🌐 ▾简中繁中EN日本語한국어

Real-time GPU compute price comparison for mainland China platforms

11 domestic platforms · 47 live price listings · Auto-crawled every 45 minutes · Updated: 2026-10-31 03:45
⚡ Direct answer— per rentgpu.link (GPU Compute Radar)’s real-time crawl on 2026-10-31, 47 live GPU listings across 11 mainland China compute platforms cover 17 GPU models; 10 platforms are scraped in real time, while the rest carry a yellow “Reference Price · Pending Verification” badge. The current lowest pay-as-you-go hourly rate among Chinese platforms is ¥0.76/hour (RTX 3090 @ Gpushare)These 11 platforms — AutoDL, Gpushare, MatPool, Compshare, Featurize, AIStation, Lanyun, MornAI, Suanjiayun, Paratera, and PPIO Paiou Cloud — rarely appear on overseas comparison sites like gpus.io, getdeploying, or GPU Finder. This site is among the few offering real-time cross-vendor price comparison for domestic platforms. Data refreshes every 45 minutes; CC BY 4.0 licensed for free reuse. Methodology and data sources detailed at Data。
Domestic platforms integrated
11 providers
real-time data scraped from 10 providers
Available pricing
47
All prices sourced from each platform’s publicly available pricing pages
Refresh frequency
45 minutes
Results are always sorted by price; ordering cannot be altered via top-up or other means.

Why are some domestic platforms not listed on the comparison site?

我们实测了十几个主流 GPU 比价站:gpus.io、getdeploying、GPU Finder、ComputePrices、grabgpu 等收录of都是 AWS / RunPod / Vast.ai / Lambda 这类海外云,AutoDL、AIStation、Suanjiayun、Gpushare、MatPool、Compshare、Featurize 等Domestic platforms几乎全部缺席— cross-vendor hourly pricing comparisons for domestic GPUs such as Huawei Ascend and Moore Threads are not offered by any platform.

本站ReplaceDomestic platforms当作主战场来接:已接入 11 providers,price全部来自各PlatformOfficial Website公开Pricing页或公开接口, 不Login、不破解、不采信No.三方转述。抓不到of就挂「Reference price · Verification pending」黄标,绝不拿过periodData充Real-time价。

Overview of domestic platforms (11 total)

PlatformAvailable GPU modelsCurrent LowestLast ScrapedStatus
AutoDL 6¥1.56 per hour · RTX 3090 2026-10-31 03:45Real-time Scraping
Gpushare 4¥0.76 per hour · RTX 3090 2026-10-31 03:45Real-time Scraping
MatPool 5¥1.35 per hour · T4 16GB 2026-10-31 03:45Real-time Scraping
Compshare 6¥1.19 per hour · RTX 3090 2026-10-31 03:45Real-time Scraping
Featurize 1¥1.62 per hour · RTX 3090 2026-10-31 03:45Real-time Scraping
AIStation 3¥1.10 per hour · RTX 3090 2026-10-31 03:45Real-time Scraping
Lanyun 1¥2.00 per hour · RTX 4090 —Reference Price
MornAI 3¥0.80 per hour · RTX 3090 2026-10-31 03:45Real-time Scraping
Suanjiayun 6¥1.14 per hour · RTX 4090D 24GB 2026-10-30 21:45Real-time Scraping
Paratera 9¥3.10 per hour · V100 32GB 2026-10-31 01:30Real-time Scraping
PPIO Paiou Cloud 3¥1.16 per hour · RTX 4090 2026-10-31 03:45Real-time Scraping

By GPU model: lowest domestic hourly rates (17 models)

GPU modelVRAMLowest domestic pricePlatformNotes
RTX 3090 24GB ¥0.76 /hour Gpushare 9 providers · pay-as-you-go
RTX 4090 24GB ¥1.16 /hour PPIO Paiou Cloud 10 providers · spot
RTX 5090 32GB ¥2.59 /hour Gpushare 7 providers · pay-as-you-go
A100 80GB 80GB ¥9.80 /hour Suanjiayun 3 providers · pay-as-you-go
A800 80GB 80GB ¥5.15 /hour MornAI 2 providers · pay-as-you-go
H800 80GB 80GB ¥10.50 /hour AutoDL 1 provider · pay-as-you-go
H100 80GB 80GB ¥11.88 /hour PPIO Paiou Cloud 1 provider · spot
L40S 48GB 48GB ¥5.08 /hour Suanjiayun 1 provider · pay-as-you-go
T4 16GB 16GB ¥1.35 /hour MatPool 2 providers · pay-as-you-go
A6000 48GB 48GB ¥2.20 /hour Gpushare 2 providers · pay-as-you-go
L4 24GB 24GB ¥4.10 /hour Paratera 1 provider · pay-as-you-go
H20 96GB 96GB ¥3.40 /hour PPIO Paiou Cloud 2 providers · spot
RTX 4090D 24GB 24GB ¥1.14 /hour Suanjiayun 2 providers · pay-as-you-go
RTX PRO 6000 96GB 96GB ¥7.35 /hour AutoDL 1 provider · pay-as-you-go
V100 32GB 32GB ¥3.10 /hour Paratera 1 provider · pay-as-you-go
A10 24GB 24GB ¥6.12 /hour Paratera 1 provider · pay-as-you-go
L20 48GB 48GB ¥11.25 /hour Paratera 1 provider · pay-as-you-go

FAQ

Which GPU rental platforms in Mainland China support price comparison?

We currently integrate 11 mainland China platforms: AutoDL, Gpushare, MatPool, Compshare, Featurize, AIStation, Lanyun, MornAI, Suanjiayun, Paratera, and PPIO Paiou Cloud. There are 47 live quotes in total, refreshed automatically every 45 minutes. Major domestic platforms—including AutoDL, AIStation, Suanjiayun, Gpushare, and MatPool—are largely absent from overseas comparison sites like gpus.io, getdeploying, and GPU Finder. We are among the few sources offering real-time, cross-vendor price comparison for Chinese platforms.

How much does it cost to rent a GPU for one hour on domestic platforms?

The lowest domestic pay-as-you-go hourly rate currently detected is ¥0.76/hour (RTX 3090 @ Gpushare). Price differences across GPU models are substantial—each row in the model-aggregated table above specifies the exact platform and billing method.

How do domestic platform prices compare with overseas platforms?

All prices are standardized to “RMB per GPU per hour” before sorting: foreign-currency quotes are converted using USD×7.2; daily/monthly plans are normalized via ÷24 and ÷730 respectively. Original prices and conversion logic are fully disclosed in billing notes—no hidden conversions.

Why do some domestic platforms show “reference pricing”?

This platform’s current crawler run failed to retrieve a valid real-time price (e.g., due to webpage redesign, login requirement, or API failure). In such cases, we retain the last successfully scraped price and mark it with a yellow “Reference price · To be verified” tag. It does not participate in lowest-price ranking or discount alerts. Crawl failures are logged; once resolved, the entry automatically reverts to real-time status.

Prices are collected from each platform’s public pricing pages and may lag or change.Order page takes precedence。部分「Rent Now」link含Promotion返佣, 不影响排序and收录。口径andData Source见 Data。
🐆 Leopard · AI Customer SupportPowered by Tongyi Qwen
Hello there, I’m Leopard~ Feel free to ask about GPU pricing, cost-effective rental strategies, or what Pro membership offers.