This page answers common billing questions for Function Compute, including overdue payments, unexpected charges, and resource plan behavior.
Stopping charges and deactivating Function Compute
How do I stop being charged or deactivate Function Compute?
Delete the functions you no longer use. For instructions, see Manage functions.
When the system detects that a function is deleted, it reclaims all running instances associated with the function. Proceed with caution when you delete a function. If you use provisioned instances, release them before deleting functions. For instructions, see Configure provisioned instances and auto scaling rules.
Why am I still charged after I deactivate Function Compute?
Function Compute bills resource usage from each clock hour at the end of the following clock hour. For example, if you invoke a function between 13:00 and 14:00 and stop the service at 14:00, the bill for that period arrives around 15:00. Any charges you see after deactivation are for usage that occurred before it.
Also check whether associated Alibaba Cloud services are generating charges. If those service instances are no longer needed, delete them in their respective consoles.
How do I settle an overdue payment?
Log in to Expenses and Costs. On the Homepage, check your Account Available Credit. A negative balance indicates that your account has an overdue payment. Click Top-up to pay the overdue amount.
Can I still use Function Compute if I have overdue payments for other Alibaba Cloud services?
Yes, as long as no pay-as-you-go fees are incurred for Function Compute. Once the first pay-as-you-go bill is issued for Function Compute, your account will have an overdue payment for it as well, and the service will be frozen. If your resource plans and vouchers cover all Function Compute usage with no outstanding bills greater than 0, Function Compute continues to work normally.
Unexpected charges
Why are fees generated when my functions are not processing requests?
The most likely cause is provisioned instances. The two instance types have different billing triggers:
| Instance type | Billing trigger | Ends when |
|---|---|---|
| On-demand | Billed only during function execution | Execution completes |
| Provisioned | Billed from the moment Function Compute starts the instance | You manually release it |
Release any provisioned instances you no longer need. For instructions, see Configure provisioned instances and auto scaling rules.
Also check whether other Alibaba Cloud services in your account are generating fees. Log on to the Expenses and Costs console to review billing details. For more information, see Expenses and CostsExpenses and CostsView bills.
Why am I charged after I purchase a free trial plan?
The free trial plan is available to first-time users. Once the trial quota is exhausted or the plan expires, billing switches to pay-as-you-go. For more information, see For the Chinese mainland regions on the Alibaba Cloud China Website (www.aliyun.com), a 50% discount for nighttime usage is also available. These two discounts can be combined.Billing overview.
After I purchased a trial plan, my account has overdue payments. Why?
Overdue payments occur when your resource usage exceeds the trial quota, or when other Alibaba Cloud services you use are not covered by resource plans.
To view the details, log on to the Expenses and Costs console. For overdue payment details, see View overdue payments. To review charges by billable item, see View bills.
Resource plans
When does a resource plan expire?
Resource plans are valid for 12 months from the purchase date. For example, a plan purchased on August 14, 2023 expires on August 14, 2024.
To view expiration dates, log on to the Function Compute console and check the Resource Plan section on the right side of the Overview page.
Annual resource plans are valid for 12 months. For example, if you purchase a plan on August 14, 2024, it expires on August 14, 2025.
Monthly resource plans are valid for 1 month. For example, if you purchase a plan on June 1, 2025, it expires on July 1, 2025.
Can I use a resource plan across regions?
Yes. A resource plan offsets resource usage in any region.
Does Function Compute support auto-renewal of resource plans?
No. Purchase a new resource plan when the existing one expires.
Why am I still charged after I purchase a resource plan?
After a resource plan is fully used up or expires, billing reverts to pay-as-you-go. To check the remaining quota on your plan, see Manage resource plans.
Why isn't my newly purchased CU resource plan being applied?
Resource plans are applied in order of expiration date, starting with the plan that expires soonest. If you have existing plans that have not expired or been fully used, those are applied first. Your new plan takes effect only after all prior plans are exhausted.
CU-based billing
Why can I only buy CU resource plans now?
Function Compute now uses a single billable item: Compute Unit (CU) usage. All resource types—including vCPU, memory, GPU, function invocations, and outbound Internet traffic—are converted to CU usage based on defined conversion factors. Accordingly, only CU resource plans are available for purchase. Legacy resource plans for GPU, vCPU, memory, function invocations, and outbound Internet traffic are no longer offered.
After purchasing a CU resource plan, all resource usage is automatically converted to CU usage and deducted from the plan. For conversion details, see CU conversion factors.
Will the new CU-based billing increase my costs?
Not in most cases. The CU billing model is designed to balance costs across resource types, so overall spending rarely increases or decreases significantly. During the promotion period, Function Compute applies a 20% discount on CU usage. To reduce costs further, use idle mode and resource plans. For the full billing overview, see Billing overview.
GPU-accelerated instances
Can I switch between CPU instances and GPU-accelerated instances?
No. Switching between CPU instances and GPU-accelerated instances is not supported.
What resource usage items apply when running GPU-accelerated instances?
The following items are billed during function executions on GPU-accelerated instances: function invocations, active vCPU usage, idle vCPU usage, memory usage, disk usage, active GPU usage, idle GPU usage, and outbound Internet traffic (if Internet access is involved). For more information, see Billing overview.
GPU-accelerated instances also require CPU resources. Available specifications are listed below.
Instance specifications
| Instance | Full GPU size | FP16 (TFLOPS) | FP32 (TFLOPS) | vGPU memory | vGPU computing power | vCPUs | Memory | Supported modes |
|---|---|---|---|---|---|---|---|---|
| fc.gpu.tesla.1 | 16 GB | 65 | 8 | 1,024–16,384 MB (1–16 GB); must be a multiple of 1,024 | Calculated as: vGPU memory (GB) ÷ 16. Auto-allocated. | 0.05 to vGPU memory (GB) ÷ 2; must be a multiple of 0.05. See GPU specifications. | 128 MB to vGPU memory (GB) × 2,048 MB; must be a multiple of 64. See GPU specifications. | On-demand, Regular provisioned, Idle provisioned |
| fc.gpu.ada.1 | 48 GB | 119 | 60 | 49,152 MB (48 GB only) | Full GPU computing power. Auto-allocated. | 8 only | 65,536 MB (64 GB only) | On-demand, Regular provisioned, Idle provisioned |
Notifications
How do I unsubscribe from Function Compute notifications?
Log on to the portal and click Alibaba Cloud official websiteConsoleAddresses and Contacts in the upper-right corner.
In the Alibaba Cloud Management Console, click the bell icon in the top navigation bar to go to the Message Center console.
In the left-side navigation pane, choose Message Settings > Common Settings.
Click Manage Contacts in the upper-right corner, then add or remove contacts as needed.
For more information, see How do I configure notifications?