🌐 ▾Simplified ChineseTraditional ChineseENJapanese한국어

GPU rental prices in the US (real-time comparison)

Data updated: 2026-11-12 15:45 · Regions are defined by each platform’s primary data center location · Prices reflect RMB per GPU per hour; USD amounts converted at 7.2.
Direct answer:As of 2026-11-12, this site monitors real-time quotes from 12 GPU platforms in North America; the lowest entry price is Vast.ai (USA) (¥0.63/hour, RTX 3090). A total of 18 GPU models are available; compared to this site’s global lowest price, 0 are cheaper, 6 are more expensive, and 12 match exactly—see per-GPU differences in the table below.

Platforms and entry-level pricing

PlatformLowest price on this platformCorresponding GPU modelAvailability status
Vast.ai (USA)¥0.63/hourRTX 30904 listings · spot
Salad (USA)¥0.65/hourRTX 30903 listings · spot
RunPod (USA)¥1.37/hourV100 16GB15 listings · spot
Cloudrift (USA)¥1.80/hourV100 16GB4 listings · pay-as-you-go
Thunder Compute (USA)¥2.52/hourA6000 48GB4 listings · pay-as-you-go
Akamai / Linode (USA)¥3.74/hourRTX 4000 Ada 20GB2 listings · pay-as-you-go
Massed Compute (USA)¥4.10/hourA6000 48GB9 listings · pay-as-you-go
Paperspace (USA)¥5.47/hourA100 80GB3 listings · pay-as-you-go
Lambda (USA)¥5.69/hourV100 16GB4 listings · pay-as-you-go
Crusoe (USA)¥10.80/hourL40S 48GB5 listings · pay-as-you-go
Vultr (USA)¥12.03/hourL40S 48GB1 listings · pay-as-you-go
Latitude.sh (Canada)¥12.10/hourH100 80GB1 listings · pay-as-you-go

Per GPU model: Lowest in this zone vs. Global Lowest Observed

GPU modelLowest in this regionPlatformGlobal Lowest ObservedThis region vs. global
RTX 3090¥0.63/hourVast.ai (USA)¥0.63/hourNo change
RTX 4090¥1.15/hourSalad (USA)¥1.15/hourNo change
RTX 5090¥1.53/hourVast.ai (USA)¥1.53/hourNo change
A100 80GB¥5.47/hourPaperspace (United States)¥5.47/hourNo change
H100 80GB¥8.28/hourPaperspace (United States)¥8.28/hourNo change
L40S 48GB¥4.97/hourCloudrift (United States)¥4.97/hourNo change
A6000 48GB¥2.38/hourRunPod (USA)¥2.38/hourNo change
H200 141GB¥25.85/hourRunPod (USA)¥16.49/hour56.8% higher
B200 180GB¥39.11/hourMassed Compute (United States)¥33.56/hour16.5% higher
L4 24GB¥3.17/hourRunPod (USA)¥2.09/hour51.7% higher
L40 48GB¥4.97/hourRunPod (USA)¥4.50/hour10.4% higher
A40 48GB¥2.52/hourRunPod (USA)¥2.52/hourNo change
RTX PRO 6000 96GB¥15.77/hourMassed Compute (United States)¥8.28/hour90.5% higher
RTX 6000 Ada 48GB¥5.33/hourRunPod (USA)¥5.33/hourNo change
V100 16GB¥1.37/hourRunPod (USA)¥0.88/hour55.7% higher
MI300X 192GB¥3.60/hourRunPod (USA)¥3.60/hourNo change
RTX 6000 24GB¥10.80/hourAkamai / Linode (United States)¥10.80/hourNo change
RTX 4000 Ada 20GB¥3.74/hourAkamai / Linode (United States)¥3.74/hourNo change

Metric Explanation

Only real-time quotes auto-captured today are included (reference prices and expired data excluded; platforms without full compatibility excluded). ‘Global Lowest Observed’ is the lowest live price across all platforms covered by this site—including mainland China platforms. Regions are defined by primary data center locations (North America includes U.S. and Canadian platforms); each platform appears in only one region. USD prices are converted at a 7.2 exchange rate for comparison—differences arise from both exchange rates and platform pricing strategies.

Related pages

China GPU Platform Zone · Cross-border Price Gap Page · Market Terminal · Price Difference for the Same GPU Across Borders · GHPI Price Index

🐆 Xiao Bao · AI AssistantPowered by Tongyi Qwen
Hello there, I’m Leopard~ Feel free to ask about GPU pricing, cost-effective rental strategies, or what Pro membership offers.