Billing Items

Billing Description

The billing items of FunctionGraph consist of requests, execution duration, and others. For details, see Table 1.

For details about FunctionGraph prices, see FunctionGraph Pricing Details.

**Table 1** FunctionGraph billing items
Billing Item	Description	Billing Mode	Formula
Price for requests	Total number of times all of your functions are invoked.	Pay-per-use	The first 1,000,000 requests every month are free of charge. Billing beyond the free quota = (Total number of requests – 1,000,000) x Unit price per 1,000,000 requests For details about the unit price of requests, see Pricing Details.
Execution duration (non-reserved instances)	The execution duration fee depends on the execution duration and memory allocated to the function. The following describes how computing resources are calculated: FunctionGraph calculates the resource consumption of a function by multiplying the selected memory by the execution duration. The execution duration is calculated from the time when your function code starts to execute to the time when it returns a response or terminates. It is rounded up to the nearest 1 ms. If the duration is less than 1 ms, you will be billed for 1 ms. For example, if function A is executed for 0.5 ms and function B is executed for 2.3 ms, the execution duration of function A is 1 ms and that of function B is 3 ms.	Pay-per-use	Compute resource consumption = Memory size (GB) × Execution duration (s) The execution duration is measured at a granularity of 1 ms and converted to seconds for resource consumption calculations. The first 400,000 GB-seconds are free of charge. Billing beyond the free quota = (Total compute resources consumed by all functions – 400,000 GB-seconds) × Unit price per GB-second For details about the unit price of execution duration, see Pricing Details.
Execution duration (reserved instances)	The execution duration fee depends on the execution duration and memory allocated to the function. The following describes how computing resources are calculated: FunctionGraph calculates the resource consumption of a function by multiplying the selected memory by the execution duration. The execution duration is calculated from the time when reserved instances are successfully created to the time when you call an API to release these instances. Reserved instances that stay alive for less than 1 minute will be billed for 1 minute. Those that stay alive for a longer time will be rounded up to the nearest 1s. For details, see Reserved Instance Billing and Billing Analysis (Scenario I).	Pay-per-use	Idle mode disabled The compute resources consumption of a function is calculated as follows: Memory size (GB) × Reserved instances' running duration (s). The first 400,000 GB-seconds are free of charge. Billing beyond the free quota = (Total compute resources consumed by all functions – 400,000 GB-seconds) × Unit price per GB-second Idle mode enabled The compute resources consumption of a function is calculated as follows: Idle resource consumption + Execution resource consumption. The idle resource consumption (not included in the 400,000 GB-second free quota) is calculated as follows: Memory size (GB) × [Running duration of reserved instances – Execution duration of reserved instances (s)] × Unit price per GB-second in idle mode The execution resource consumption (included in the 400,000 GB-second free quota) is calculated as follows: The first 400,000 GB-seconds are free of charge. Billing beyond the free quota is as follows: [Memory size of the function (GB) × Execution duration of reserved instances –- 400,000 GB-second] × Unit price per GB-second in idle mode. For details about the unit price of execution duration, see Pricing Details.
Others	Additional charges will apply if other cloud services, such as OBS and SMN, are used with FunctionGraph.	Billing modes of involved services	The billing modes of other cloud services are different. For details, see Pricing Details.

Reserved Instance Billing

You can create and release reserved instances and will be billed based on their execution duration. Reserved instances stay alive in the execution environment, eliminating the influence of cold starts on latency.

If you call an API to create a reserved instance, billing will begin as soon as the reserved instance creation is complete.
If you call another API to release a reserved instance, new requests will not be routed to the reserved instance. The reserved instance will be released within a certain period, and the billing will stop.
Figure 1 Lifecycle of a reserved instance

As shown in Figure 1, the billing lasts from T1 to T4.

Reserved instances are metered at a granularity of second. If a reserved instance runs for any fraction of a minute, you will be billed for the full minute. Otherwise, you will be billed based on the actual execution duration.

For example, if a reserved instance runs for 51s, you will be billed for 1 minute. If the reserved instance runs for 61s, you will be billed for 61s.
The unit of execution duration is GB-second, which means that when the memory size is 1 GB, the duration is charged by the second.