Inference mode metering and billing rules

更新时间:
复制 MD 格式

Introduction to inference mode

Inference mode tracks the time you or your users spend using the image generation service. Inference time is the duration from when an image generation request is submitted until the result is returned.

Billing for inference mode is based on the total time the image generation service is used by each user in your tenant. Each user is identified by a token. Two billing methods are available: pay-as-you-go and subscription. The pay-as-you-go method uses a monthly tiered pricing model. On the first day of each month, the total usage for all your users during the previous calendar month is calculated. This usage is measured in seconds, converted to hours, and then multiplied by the corresponding tiered unit price to determine your bill. Pay-as-you-go is the default billing method after you activate the AIGC Workshop service.

With the subscription method, you purchase resource plans in advance. Plans are available in sizes such as 1,000 hours, 3,000 hours, and 5,000 hours. These plans are valid for six months from the date of purchase. At the end of each month, the system first deducts your usage from your resource plan. If your resource plan is depleted, any remaining usage is billed using the pay-as-you-go method. Purchase a resource plan

Pay-as-you-go unit prices

Your tiered unit price is determined by your total usage within a calendar month. The pricing tiers are as follows:

Tier

Duration (hours)

Unit price

Tier 0

(0, 300]

CNY 20.50/hour

Tier 1

(300, 1,000]

CNY 17.50/hour

Tier 2

(1,000, 2,500]

CNY 15.00/hour

Tier 3

>2,500

CNY 13.50/hour

Billing formula

  • Duration (in hours) = Monthly session pattern usage (in seconds) / 3600

  • Determine the tier that corresponds to the duration,

  • Inference mode fee = Unit price for the tier × Duration

  • For example, if your usage in March is 500 hours, it falls into Tier 1. The total inference mode fee for March is calculated as follows:

500 × 17.5 = CNY 8,750

Resource plan pricing

Three plan sizes are available: 1,000 hours, 3,000 hours, and 5,000 hours. You can purchase multiple resource plans.

Resource plan contents

Resource plan specification

Resource plan price

1,000-hour inference mode resource plan

Includes 1,000 hours of deductible inference mode usage

CNY 18,600

3,000-hour inference mode resource plan

Includes 3,000 hours of deductible inference mode usage

CNY 48,000

5,000-hour inference mode resource plan

Includes 5,000 hours of deductible inference mode usage

CNY 72,000

Resource plan validity rules

A resource plan becomes active in the month of purchase and is valid for six months from the purchase date. The plan can be used to deduct usage for the entire month in which it was purchased. A plan cannot be applied to a bill if the plan expires before the bill's settlement date. Therefore, a resource plan can be used to deduct usage for a total of six calendar months.

If your resource plan does not have enough hours to cover your usage in a given month, the overage is billed using the pay-as-you-go method. The pay-as-you-go unit price is based on your total monthly usage for inference mode, before any deductions from the resource plan are applied.

For example, your total inference mode usage in April is 1,400 hours, but your resource plan only has 800 hours remaining. The resource plan covers the first 800 hours, and you must pay for the remaining 600 hours. The unit price is determined by your total usage of 1,400 hours, which falls into Tier 1 (CNY 17.50/hour). Therefore, your overage charge is 600 × 17.50 = CNY 10,500.