Marketplace / RTX 5090
Rent an RTX 5090 by the hour
The fastest single consumer GPU. Best tokens per second for models that fit in 32 GB. You get the whole machine and root over SSH, with no minimum term.
- From, per GPU-hour
- $0.515
- Typical, per GPU-hour
- $0.881
- Memory per GPU
- 32 GB
- Machines free now
- 118
Free right now
Live, read 2026-11-15 16:48 UTC, 24 countries
Loading…
118 machines
| GPU | CPU and RAM | Location | Network | Rent | ||||
|---|---|---|---|---|---|---|---|---|
| RTX 5090 | 32 GB | AMD EPYC 7K6224 cores, 63 GB, 301 GB disk | China, CN | 833 Mbps 735 Mbps | 98.6% | $0.0174 | $0.556 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 7K6224 cores, 41 GB, 262 GB disk | China, CN | 211 Mbps 81 Mbps | 99.5% | $0.0179 | $0.572 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 7 7800X3D16 cores, 30 GB, 523 GB disk | South Korea, KR | 378 Mbps 323 Mbps | 99.8% | $0.0187 | $0.597 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 7 7800X3D16 cores, 30 GB, 406 GB disk | South Korea, KR | 500 Mbps 408 Mbps | 99.8% | $0.0187 | $0.597 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 7 7800X3D8 cores, 30 GB, 198 GB disk | South Korea, KR | 497 Mbps 480 Mbps | 99.8% | $0.0187 | $0.597 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 7 7800X3D8 cores, 30 GB, 614 GB disk | South Korea, KR | 403 Mbps 208 Mbps | 99.6% | $0.0187 | $0.597 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 7 7800X3D8 cores, 30 GB, 316 GB disk | South Korea, KR | 457 Mbps 506 Mbps | 99.8% | $0.0187 | $0.597 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 72828 cores, 47 GB, 103 GB disk | Japan, JP | 660 Mbps 68 Mbps | 99.6% | $0.0191 | $0.611 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 7 7800X3D16 cores, 30 GB, 192 GB disk | South Korea, KR | 531 Mbps 524 Mbps | 99.0% | $0.0202 | $0.645 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 776364 cores, 63 GB, 111 GB disk | Anhui, CN | 1.7 Gbps 848 Mbps | 99.3% | $0.0211 | $0.674 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 7B1232 cores, 63 GB, 57 GB disk | Thailand, TH | 659 Mbps 888 Mbps | 99.3% | $0.0222 | $0.711 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 9475F12 cores, 31 GB, 282 GB disk | Bulgaria, BG | 813 Mbps 540 Mbps | 99.4% | $0.0231 | $0.740 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 5 360012 cores, 63 GB, 179 GB disk | Poland, PL | 875 Mbps 569 Mbps | 99.4% | $0.0234 | $0.750 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 751316 cores, 63 GB, 172 GB disk | Texas, US | 836 Mbps 827 Mbps | 99.5% | $0.0241 | $0.771 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 745213 cores, 63 GB, 1129 GB disk | Estonia, EE | 854 Mbps 821 Mbps | 98.5% | $0.0243 | $0.776 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 7B1332 cores, 63 GB, 161 GB disk | Thailand, TH | 766 Mbps 883 Mbps | 99.6% | $0.0243 | $0.779 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen Threadripper PRO 7965WX 24-Cores8 cores, 42 GB, 119 GB disk | Vietnam, VN | 810 Mbps 732 Mbps | 99.5% | $0.0246 | $0.786 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 9B1443 cores, 42 GB, 592 GB disk | China, CN | 6.7 Gbps 7.2 Gbps | 99.3% | $0.0253 | $0.809 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 973456 cores, 39 GB, 510 GB disk | Michigan, US | 19.6 Gbps 14.9 Gbps | 99.8% | $0.0258 | $0.827 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen Threadripper 9960X 24-Cores16 cores, 42 GB, 360 GB disk | Illinois, US | 1.6 Gbps 280 Mbps | 99.4% | $0.0261 | $0.836 | Rent |
| RTX 5090 | 32 GB | AMD EPYC 764224 cores, 63 GB, 644 GB disk | British Columbia, CA | 1.4 Gbps 197 Mbps | 99.6% | $0.0266 | $0.851 | Rent |
| RTX 5090 | 32 GB | Core™ i7-5930K6 cores, 31 GB, 629 GB disk | South Korea, KR | 592 Mbps 397 Mbps | 99.3% | $0.0266 | $0.851 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 9 7940HX with Radeon Graphics32 cores, 61 GB, 294 GB disk | Texas, US | 717 Mbps 754 Mbps | 98.3% | $0.0267 | $0.856 | Rent |
| RTX 5090 | 32 GB | AMD Ryzen 9 9950X16 cores, 62 GB, 1491 GB disk | Minnesota, US | 947 Mbps 936 Mbps | 99.5% | $0.0271 | $0.866 | Rent |
| RTX 5090 | 32 GB | Core™ Ultra 9 285K24 cores, 63 GB, 2679 GB disk | Australia, AU | 1.8 Gbps 537 Mbps | 99.6% | $0.0272 | $0.869 | Rent |
What fits in 32 GB
Models up to about 28 GB: 30B class at 6-bit, 70B class does not fit on one card. Fast image and video generation.
Machines with two or four cards add their memory together for runtimes that split a model across GPUs.
Next to the other GPUs, lowest price per GPU-hour
The same list over the API
No key needed. The response is JSON with what each machine costs right now.
$ curl "gpuvox.com/api/v1/offers?gpu=RTX%205090&limit=5"