Managed models billing

Updated at:

Lists pay-as-you-go unit prices for Hologres managed models, covering token-based, image-based, and video/audio-based models across supported regions.

Token-based model billing

Cache billing rules

For models that are billed by token, cached input tokens are charged at a fixed ratio of the input unit price. The following table describes the ratios.

Billing item

Ratio to input unit price

Description

Explicit cache write

Input unit price × 1.25

Charged when content is written to the cache for the first time. The price is 25% higher than that of a regular input token.

Explicit cache hit

Input unit price × 0.10

Charged when a request hits an existing explicit cache. The price is one tenth of that of a regular input token.

Implicit cache hit

Input unit price × 0.20

Charged when a request hits an implicit cache. The price is one fifth of that of a regular input token.

For example, the input unit price of Qwen3.7-Max in the Beijing region is CNY 0.01440 per 1,000 tokens. The cache prices are derived as follows:

  • Explicit cache write price = 0.01440 × 1.25 = CNY 0.01800 per 1,000 tokens

  • Explicit cache hit price = 0.01440 × 0.10 = CNY 0.00144 per 1,000 tokens

  • Implicit cache hit price = 0.01440 × 0.20 = CNY 0.00288 per 1,000 tokens

Region

Model name

Model category

Input token tier

Input unit price (CNY per 1,000 tokens)

Output unit price (CNY per 1,000 tokens)

  • Beijing

  • Shanghai

  • Hangzhou

  • Shenzhen

Qwen3.8-Max

Text generation (non-thinking + thinking)

-

0.01440

0.04320

Qwen3.7-Max

Text generation (non-thinking + thinking)

-

0.01440

0.04320

Qwen3.7-Plus

Text generation (non-thinking + thinking)

≤256K

0.00240

0.00960

256K~1M

0.00720

0.02880

Qwen3.6-Plus

Text generation (non-thinking + thinking)

≤256K

0.00240

0.01440

256K~1M

0.00960

0.05760

Qwen3.6-Flash

Text generation (non-thinking + thinking)

≤256K

0.00144

0.00864

256K~1M

0.00576

0.03456

Qwen3-VL-Plus

Text generation (non-thinking + thinking)

≤32K

0.00120

0.01200

32K~128K

0.00180

0.01800

128K~256K

0.00360

0.03600

DeepSeek-V4-Pro

Text generation (non-thinking + thinking)

-

0.01440

0.02880

GLM-5.2

Text generation (non-thinking + thinking)

-

0.00960

0.03360

DeepSeek-V4-Flash

Text generation (non-thinking + thinking)

-

0.00120

0.00240

Qwen-MT-Plus

Translation

-

0.00216

0.00648

Text-Embedding-V4

Embedding

-

0.00060

-

Qwen3-VL-Embedding

Multimodal embedding - text

-

0.00084

-

Qwen3-VL-Embedding

Multimodal embedding - image/video

-

0.00216

-

qwen3.7-text-embedding

Text embedding

1

0.00060

-

  • Singapore

Qwen3.8-Max

Text generation (non-thinking + thinking)

-

0.01799

0.05396

Qwen3.7-Max

Text generation (non-thinking + thinking)

-

0.02248

0.06745

Qwen3.7-Plus

Text generation (non-thinking + thinking)

≤256K

0.00352

0.01409

256K~1M

0.01057

0.04228

Qwen3.6-Plus

Text generation (non-thinking + thinking)

≤256K

0.00450

0.02698

256K~1M

0.01799

0.05396

Qwen3.6-Flash

Text generation (non-thinking + thinking)

≤256K

0.00225

0.01349

256K~1M

0.00899

0.03597

Qwen3-VL-Plus

Text generation (non-thinking + thinking)

≤32K

0.00176

0.01409

32K~128K

0.00264

0.02114

128K~256K

0.00528

0.04227

DeepSeek-V4-Pro

Text generation (non-thinking + thinking)

-

0.02158

0.04317

GLM-5.2

Text generation (non-thinking + thinking)

-

0.00960

0.03360

DeepSeek-V4-Flash

Text generation (non-thinking + thinking)

-

0.00180

0.00360

Qwen-MT-Plus

Translation

-

0.02167

0.06491

Text-Embedding-V4

Embedding

-

0.00062

-

Image-based model billing

Region

Model name

Model category

Unit price (CNY per image)

  • Beijing

  • Shanghai

  • Hangzhou

  • Shenzhen

Qwen-Image-2.0-Pro

Image generation/editing

0.6000

Wan2.7-Image-Pro

0.6000

  • Singapore

Qwen-Image-2.0-Pro

0.6605

Wan2.7-Image-Pro

0.6745

Video and audio model billing

Region

Model name

Billing rule

Resolution/Specification

Unit price (CNY per second)

  • Beijing

  • Shanghai

  • Hangzhou

  • Shenzhen

Fun-ASR

Billed by audio duration

-

0.000264

Wan2.7-T2V

Billed per second of output video

720P

0.7200

1080P

1.2000

Wan2.7-I2V

Billed per second of output video

720P

0.7200

1080P

1.2000

Wan2.7-R2V

Billed per second for both input and output video

720P

0.72

1080P

1.20

Wan2.7-VideoEdit

Billed per second for both input and output video

720P

0.7200

1080P

1.2000

HappyHorse-1.0-T2V

Billed per second of output video

720P

1.0800

1080P

1.9200

HappyHorse-1.0-I2V

Billed per second of output video

720P

1.0800

1080P

1.9200

HappyHorse-1.0-R2V

Billed per second of output video

720P

1.0800

1080P

1.9200

HappyHorse-1.0-VideoEdit

Billed per second for both input and output video

720P

1.0800

1080P

1.9200

HappyHorse-1.1-T2V

Billed per second of output video

720P

1.0800

1080P

1.4400

HappyHorse-1.1-I2V

Billed per second of output video

720P

1.0800

1080P

1.4400

HappyHorse-1.1-R2V

Billed per second of output video

720P

1.0800

1080P

1.4400

  • Singapore

Fun-ASR

Billed by audio duration

-

0.000264

Wan2.7-T2V

Billed per second of output video

720P

0.8807

1080P

1.3211

Wan2.7-I2V

Billed per second of output video

720P

0.8807

1080P

1.3211

Wan2.7-R2V

Billed per second for both input and output video

720P

0.8807

1080P

1.3211

Wan2.7-VideoEdit

Billed per second for both input and output video

720P

0.8807

1080P

1.3211

HappyHorse-1.0-T2V

Billed per second of output video

720P

1.2590

1080P

2.1583

HappyHorse-1.0-I2V

Billed per second of output video

720P

1.2590

1080P

2.1583

HappyHorse-1.0-R2V

Billed per second of output video

720P

1.2590

1080P

2.1583

HappyHorse-1.0-VideoEdit

Billed per second for both input and output video

720P

1.2590

1080P

2.1583

HappyHorse-1.1-T2V

Billed per second of output video

720P

1.2590

1080P

1.6187

HappyHorse-1.1-I2V

Billed per second of output video

720P

1.2590

1080P

1.6187

HappyHorse-1.1-R2V

Billed per second of output video

720P

1.2590

1080P

1.6187