs from anywhere. Save up to 90% vs AWS, Azure, GCP with a few clicks.
🇺🇸Los Angeles, CAFounded2018MarketplaceUpdated 34 minutes ago
Cheapest / hr
$0.01
Tesla P4
Most expensive
$12.00
T4
Average / GPU
$0.91
across 599 configs
GPU types
79
599 configurations
VRAM range
4–288GB
per single GPU
Regions
47
with offerings
GPU pricing by type
Median price per GPU/hour across all configurations offered by Vast.ai.
GPU configurations
1,011 offerings across 79 GPU types. Prices update every 4 hours.
599of 1,011
| Region | Total VRAM | Storage | Status | ||||||
|---|---|---|---|---|---|---|---|---|---|
4 GB VRAM per GPU· 2 configs |  | ×1 | 4 GB | 2 | 8 GB | 221 GB | $0.01–$0.28 | $0.01 | Available |
8 GB VRAM per GPU· 4 configs |  | ×1 | 8 GB | 0 | 16 GB | 163 GB | $0.02–$0.03 | $0.02 | Available |
6 GB VRAM per GPU· 4 configs | | ×1 | 6 GB | 8 | 8 GB | 256 GB | $0.02–$0.34 | $0.02 | Available |
16 GB VRAM per GPU· 19 configs | | ×1 | 16 GB | 10 | 31 GB | 66 GB | $0.03–$2.67 | $0.03 | Available |
10 GB VRAM per GPU· 11 configs |  | ×1 | 10 GB | 16 | 63 GB | 232 GB | $0.03–$0.12 | $0.03 | Available |
12 GB VRAM per GPU· 9 configs |  | ×2 | 24 GB | 0 | 15 GB | 59 GB | $0.03–$0.49 | $0.06 | Available |
6 GB VRAM per GPU· 5 configs | | ×2 | 12 GB | 1 | 4 GB | 146 GB | $0.03–$0.07 | $0.06 | Available |
8 GB VRAM per GPU· 5 configs | | ×2 | 16 GB | 4 | 15 GB | 292 GB | $0.03–$0.05 | $0.06 | Available |
6 GB VRAM per GPU· 4 configs | | ×6 | 36 GB | 64 | 126 GB | 3.06 TB | $0.03–$1.34 | $0.18 | Available |
12 GB VRAM per GPU· 2 configs |  | ×1 | 12 GB | 32 | 16 GB | 115 GB | $0.03–$0.06 | $0.03 | Available |
8 GB VRAM per GPU· 3 configs | | ×1 | 8 GB | 36 | 63 GB | 715 GB | $0.04–$0.06 | $0.04 | Available |
12 GB VRAM per GPU· 20 configs | | ×1 | 12 GB | 12 | 24 GB | 820 GB | $0.04–$0.68 | $0.04 | Available |
8 GB VRAM per GPU· 4 configs | | ×4 | 32 GB | 1 | 7 GB | 65 GB | $0.05–$0.07 | $0.20 | Available |
8 GB VRAM per GPU· 5 configs |  | ×1 | 8 GB | 4 | 16 GB | 40 GB | $0.05–$0.53 | $0.05 | Available |
8 GB VRAM per GPU· 2 configs |  | ×1 | 8 GB | 4 | 8 GB | 82 GB | $0.05–$0.07 | $0.05 | Available |
11 GB VRAM per GPU· 6 configs |  | ×2 | 22 GB | 40 | 47 GB | 764 GB | $0.05–$0.81 | $0.10 | Available |
12 GB VRAM per GPU· 5 configs | | ×4 | 48 GB | 8 | 31 GB | 1,007 GB | $0.05–$0.06 | $0.20 | Available |
8 GB VRAM per GPU· 2 configs | | ×2 | 16 GB | 19 | 21 GB | 372 GB | $0.05 | $0.10 | Available |
8 GB VRAM per GPU· 11 configs |  | ×1 | 8 GB | 5 | 8 GB | 195 GB | $0.06–$0.14 | $0.06 | Available |
11 GB VRAM per GPU· 8 configs | | ×1 | 11 GB | 3 | 9 GB | 101 GB | $0.06 | $0.06 | Available |
8 GB VRAM per GPU· 3 configs |  | ×4 | 32 GB | 32 | 63 GB | 340 GB | $0.06–$2.67 | $0.24 | Available |
8 GB VRAM per GPU· 3 configs | | ×1 | 8 GB | 11 | 21 GB | 233 GB | $0.07–$0.12 | $0.07 | Available |
12 GB VRAM per GPU· 4 configs | | ×8 | 96 GB | 80 | 252 GB | 929 GB | $0.07–$0.19 | $0.56 | Available |
16 GB VRAM per GPU· 7 configs | | ×1 | 16 GB | 12 | 31 GB | 809 GB | $0.07–$0.25 | $0.07 | Available |
12 GB VRAM per GPU· 8 configs |  | ×2 | 24 GB | 36 | 63 GB | 535 GB | $0.08–$0.14 | $0.16 | Available |
8 GB VRAM per GPU· 17 configs | | ×2 | 16 GB | 24 | 78 GB | 575 GB | $0.08–$0.80 | $0.16 | Available |
16 GB VRAM per GPU· 19 configs | | ×2 | 32 GB | 8 | 16 GB | 823 GB | $0.08–$0.47 | $0.16 | Available |
12 GB VRAM per GPU· 9 configs | | ×8 | 96 GB | 96 | 126 GB | 1.35 TB | $0.08–$0.13 | $0.64 | Available |
16 GB VRAM per GPU· 9 configs | | ×4 | 64 GB | 24 | 93 GB | 2.71 TB | $0.09–$0.20 | $0.36 | Available |
24 GB VRAM per GPU· 35 configs | | ×2 | 48 GB | 24 | 63 GB | 134 GB | $0.09–$0.87 | $0.18 | Available |
16 GB VRAM per GPU· 26 configs |  | ×8 | 128 GB | 64 | 126 GB | 1.37 TB | $0.09–$1.27 | $0.72 | Available |
12 GB VRAM per GPU· 2 configs | | ×2 | 24 GB | 48 | 63 GB | 788 GB | $0.09 | $0.18 | Available |
8 GB VRAM per GPU | -- | ×1 | 8 GB | 32 | 31 GB | 304 GB | $0.10 | $0.10 | Available |
24 GB VRAM per GPU· 4 configs | | ×2 | 48 GB | 10 | 31 GB | 328 GB | $0.11–$0.15 | $0.22 | Available |
24 GB VRAM per GPU |  | ×2 | 48 GB | 48 | 189 GB | 800 GB | $0.11 | $0.22 | Available |
20 GB VRAM per GPU | | ×1 | 20 GB | 8 | 251 GB | 354 GB | $0.11 | $0.11 | Available |
24 GB VRAM per GPU· 8 configs | | ×1 | 24 GB | 18 | 126 GB | 69 GB | $0.11–$0.67 | $0.11 | Available |
12 GB VRAM per GPU· 4 configs | | ×2 | 24 GB | 26 | 75 GB | 460 GB | $0.11–$0.19 | $0.22 | Available |
24 GB VRAM per GPU· 46 configs | | ×1 | 24 GB | 18 | 126 GB | 24 GB | $0.12–$4.00 | $0.12 | Available |
16 GB VRAM per GPU· 7 configs | | ×1 | 16 GB | 14 | 31 GB | 353 GB | $0.12–$0.40 | $0.12 | Available |
12 GB VRAM per GPU· 7 configs | | ×1 | 12 GB | 56 | 16 GB | 19 GB | $0.12–$0.23 | $0.12 | Available |
16 GB VRAM per GPU· 3 configs | | ×1 | 16 GB | 36 | 31 GB | 677 GB | $0.12–$12.00 | $0.12 | Available |
24 GB VRAM per GPU· 3 configs | | ×1 | 24 GB | 11 | 31 GB | 264 GB | $0.13–$0.15 | $0.13 | Available |
16 GB VRAM per GPU· 18 configs | | ×2 | 32 GB | 64 | 31 GB | 617 GB | $0.13–$1.34 | $0.26 | Available |
16 GB VRAM per GPU· 12 configs | | ×1 | 16 GB | 24 | 31 GB | 228 GB | $0.13–$0.80 | $0.13 | Available |
24 GB VRAM per GPU· 6 configs |  | ×2 | 48 GB | 24 | 94 GB | 1.64 TB | $0.13–$0.32 | $0.26 | Available |
8 GB VRAM per GPU | | ×2 | 16 GB | 8 | 31 GB | 617 GB | $0.13 | $0.26 | Available |
8 GB VRAM per GPU | | ×1 | 8 GB | 24 | 15 GB | 737 GB | $0.14 | $0.14 | Available |
16 GB VRAM per GPU· 2 configs | | ×1 | 16 GB | 12 | 63 GB | 1.26 TB | $0.19–$0.21 | $0.19 | Available |
24 GB VRAM per GPU· 3 configs |  | ×1 | 24 GB | 24 | 252 GB | 717 GB | $0.20–$0.33 | $0.20 | Available |
Showing first 50 of 79 matching configurations
Similar providers
See incorrect data?
Frequently asked questions
What GPUs does Vast.ai offer?
Vast.ai offers 79 GPU types across 599 configurations, including Tesla V100 16GB, Tesla P4, RTX 3080, RTX A2000, GTX 1050 Ti, RTX 3060 Laptop GPU and 73 more. The full list with live hourly pricing is shown above.
How much does Vast.ai cost?
Vast.ai GPU pricing starts at $0.01 per GPU-hour, with an average of $0.91 across its on-demand configurations. Spot and reserved options can be cheaper for interruptible or committed workloads.