Product Pricing

Last Updated：2020-07-20

Billing Mode

BML products support two billing methods: pay per use (post-paid) and purchase package (prepaid).

Pay per use supported areas: North China-Beijing Package purchase supported areas: North China-Beijing

Pay per Use

BML products are billed per use. According to the selected resource package configuration and the number of running instances, the real-time billing is based on the usage time (minute-level). The specific charging rules are as follows:

Charge by minute, less than 1 minute is counted as 1 minute.
Charge by hour, i.e. hourly deducting fees and generating bills on Beijing time. The billing time is within 1 hour after the end of the current billing cycle. For example, the bill during 10:00-11:00 is generated before 12:00. The specific time depends on the system billing time.
Before using BML, you need to ensure that the account has no arrears.

Billing formula
Fee = package unit price x number of package instances x service time

Time measurement method: the use time of workspace module, training module and prediction module only includes the statistical time when the task status is "running".

Package

BML products support the purchase of resource package in the form of usage (prepaid). After purchasing the package, your BML consumption will be deducted from the package first, and the excess will be paid per use. If more than one package is purchased, the package purchased first will be deducted in priority according to the purchase order. Currently, the resource package only supports two package configuration types used by the workarea feature module: GPU Instance_P4_12 Core 32GB Memory x1 card and GPU Instance_V100_12 core 50GB Memory x1 Card.

You can purchase package directly on the "BML - > Package" page of Baidu AI Cloud management console. When purchasing, and select the area, package configuration, purchase duration and effective time according to the prompts.

Pricing

The package price of BML product pay per use is as follows:

Modules	Resource package configuration (instance)	Unit price (CNY)/minute/piece	Unit price (CNY)/hour/piece	Duration of discount
Workarea	CPU Instance_4 Core_8GB Memory	0.021	1.26	Free for the first 72 hours of each month
Workarea	GPU Instance_ Deep learning development card_6 Core 32GB Memory x1 Card	0.081	4.86	——
Workarea	GPU Instance_K40_6 Core 32GB Memory x1 Card	0.104	6.24	——
Workarea	GPU Instance_V100_12 Core 50GB Memory x1 Card	0.185	11.10	——
Workarea	GPU Instance_V100_48 Core 200GB Memory x4 Card	0.736	44.16	——
Workarea	GPU Instance_P4_12 Core 32GB Memory x1 Card	0.11	6.06	Free for the first 72 hours of each month
Workarea	GPU Instance_P4_48 Core 128GB Memory x4 Card	0.376	22.56	——
Workarea	CDS high performance storage 5G	0.00	0.00	Free
Workarea	CDS high performance storage 40G	0.0003267	0.019602	——
Workarea	CDS high performance storage 100G	0.0008167	0.049002	——
Workarea	CDS high performance storage 200G	0.0016333	0.097998	——
Workarea	CDS high performance storage 500G	0.0040833	0.244998	——
Train	CPU Instance_8 Core_32GB Memory	0.061	3.66	Free for the first 72 hours of each month
Train	GPU Instance_ Deep learning development card_6 Core 32GB Memory x1 Card	0.081	4.86	——
Train	GPU Instance_K40_6 Core 32GB Memory x1 Card	0.104	6.24	——
Train	GPU Instance_K40_24 Core 128GB Memory x4 Card	0.438	26.28	——
Train	GPU Instance_V100_12 Core 50GB Memory x1 Card	0.185	11.10	——
Train	GPU Instance_V100_48 Core 200GB Memory x4 Card	0.736	44.16	——
Train	GPU Instance_P4_12 Core 32GB Memory x1 Card	0.11	6.06	Free for the first 72 hours of each month
Train	GPU Instance_P4_48 Core 128GB Memory x4 Card	0.375	22.5	——
Prediction	CPU Instance_4 Core_8GB Memory	0.021	1.26	Free for the first 72 hours of each month
Prediction	CPU Instance_8 Core_32GB Memory	0.061	3.66	Free for the first 72 hours of each month
Prediction	GPU Instance_P4_12 Core 32GB Memory x1 Card	0.11	6.06	Free for the first 72 hours of each month
Prediction	GPU Instance_P4_48 Core 128GB Memory x4 Card	0.375	22.5	——

The supported package and price of BML product purchase consumption package are as follows:

Module	Package Configuration	Hour	Unit (CNY)/package
Workarea	GPU Instance_P4_12-core 32GB RAM x 1 Card	20	73
		50	182
		100	364
		200	728
		300	1090
		500	1818
Module	Package Configuration	Hour	Unit (CNY)/package
workarea	GPU Instance_V100_12-core 50GB RAMx1 Card	20	133
		50	333
		100	666
		200	1332
		300	1998
		500	3330

Arrears Rules

Low Balance Reminders and Arrears

Reminder of Insufficient Balance: Whether your account balance (including voucher available) is sufficient to pay the bill in 3 days is judged according to your bill amount of the latest 3 days. If not sufficient, system sends the renewal reminder. Whether your account balance (including voucher available) is sufficient to pay the bill in next day is judged according to your bill amount of the latest day. If not sufficient, system sends the renewal reminder.

Handling with the arrear payment: Whether your account balance is sufficient to pay the BML bill is hourly checked up on the Beijing time. For example, whether your account balance is sufficient to pay the bill from 10:00 to 11:00 is checked up on the 11:00. If not sufficient, system confirms the payment arrear and sends the renewal reminder. Service is stopped immediately after arrear. System sends the service suspension notice in arrear. Note: The running workspace instance will be stopped after the service is stopped, and the data under the non-mount point will be released and cannot be saved. The running training job will be forced to stop, the job will fail, and the prediction endpoint in the service will be terminated. Task configuration information in the training module and prediction module consoles will be retained by default. Please recharge and pay in time.

Product Description

Operation Guide

百度智能云

Baidu Machine Learning