Pay-as-you-go

Updated at:

Pay-as-you-go instances are billed after use. Each billing item is metered hourly, with any partial hour charged as a full hour. You can purchase a GA resource plan in advance to offset the instance fee and CU fee and reduce costs. Before purchasing, use the GA (pay-as-you-go) CU calculator to estimate CU consumption.

Billing item

Applicable instances

Billing product

Instance fee

Standard + Basic

Global Accelerator (GA)

CU fee

Standard

Global Accelerator (GA)

Data transfer fee

Standard + Basic

Cloud Data Transfer (CDT)

To use pay-as-you-go GA instances for the first time, activate the service on the service activation page.
All unit prices in this document are catalog prices for reference only. Actual prices are shown on the console purchase page or in your bill details.

Instance fee

The instance fee covers instance occupancy and is charged based on the unit price and billing duration. Both standard and basic GA instances are charged an instance fee.

  • Formula: Instance fee = Unit price (CNY/hour) x Billing duration (hours)

  • Unit price: 0.137 CNY/hour

  • Billing duration:

    • Metered hourly; any partial hour is charged as a full hour.

    • Billing starts the moment the instance is created, regardless of whether you finish configuring or using it. Instances in a pending-configuration state are billed the same way. Releasing the instance is the only way to stop billing — release it promptly when no longer needed.

CU fee

GA measures the resource consumption of standard instances in CUs (Capacity Units). The CU fee is based on the hourly CU usage and billing duration. Basic instances are not charged a CU fee.

  • Formula:

  Hourly CU fee = CU unit price (CNY/CU/hour) x Number of CUs per hour
  Number of CUs per hour = MAX {New Connections CU count, Concurrent Connections CU count, Data Processing CU count, Rule Evaluation CU count}
  • CU unit price: 0.386 CNY/CU/hour

  • Billing rule:

    • The hourly CU count is converted from various performance metrics, and you are charged based on the metric with the highest CU usage. Currently only Data Processing CUs are charged; new connections, concurrent connections, and rule evaluations are not yet billed.

    • The CU fee for a single instance is the sum of CU fees across all its listeners.

    • Minimum precision: 0.000001 CU.

CU usage conversion

The performance metrics contained in one CU depend on the listener protocol. Divide each metric by its CU coefficient to get the number of CUs.

Traffic type

Metric

CU coefficient

Metering interval

TCP

New connections

800

Second

TCP

Concurrent connections

100,000

Minute

TCP

Data Processing

1 GB

Hour

UDP

New connections

400

Second

UDP

Concurrent connections

50,000

Minute

UDP

Data Processing

1 GB

Hour

HTTP(S)

New connections

25

Second

HTTP(S)

Concurrent connections

3,000

Minute

HTTP(S)

Data Processing

1 GB

Hour

HTTP(S)

Rule evaluations

1,000

N/A

HTTP(S) rule evaluations are affected by the number of forwarding rules and extended certificates. Both have a free quota of 25 — only the excess is billed:

  • When the processed rule count exceeds the free quota: Rule evaluations = QPS x (excess forwarding rules + excess extended certificates)

  • When the processed rule count does not exceed the free quota: Rule evaluations = QPS

CU billing example

On 2026-06-02 at 08:10:00 a pay-as-you-go standard GA instance is created with a TCP listener configured. The instance is released at 08:50:00 the same day. During the 08:10–08:50 billing cycle, the TCP listener records: max new connections 4,000, max concurrent connections 720,000, processed data volume 10 GB.

New Connection CUs = 4,000 / 800 = 5
Concurrent Connection CUs = 720,000 / 100,000 = 7.2
Data Processing CUs = 10 / 1 = 10
Hourly CU fee = 0.386 x MAX{5, 7.2, 10} = 0.386 x 10 = 3.86 CNY

Since only Data Processing CUs are currently charged: hourly CU fee = 0.386 x 10 = 3.86 CNY.

Data transfer fee

Data transfer fees for pay-as-you-go GA instances are settled by Cloud Data Transfer (CDT).

Data transfer types

image
  • Internet traffic from client to acceleration region: Access lines support BGP multi-line (default, available in all acceleration regions) and BGP Pro (available only in China (Hong Kong) and Japan (Tokyo), optimized for overseas-to-Chinese-mainland traffic).

  • Transfer traffic over the accelerated network between acceleration region and endpoint group:

    • Cross-region traffic: Chinese mainland region to Chinese mainland region, or overseas region to overseas region.

    • Cross-border traffic: Chinese mainland region to Hong Kong (China), Macao (China), Taiwan (China), or other countries and regions.

      Ensure the accelerated application domain has completed ICP filing, otherwise GA cannot be used.

      Choose the transmission quality type:

      • Premium Bandwidth For Cross-border Acceleration: uses a BGP Pro line and simplifies the cross-border qualification process.

      • China Unicom Cross-border Leased Line: uses a Unicom dedicated line for better acceleration but requires applying for a China Unicom cross-border business qualification and completing cross-border compliance certification.

    • No transfer fee is charged when the acceleration region and endpoint region are the same.

  • Internet traffic from endpoint group to backend service: Internet traffic is generated based on the backend connection method. Private-network traffic is not charged.

Billing rules

Traffic type

Billing rule

Service tier

Price

Internet traffic (client to acceleration region, endpoint group to backend)

Monthly tiered billing per region. Tiers reset to zero at the start of each calendar month.

Internet traffic

Cross-region traffic

Charged per region pair.

Gold

Inter-region data transfer

Cross-border traffic (Premium Bandwidth For Cross-border Acceleration)

Charged per region pair.

Gold

6.4 CNY/GB

Cross-border traffic (China Unicom Cross-border Leased Line)

Charged per region pair.

Gold

5.6 CNY/GB (operated and sold by China Unicom)

Fees charged by scenario

Acceleration region

Endpoint group region

Backend service

Acceleration-region internet fee

Endpoint-region internet fee

Cross-region fee

Cross-border fee

Chinese mainland

Chinese mainland

Alibaba Cloud

Charged

N/A

Charged

N/A

Non-Alibaba Cloud

Charged

Charged

Charged

N/A

HK/Macao/Taiwan (China) or other countries

Alibaba Cloud

Charged

N/A

N/A

Charged

Non-Alibaba Cloud

Charged

Charged

N/A

Charged

HK/Macao/Taiwan (China) or other countries

HK/Macao/Taiwan (China) or other countries

Alibaba Cloud

Charged

N/A

Charged

N/A

Non-Alibaba Cloud

Charged

Charged

Charged

N/A

Chinese mainland

Alibaba Cloud

Charged

N/A

N/A

Charged

Non-Alibaba Cloud

Charged

Charged

N/A

Charged

Data transfer billing examples

Example 1: Chinese mainland client accessing a non-Alibaba Cloud application in the Chinese mainland

Acceleration region: China (Shanghai). Endpoint group region: China (Beijing). Internet connection to backend. Traffic used in one month: 1,000 GB.

Chinese mainland internet traffic unit price = 0.80 CNY/GB
Chinese mainland cross-region traffic unit price = 0.48 CNY/GB
Shanghai internet traffic fee = 0.80 x 1000 = 800 CNY
Beijing internet traffic fee = 0.80 x 1000 = 800 CNY
Cross-region traffic fee = 0.48 x 1000 = 480 CNY
Total data transfer fee = 800 + 800 + 480 = 2,080 CNY

Example 2: Chinese mainland client accessing an overseas Alibaba Cloud application

Acceleration region: China (Shanghai). Endpoint group region: US (Silicon Valley). Private-network connection to backend. Traffic used in one month: 1,000 GB. Transmission quality: premium bandwidth for cross-border acceleration.

Chinese mainland internet traffic unit price = 0.80 CNY/GB
Premium bandwidth cross-border traffic unit price = 6.4 CNY/GB
Shanghai internet traffic fee = 0.80 x 1000 = 800 CNY
US (Silicon Valley) internet traffic fee = 0 CNY (private-network connection, no charge)
Cross-border traffic fee = 6.4 x 1000 = 6,400 CNY
Total data transfer fee = 800 + 0 + 6400 = 7,200 CNY

Cost optimization

GA resource plan

A GA resource plan is a prepaid offset package using CU as the unified unit. It automatically offsets the instance fee and CU fee of pay-as-you-go GA instances at a defined ratio. It does not offset data transfer fees.

Plan specification

Price (CNY)

Approx. instance fee coverage

Approx. processed-data CU fee coverage

1,000 CU

378

30 days

740 GB

5,000 CU

1,853

90 days

4,230 GB

10,000 CU

3,667

365 days

6,890 GB

50,000 CU

17,756

365 days

46,890 GB

  • Purchase: Go to GA console - GA Resource Plan, choose Buy Now. Select the specification, effective time, and validity period.

    • Effective time: You can choose immediate activation after payment or a scheduled activation time.

    • Offset order: Multiple plans can be stacked.

      • They are consumed in order of expiration (earliest first). When expiration dates are the same, the plan purchased first is consumed first.

      • Usage exceeding the purchased quota is automatically billed as pay-as-you-go.

    • Offset ratio:

      • The instance fee is offset using a coefficient (approximately unit price / CU price = 0.354922 CU per hour).

      • The CU fee is offset 1:1 based on actual hourly consumption.

    • Validity: 1 year from the purchase date. Unused quota is automatically cleared at expiration and cannot be extended.

  • View usage details: Go to the Resource Plan page to check remaining quota and offset records.

  • Unsubscribe: Initiate unsubscription on the Resource Unsubscription page in the Billing Management console. Refer to Introduction to resource plans and purchasing for detailed unsubscription rules.

CDT resource plan

CDT resource plan offsets internet traffic across all billing tiers on BGP (multi-line) routes (beyond the free quota).

Data Transfer Plan also offsets CDT internet traffic, but only covers the first billing tier (0–10 TB) on BGP (multi-line) routes.

No resource plan currently offsets cross-region or cross-border traffic.