To improve user experience, ACK has updated the metering method and pricing solution of the cloud-native AI suite. This topic describes the updates for the metering method and pricing solution of the cloud-native AI suite.
Metering method update
The cloud-native AI suite is now billed based on the cluster size (numbers of vCores and GPUs). The new metering method took effect on February 7, 2023. The following metering rules are applied:
If the GPU acceleration component ack-ai-installer is deployed in your cluster, the system counts the number of GPUs provided by GPU-accelerated nodes in the cluster.
If other acceleration components are installed in your cluster, the system counts the number of vCores provided by all nodes in the cluster.
NoteThe system does not count the vCores provided by GPU-accelerated nodes whose GPUs are already counted.
Pricing solution update
The cloud-native AI suite now uses the pay-as-you-go billing method and tiered pricing, and bills are generated on a daily basis. The original two price tiers are increased to four. The unit price in each price tier is reduced. According to the current user demand, the free vCore quota will be adjusted from 480 vCores to 128 vCores after the metering method update. In addition, a free GPU quota will be provided to allow you to use eight GPUs free of charge. The new pricing solution will take effect on April 1, 2023.
The following tables describe the unit prices of vCPUs and GPUs:
Billable item | Usage (peak value of the day per Alibaba Cloud account) | Unit price (USD/day/vCPU) |
Number of vCores | [0,128] | Free |
(128,512] | 0.060 | |
(512,2048] | 0.045 | |
(2048,∞) | 0.030 |
Billable item | Usage (peak value of the day per Alibaba Cloud account) | Unit price (USD/day/GPU) |
Number of GPUs | [0,8] | Free |
(8,32] | 3.00 | |
(32,128] | 2.25 | |
(128,∞) | 1.50 |
Contact us
If you have any questions about the cloud-native AI suite, join the DingTalk group 33214567.