Pay-as-you-go
The pay-as-you-go billing method allows you to pay for resources after they are used. If you use the pay-as-you-go billing method, you pay only for the Function Compute resources that you use. You do not need to purchase resources in advance. This topic describes the resources that support the pay-as-you-go billing method and the settlement rules.
Prices
CU usage is billed monthly on a tiered basis. Tiers are calculated independently for each region. CU usage in one region is not merged with CU usage in other regions. For more information, see Billing example.
| Tier | Monthly CU usage (unit: CU) | Standard unit price | Official website discounted price (promotion period: Aug 27, 2024 to Aug 27, 2026) |
| Tier 1 | (0, 200 million] | CNY 0.00011/CU | CNY 0.000088/CU |
| Tier 2 | (200 million, 1 billion] | CNY 0.00010/CU | CNY 0.000080/CU |
| Tier 3 | > 1 billion | CNY 0.00009/CU | CNY 0.000072/CU |
CU conversion coefficients per resource usage
CPU Services
| Billing item | vCPU usage | Memory usage | Function invocations | Disk usage | Deep sleep storage usage | Snapshot storage usage |
| Unit | CU / (vCPU * second) | CU / (GB * second) | CU / 10,000 invocations | CU / (GB * second) | CU / (GB * second) | CU / (GB * second) |
| Elastic instance (active) CU conversion coefficient | 1.0 | 0.15 | 75 | 0.05 | 0 | 0 |
| Elastic instance (light sleep, formerly idle) CU conversion coefficient | 0 | 0.15 | 0 | 0.05 | 0 | 0 |
| Elastic instance (deep sleep) CU conversion coefficient | 0 | 0 | 0 | 0 | 0.05 | 0 |
| Snapshot CU conversion coefficient | 0 | 0 | 0 | 0 | 0 | 0.05 |
Active state
When an elastic instance is in the active state, you are charged for vCPU, memory, and disk usage. Disk usage includes a 10 GB free quota.
Fee = vCPU fee + Memory fee + Function invocation fee + Disk fee
Billable disk usage = max(0, Disk size - 10) GBLight sleep state
When an elastic instance enters the light sleep state, you are not charged for vCPU usage. You are charged only for memory and disk usage. Disk usage includes a 10 GB free quota.
Fee = Memory fee + Disk fee
Billable disk usage = max(0, Disk size - 10) GBDeep sleep state
When an elastic instance enters the deep sleep state, you are not charged for vCPU or memory usage. You are charged for disk usage based on deep sleep storage usage. The 10 GB free disk quota does not apply during the deep sleep state.
Fee = Deep sleep storage fee
Deep sleep storage usage = Memory size × 2 + Disk sizeSnapshot fee
After the Snapshot feature is enabled, billing follows the Snapshot pricing rules.
Snapshot storage usage = Memory size × 2 + Disk size
Fee = Snapshot storage usage × Disk unit price × DurationGPU Services
| Resource or billing item | Unit | Elastic instance (active) CU conversion coefficient | Elastic instance (light sleep) CU conversion coefficient |
| Function invocations | CU / 10,000 invocations | 75 | — |
| vCPU usage | CU / (vCPU * second) | 1 | 0 |
| Memory usage | CU / (GB * second) | 0.15 | 0.10 |
| Disk usage | CU / (GB * second) | 0.05 | 0.05 |
| tesla.1 GPU usage | CU / (GB VRAM * second) | 2.1 | 0.5 |
| ampere.1 GPU usage | CU / (GB VRAM * second) | 1.8 | 0.3 |
| ada.1 GPU usage | CU / (GB VRAM * second) | 1.7 | 0.2 |
| ada.2 GPU usage | CU / (GB VRAM * second) | 1.95 | 0.23 |
| ada.3 GPU usage | CU / (GB VRAM * second) | 1.95 | 0.23 |
| blackwell.1 GPU usage | CU / (GB VRAM * second) | 2.1 | 0.28 |
| hopper.1 GPU usage | CU / (GB VRAM * second) | 2.31 | 0.315 |
| hopper.2 GPU usage | CU / (GB VRAM * second) | 2.31 | 0.315 |
| xpu.1 GPU usage | CU / (GB VRAM * second) | 1.2 | 0.23 |
Billing granularity
The billing granularity for the execution duration of different instance types is as follows:
| Instance type | On-demand mode | Provisioned mode |
| CPU instance | Billed in 1-millisecond increments. |
Billed in 10-second increments. Usage of less than 10 seconds is rounded up to 10 seconds. Note
For example, if a provisioned instance runs for 51 seconds, you are billed for 60 seconds. If it runs for 61 seconds, you are billed for 70 seconds. |
| GPU instance |
Billed in 1-second increments. Usage of less than 1 second is rounded up to 1 second. Note
For example, if an on-demand GPU instance runs for 51 milliseconds, you are billed for 1 second. If it runs for 10.5 seconds, you are billed for 11 seconds. |
-
The execution duration refers to the actual execution time of a function. The calculation method varies by instance mode. For more information, see Billing overview.
-
The billing cycle is 1 hour. Fees accumulate by billing granularity and are settled hourly.
Billing cycle
Bills are settled hourly. Bills are typically generated 1 to 2 hours after the end of each billing cycle, subject to the actual system generation time.
The billing cycle of Function Compute is 1 hour. Within a billing cycle, the CU usage metered for each function is rounded up to the nearest integer. The CU usage of all functions is then accumulated as the total CU usage. After deductions from resource plans, if you purchased any, the total CU usage is used to calculate the total fees for the current billing cycle. Fees are automatically deducted from your account. You can view the billing details on the Expenses and Costs page.
For a function that receives more than zero invocations or continuously occupies compute resources within an hour, if the hourly CU consumption converts to less than CNY 0.01, you are charged CNY 0.01. If the actual cost exceeds this amount, you are charged based on the actual CU consumption.
References
-
To use the same amount of resources at a lower price, see Resource plans.
-
If you stopped using Function Compute but bills are still generated, see Why am I still billed after I stop using Function Compute?.
-
If you no longer want to use Function Compute, see How do I unsubscribe from the service and pay overdue fees when my account has overdue payments?.