阿里云
腾讯云
华为云
仙宫云 
OneThingAI 
优刻得9.9元/7天GPU Rental Price Comparison (Per GPU Hour)
Hourly GPU rental prices from Chinese and international platforms, collected 2026-10-23. Prices are per GPU; multi-GPU instances multiply them.
Cheapest option in this table: RTX 3090 at $0.149 per GPU hour on Vast.ai.
All GPU models at a glance
| GPU | VRAM | Cheapest (CNY/hr) | Cheapest (USD/hr) | Platforms |
|---|---|---|---|---|
| RTX 3090 | 24 GB | — | $0.149 (Vast.ai) | 3 |
| RTX 4090 | 24 GB | ¥2.15 (优云智算（UCloud）) | $0.34 (RunPod) | 4 |
| RTX 5090 | 32 GB | ¥3.32 (优云智算（UCloud）) | $0.536 (Vast.ai) | 4 |
| RTX A6000 | 48 GB | — | $0.33 (RunPod) | 3 |
| L40S | 48 GB | — | $0.79 (RunPod) | 3 |
| A100 80GB | 80 GB | — | $0.934 (Vast.ai) | 2 |
| H100 80GB | 80 GB | — | $1.788 (Vast.ai) | 6 |
| H200 | 141 GB | — | $3.59 (RunPod) | 3 |
Prices are per GPU per hour, collected 2026-10-23. Currencies are not converted.
Choose a GPU model
RTX 3090 (24 GB)
RTX 4090 (24 GB)
RTX 5090 (32 GB)
RTX A6000 (48 GB)
L40S (48 GB)
A100 80GB (80 GB)
H100 80GB (80 GB)
H200 (141 GB)
FAQ
What does a GPU hour cost in 2026?
It depends on the card. Consumer cards like the RTX 4090 are typically under $1 per hour on international platforms, while data-center cards such as the H100 and H200 cost several dollars per hour. Prices in this table are collected daily; see each model page for details.
Are these prices per card or per machine?
Per GPU. An 8-GPU server bills roughly eight times the per-card hourly rate. Some platforms offer a small discount for multi-GPU instances.
Why do Chinese and international platforms differ so much?
They are priced in different currencies and serve different markets; we list each in its original currency and do not convert. International platforms are often cheaper per hour but add network latency for users in mainland China.
How much VRAM do I need?
It depends on model size, precision and context length. Use our VRAM calculator to estimate requirements first, then compare prices for the GPUs that fit.