Skip to main content
Billing

Model inference pricing

Model API calls are billed on a pay-as-you-go basis by default.
This document only lists standard prices. For the latest promotions, visit the Model Studio console.
Some models support context caching (explicit cache and implicit cache). Cache-hit input tokens and the tokens used to create an explicit cache are billed at unit prices different from the standard input price (for example, explicit cache creation is billed at 125% of the standard input price, and cache hits at 10%). The input prices in the tables below do not include cache prices. For cache billing rules, discount rates, and supported models, see Context Cache.

Tiered pricing rules

Some Model Studio models use tiered pricing. The unit price is determined by the total number of input tokens in a single request. All tokens in the request are billed at the unit price of the corresponding tier. In the pricing tiers, K means 1,000 and M means 1,000,000. For example, 128K equals 128,000 tokens, 256K equals 256,000 tokens, and 1M equals 1,000,000 tokens. For example, a model has two pricing tiers: 0 < tokens ≤ 32K and 32K < tokens ≤ 128K. If a request contains 100K input tokens, it falls into the second tier (32K < 100K ≤ 128K), and all tokens are billed at the unit price of the second tier.

Text generation - Qwen

Qwen-Max

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price. If the model supports context cache, only input tokens receive a discount. These two discounts cannot apply simultaneously.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)
Model IDModeInput tokens per requestInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought + answer
Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3.8-max-prime
For more details, see Prime mode
Non-Thinking and Thinking modes0<Token≤1MCNY 24CNY 72No free quota
qwen3.8-max
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤1MCNY 12CNY 361 million tokens
qwen3.8-max-0902
context caching discount
Non-Thinking and Thinking modes0<Token≤1MCNY 12CNY 361 million tokens
qwen3.7-max
Currently equivalent to qwen3.7-max-2026-05-20
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤1MCNY 12CNY 361 million tokens
qwen3.7-max-2026-06-08Non-Thinking and Thinking modes0<Token≤1MCNY 12CNY 361 million tokens
qwen3.7-max-2026-05-20Non-Thinking and Thinking modes0<Token≤1MCNY 12CNY 361 million tokens
qwen3.7-max-preview
Currently equivalent to qwen3.7-max-2026-05-17
Thinking mode only0<Token≤1MCNY 12CNY 361 million tokens
qwen3.7-max-2026-05-17Thinking mode only0<Token≤1MCNY 12CNY 361 million tokens
More models

Model ID

Deployment scope

Mode

Input tokens per request

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-max

Chinese mainland

Non-Thinking mode only

No tiered pricing

CNY 2.4

CNY 9.6

1 million tokens

Qwen-Plus

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)
Model IDInput tokens per requestInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
Non-Thinking modeThinking mode (chain of thought + answer)
qwen3.7-plus
Currently equivalent to qwen3.7-plus-2026-05-26
50% batch inference discount
context caching discount
0<Token≤256KList price CNY 2 (Limited-time 20% off)List price CNY 8 (Limited-time 20% off)List price CNY 8 (Limited-time 20% off)1 million tokens
256K<Token≤1MList price CNY 6 (Limited-time 20% off)List price CNY 24 (Limited-time 20% off)List price CNY 24 (Limited-time 20% off)
qwen3.7-plus-2026-05-26
context caching discount
0<Token≤256KCNY 2CNY 8CNY 81 million tokens
256K<Token≤1MCNY 6CNY 24CNY 24
qwen3.6-plus
Currently equivalent to qwen3.6-plus-2026-04-02
0<Token≤256KCNY 2CNY 12CNY 121 million tokens
256K<Token≤1MCNY 8CNY 48CNY 48
qwen3.6-plus-2026-04-020<Token≤256KCNY 2CNY 12CNY 121 million tokens
256K<Token≤1MCNY 8CNY 48CNY 48
qwen3.5-plus
Currently equivalent to qwen3.5-plus-2026-02-15
0<Token≤128KCNY 0.8CNY 4.8CNY 4.81 million tokens
128K<Token≤256KCNY 2CNY 12CNY 12
256K<Token≤1MCNY 4CNY 24CNY 24
qwen3.5-plus-2026-04-200<Token≤128KCNY 0.8CNY 4.8CNY 4.81 million tokens
128K<Token≤256KCNY 2CNY 12CNY 12
256K<Token≤1MCNY 4CNY 24CNY 24
qwen3.5-plus-2026-02-150<Token≤128KCNY 0.8CNY 4.8CNY 4.81 million tokens
128K<Token≤256KCNY 2CNY 12CNY 12
256K<Token≤1MCNY 4CNY 24CNY 24
qwen-plus
Currently equivalent to qwen-plus-2025-12-01
50% batch inference discount
0<Token≤128KCNY 0.8CNY 2CNY 81 million tokens
128K<Token≤256KCNY 2.4CNY 20CNY 24
256K<Token≤1MCNY 4.8CNY 48CNY 64
qwen-plus-latest
50% batch inference discount
0<Token≤128KCNY 0.8CNY 2CNY 81 million tokens
128K<Token≤256KCNY 2.4CNY 20CNY 24
256K<Token≤1MCNY 4.8CNY 48CNY 64
qwen-plus-2025-12-010<Token≤128KCNY 0.8CNY 2CNY 81 million tokens
128K<Token≤256KCNY 2.4CNY 20CNY 24
256K<Token≤1MCNY 4.8CNY 48CNY 64
qwen-plus-2025-07-14No tiered pricingCNY 0.8CNY 2CNY 81 million tokens
qwen-plus-2025-04-28No tiered pricingCNY 0.8CNY 2CNY 81 million tokens
More models

Model ID

Deployment scope

Input tokens per request

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-plus-2025-01-25

Chinese mainland

No tiered pricing

CNY 0.8

CNY 2

1 million tokens

qwen-plus-2025-01-12

Chinese mainland

No tiered pricing

CNY 0.8

CNY 2

1 million tokens

qwen-plus-2024-12-20

Chinese mainland

No tiered pricing

CNY 0.8

CNY 2

1 million tokens

Qwen-Flash

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price. If the model supports context cache, only input tokens receive a discount. These two discounts cannot apply simultaneously.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)
Model IDModeInput tokens per requestInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought + answer
Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3.8-flash
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤1MCNY 0.8CNY 2.71 million tokens
qwen3.7-flash
Currently equivalent to qwen3.7-flash-2026-07-15
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤32KCNY 0.2CNY 0.81 million tokens
32K<Token≤256KCNY 0.6CNY 2.4
256K<Token≤1MCNY 1.2CNY 4.8
qwen3.7-flash-2026-07-15Non-Thinking and Thinking modes0<Token≤32KCNY 0.2CNY 0.81 million tokens
32K<Token≤256KCNY 0.6CNY 2.4
256K<Token≤1MCNY 1.2CNY 4.8
qwen3.6-flash
Currently equivalent to qwen3.6-flash-2026-04-16
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤256KCNY 1.2CNY 7.21 million tokens
256K<Token≤1MCNY 4.8CNY 28.8
qwen3.6-flash-2026-04-16Non-Thinking and Thinking modes0<Token≤256KCNY 1.2CNY 7.21 million tokens
256K<Token≤1MCNY 4.8CNY 28.8
qwen3.5-flash
Currently equivalent to qwen3.5-flash-2026-02-23
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤128KCNY 0.2CNY 21 million tokens
128K<Token≤256KCNY 0.8CNY 8
256K<Token≤1MCNY 1.2CNY 12
qwen3.5-flash-2026-02-23Non-Thinking and Thinking modes0<Token≤128KCNY 0.2CNY 21 million tokens
128K<Token≤256KCNY 0.8CNY 8
256K<Token≤1MCNY 1.2CNY 12
qwen-flash
Currently equivalent to qwen-flash-2025-07-28
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤128KCNY 0.15CNY 1.51 million tokens
128K<Token≤256KCNY 0.6CNY 6
256K<Token≤1MCNY 1.2CNY 12
qwen-flash-2025-07-28Non-Thinking and Thinking modes0<Token≤128KCNY 0.15CNY 1.51 million tokens
128K<Token≤256KCNY 0.6CNY 6
256K<Token≤1MCNY 1.2CNY 12

Qwen-Long

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price.
  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen-long
50% batch inference discount
CNY 0.5CNY 21 million tokens

Qwen-Omni

Pricing rule: billed by input tokens and output tokens. For the token calculation rules of different modalities, see Billing and rate limits.
The Qwen3.5-Omni and Qwen3-Omni series offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
  • China (Hong Kong)
  • Japan (Tokyo)
  • Germany (Frankfurt)
  • US (Virginia)
Model IDDeployment scopeInput price (per million tokens)Cache-hit input price (per million tokens)Output price (per million tokens)
qwen3.8-omni-flashChinese mainlandCNY 0.8CNY 0.1CNY 2.7
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
Text/Image/videoAudioText
Multimodal input
Text + audio
Audio only billed
qwen3.5-omni-plus
Currently equivalent to qwen3.5-omni-plus-2026-03-15
CNY 7CNY 53CNY 40CNY 2131 million tokens
qwen3.5-omni-plus-2026-03-15CNY 7CNY 53CNY 40CNY 2131 million tokens
qwen3.5-omni-flash
Currently equivalent to qwen3.5-omni-flash-2026-03-15
CNY 2.2CNY 18CNY 13.3CNY 721 million tokens
qwen3.5-omni-flash-2026-03-15CNY 2.2CNY 18CNY 13.3CNY 721 million tokens
Model IDDeployment scopeModeInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
TextAudioImage/videoText
Text-only input
Text
Multimodal input
Text + audio
Audio only billed
qwen3-omni-flashChinese mainlandNon-Thinking and Thinking modesCNY 1.8CNY 15.8CNY 3.3CNY 6.9CNY 12.7CNY 62.61 million tokens

Qwen-Omni-Realtime

Pricing rule: billed by input tokens and output tokens. For the token calculation rules of different modalities, see Billing and rate limits. For speech output, qwen3.8-omni-flash-realtime bills both the audio and its corresponding text at their respective output rates. Qwen3.5-Omni-Realtime models bill only the audio; the corresponding text is not charged.
The free quotas listed in the tables apply only to China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Prices in CNY per million tokens.

Model ID

Input: text/images/video

Input: audio

Output: text

Output: audio

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen3.8-omni-flash-realtime

1.5

6

4.5

12

1 million tokens

Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
Text/imageAudioText
Multimodal input
Text + audio
Audio only billed
qwen3.5-omni-plus-realtime
Currently equivalent to qwen3.5-omni-plus-realtime-2026-03-15
CNY 10CNY 80CNY 60CNY 3001 million tokens
qwen3.5-omni-plus-realtime-2026-03-15CNY 10CNY 80CNY 60CNY 3001 million tokens
qwen3.5-omni-flash-realtime
Currently equivalent to qwen3.5-omni-flash-realtime-2026-03-15
CNY 3.3CNY 27CNY 20CNY 1071 million tokens
qwen3.5-omni-flash-realtime-2026-03-15CNY 3.3CNY 27CNY 20CNY 1071 million tokens
Model IDDeployment scopeInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
TextAudioImageText
Text-only input
Text
Multimodal input
Text + audio
Audio only billed
qwen3-omni-flash-realtime-2025-09-15Chinese mainlandCNY 2.2CNY 18.9CNY 3.9CNY 8.3CNY 15.2CNY 75.11 million tokens

Qwen-VL

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
Model IDModeInput tokens per requestInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought + answer
Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3-vl-plus
50% batch inference discount
context caching discount
Non-Thinking and Thinking modes0<Token≤32KCNY 1CNY 101 million tokens
32K<Token≤128KCNY 1.5CNY 15
128K<Token≤256KCNY 3CNY 30

Qwen-OCR

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3.5-ocrCNY 0.5CNY 2
qwen-vl-ocr
Currently equivalent to qwen-vl-ocr-2025-11-20
50% batch inference discount
CNY 0.3CNY 0.51 million tokens
qwen-vl-ocr-2025-11-20CNY 0.3CNY 0.51 million tokens
qwen-vl-ocr-2024-10-28CNY 5CNY 51 million tokens

Qwen-Audio

You are charged for input tokens and output tokens. Audio token calculation: each second of audio corresponds to 25 tokens. If the audio duration is less than 1 second, it is counted as 25 tokens.
  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen-audio-turboCurrently available for free trial only.
After the free quota is exhausted, the model cannot be called. We recommend using Omni (Qwen-Omni) as an alternative model
100,000 tokens
qwen-audio-turbo-latest

Qwen-Coder

You are charged for input tokens and output tokens. If the model supports context cache, only input tokens receive a discount.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
Model IDInput tokens per requestInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3-coder-flash
Currently equivalent to qwen3-coder-flash-2025-07-28
0<Token≤32KCNY 1CNY 41 million tokens
32K<Token≤128KCNY 1.5CNY 6
128K<Token≤256KCNY 2.5CNY 10
256K<Token≤1MCNY 5CNY 25
qwen3-coder-flash-2025-07-280<Token≤32KCNY 1CNY 41 million tokens
32K<Token≤128KCNY 1.5CNY 6
128K<Token≤256KCNY 2.5CNY 10
256K<Token≤1MCNY 5CNY 25

Qwen-MT-Uni

Billed by input_tokens. Prices vary by input modality (text / document / image / audio) and are identical for synchronous and asynchronous calls.
  • China (Beijing)

Model ID

Input price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Text

Document

Image

Audio

qwen-mt-uni

CNY 65

CNY 20

CNY 32

CNY 400

1 million tokens

Qwen Translation

You are charged for input tokens and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-mt-plus

CNY 1.8

CNY 5.4

1 million tokens

qwen-mt-flash

CNY 0.7

CNY 1.95

1 million tokens

qwen-mt-lite

CNY 0.6

CNY 1.6

1 million tokens

Qwen Data Mining

You are charged for input tokens and output tokens.
  • China (Beijing)

Model ID

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

qwen-doc-turbo

CNY 0.6

CNY 1

No free quota

Qwen Deep Research

You are charged for input tokens and output tokens.
  • China (Beijing)

Model ID

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

qwen-deep-research

CNY 54

CNY 163

No free quota

qwen-deep-research-2025-12-15

CNY 79

CNY 236

No free quota

Tongyi Xiaomi Conversation Analysis

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price. If the model supports context cache, only input tokens receive a discount. These two discounts cannot apply simultaneously.
  • China (Beijing)

Model ID

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

tongyi-xiaomi-analysis-flash

Context Cache context caching discount

CNY 0.2

CNY 0.4

1 million tokens

tongyi-xiaomi-analysis-pro

Context Cache context caching discount

CNY 1.0

CNY 2.7

1 million tokens

Text generation - Qwen (open source)

Qwen-3.8

You are charged for input tokens and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDModeInput tokens per requestInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought + answer
Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3.8-2.4t-a95b
context caching discount
Non-Thinking and Thinking modes0<Token≤1MCNY 12CNY 361 million tokens
qwen3.8-27b
context caching discount
Non-Thinking and Thinking modes0<Token≤1MCNY 3CNY 121 million tokens

Qwen3.6

You are charged for input tokens and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Input tokens per request

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Non-Thinking mode

Thinking mode (chain of thought + answer)

qwen3.6-35b-a3b

0<Token≤256K

CNY 1.8

CNY 10.8

CNY 10.8

1 million tokens

qwen3.6-27b

0<Token≤256K

CNY 3

CNY 18

CNY 18

1 million tokens

Qwen3.5

You are charged for input tokens and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Input tokens per request

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Non-Thinking mode

Thinking mode (chain of thought + answer)

qwen3.5-397b-a17b

0<Token≤128K

CNY 1.2

CNY 7.2

CNY 7.2

1 million tokens

128K<Token≤256K

CNY 3

CNY 18

CNY 18

qwen3.5-122b-a10b

0<Token≤128K

CNY 0.8

CNY 6.4

CNY 6.4

1 million tokens

128K<Token≤256K

CNY 2

CNY 16

CNY 16

qwen3.5-27b

0<Token≤128K

CNY 0.6

CNY 4.8

CNY 4.8

1 million tokens

128K<Token≤256K

CNY 1.8

CNY 14.4

CNY 14.4

qwen3.5-35b-a3b

0<Token≤128K

CNY 0.4

CNY 3.2

CNY 3.2

1 million tokens

128K<Token≤256K

CNY 1.6

CNY 12.8

CNY 12.8

Qwen-Audio

You are charged for input tokens and output tokens.
  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen2-audio-instructCurrently available for free trial only.
After the free quota is exhausted, the model cannot be called. We recommend using Omni (Qwen-Omni) as an alternative model.
100,000 tokens
qwen-audio-chat

Text generation - third-party models

DeepSeek

You are charged for input tokens and output tokens. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
DeepSeek-V4-Flash-0731 has been adjusted to peak/off-peak pricing since 2026-08-17 00:00. The adjusted prices are shown in the table below. For more information, see DeepSeek-V4-Flash-0731 price adjustment notice.Peak hours refer to Beijing Time 8:00 - 22:00, and off-peak hours refer to Beijing Time 22:00 - 8:00 the next day.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)
  • China(Hongkong)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought + answer
Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
deepseek-v4.1-flash
context caching discount
Busy hours: CNY 2Idle hours: CNY 1Busy hours: CNY 8Idle hours: CNY 41 million tokens
deepseek-v4-pro
context caching discount
CNY 12CNY 241 million tokens
deepseek-v4-pro-0813
context caching discount
Busy hours: CNY 9Idle hours: CNY 4.5Busy hours: CNY 27Idle hours: CNY 13.51 million tokens
deepseek-v4-flash-0731
context caching discount
Busy hours: CNY 3Idle hours: CNY 1.5Busy hours: CNY 9Idle hours: CNY 4.51 million tokens
deepseek-r1-distill-qwen-1.5bLimited-time free
deepseek-r1-distill-llama-70bCurrently available for free trial only
After the free quota is exhausted, the model cannot be called. We recommend using Deep thinking, DeepSeek-Alibaba Cloud, Kimi - Alibaba Cloud as an alternative model
1 million tokens

DeepSeek-SiliconFlow

  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought + answer
Free quota
siliconflow/deepseek-v3.2CNY 2CNY 3None
siliconflow/deepseek-v3.1-terminusCNY 4CNY 12
siliconflow/deepseek-r1-0528CNY 4CNY 16
siliconflow/deepseek-v3-0324CNY 2CNY 8

DeepSeek-Kuaishou Wanqing

  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought + answer
Free quota
vanchin/deepseek-v3.2-think
context caching discount
CNY 2CNY 3None
vanchin/deepseek-v3.1-terminus
context caching discount
CNY 4CNY 12
vanchin/deepseek-r1
context caching discount
CNY 4CNY 16
vanchin/deepseek-v3
context caching discount
CNY 2CNY 8
vanchin/deepseek-ocrCNY 0.216CNY 0.216
vanchin/deepseek-v4-pro
context caching discount
CNY 12CNY 24
vanchin/deepseek-v4-pro-0813
context caching discount
CNY 9CNY 27
vanchin/deepseek-v4.1-flash
context caching discount
CNY 2CNY 8

Kimi

You are charged for input tokens and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Germany (Frankfurt)
  • Japan (Tokyo)
  • Singapore

Model ID

Mode

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

kimi-k3

Thinking modes only

CNY 20

CNY 100

1 million tokens

kimi-k2.7-code

Thinking modes only

CNY 6.5

CNY 27

1 million tokens

kimi-k2.6

Non-Thinking and Thinking modes

CNY 6.5

CNY 27

1 million tokens

kimi-k2.5

Non-Thinking and Thinking modes

CNY 4

CNY 21

1 million tokens

Kimi-Moonshot AI

You are charged for input tokens and output tokens.
  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)
kimi/kimi-k3
context caching discount
CNY 20CNY 100None
kimi/kimi-k2.7-code-highspeed
context caching discount
CNY 13CNY 54
kimi/kimi-k2.7-code
context caching discount
CNY 6.5CNY 27
kimi/kimi-k2.6
context caching discount
CNY 6.5CNY 27
kimi/kimi-k2.5
context caching discount
CNY 4CNY 21

GLM

You are charged for input tokens and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)
Model IDModeInput tokens per requestInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
glm-5.3Non-Thinking and Thinking modesflat-rate pricingCNY 8CNY 281 million tokens
glm-5.2Non-Thinking and Thinking modesflat-rate pricingCNY 8CNY 281 million tokens
glm-5.2-fast-previewNon-Thinking and Thinking modesflat-rate pricingCNY 16CNY56None
glm-5.1Non-Thinking and Thinking modes0<Token≤32KCNY 6CNY 241 million tokens
32K<Token≤200KCNY 8CNY 28
glm-5Non-Thinking and Thinking modes0<Token≤32KCNY 4CNY 181 million tokens
32K<Token≤198KCNY 6CNY 22

GLM-Zhipu AI

You are charged for input tokens and output tokens.
  • China (Beijing)
Model IDModeInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)
ZHIPU/GLM-5.3Thinking mode onlyCNY 8CNY 28None
ZHIPU/GLM-5.3-FlashThinking mode onlyCNY 0.8CNY 2.8None
ZHIPU/GLM-5.3-FlashXThinking mode onlyCNY 2CNY 7None
ZHIPU/GLM-5.2Non-Thinking and Thinking modesCNY 8CNY 28None
ZHIPU/GLM-5.1Non-Thinking and Thinking modesCNY 8CNY 28None
ZHIPU/GLM-5Non-Thinking and Thinking modesCNY 6CNY 22None

MiniMax

You are charged for input tokens and output tokens.
  • China (Beijing)
Model IDModeInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
MiniMax-M2.5Thinking mode onlyCNY 2.1CNY 8.41 million tokens

MiniMax-MiniMax

You are charged for input tokens and output tokens.
  • China (Beijing)
Model IDModeInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)
MiniMax/MiniMax-M3
context caching discount
Non-Thinking and Thinking modesCNY 4.2CNY 16.8None
MiniMax/MiniMax-M2.7
context caching discount
Thinking mode onlyCNY 2.1CNY 8.4
MiniMax/MiniMax-M2.5
context caching discount
Thinking mode onlyCNY 2.1CNY 8.4
MiniMax/MiniMax-M2.1
context caching discount
Thinking mode onlyCNY 2.1CNY 8.4

MiMo-Xiaomi

You are charged for input tokens and output tokens.
  • China (Beijing)
Model IDNumber of input tokensInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)
xiaomi/mimo-v2.5-pro0<Token≤256KCNY 7CNY 21None
256K<Token≤1MCNY 14CNY 42

Stepfun-StepFun

  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)
stepfun/step-3.7-flashCNY 1.35CNY 8.1None
stepfun/step-5-previewCNY 7CNY 20None

Unisound

  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)
unisound/unisound-u2CNY 1CNY 2None

Unisound

  • China (Beijing)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)
Chain of thought and answer
Free quota(Note)
unisound/unisound-u2CNY 1CNY 2None

Image generation

You are charged based on the number of input images and the number of successfully generated images. For models where the input image price is not specified, you are not charged for input. You are charged for output based on the number of successfully generated images. Formula: Cost = Input image unit price × Number of input images + Image unit price × Number of images generated. Notes: Failed requests incur no cost and do not consume your free quota.
Assume the output image unit price is CNY 0.10 per image. If you call the API to generate four images but only three image URLs return successfully, the system charges only for the three successfully generated images.
  • Number billed: 3 images.
  • Cost calculation: 0.1 × 3 = CNY 0.3.

Qwen Image Generation and Editing

Billed by the number of input and output images. For pricing rules, see Image generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)

Model ID

Output image resolution

Input unit price

Output unit price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-image-3.0-pro

1k

0.02 CNY/image

0.25 CNY/image

10 images

2k

0.5 CNY/image

qwen-image-3.0

1k

0.02 CNY/image

0.18 CNY/image

10 images

2k

0.18 CNY/image

Qwen Image Generation and Editing 2.0 series

The qwen-image-2.0 series is billed per image and supports both text-to-image and image-to-image/image editing. Prices by region:
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)
  • Hong Kong (China)

Model ID

Output price

Free quota

qwen-image-2.1-pro

0.25 CNY/image

10 images

qwen-image-2.1-turbo

0.1 CNY/image

10 images

qwen-image-2.0-pro

0.5 CNY/image

100 images

qwen-image-2.0-pro-2026-06-22

0.5 CNY/image

100 images

qwen-image-2.0-pro-2026-04-22

0.5 CNY/image

100 images

qwen-image-2.0-pro-2026-03-03

0.5 CNY/image

100 images

qwen-image-2.0

0.2 CNY/image

100 images

qwen-image-2.0-2026-03-03

0.2 CNY/image

100 images

Qwen Image Translation

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)
  • Singapore

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-mt-image-2.0

0.004 CNY/image

100 images

qwen-mt-image

0.003 CNY/image

100 images

Z-Image

Only output is billed. For pricing rules, see Image generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

z-image-turbo

Prompt rewriting disabled (prompt_extend=false): CNY 0.1/image

Prompt rewriting enabled (prompt_extend=true): CNY 0.2/image

100 images

Wanx Text-to-Image

Only output is billed. For pricing rules, see Image generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.6-t2i

0.20 CNY/image

50 images

wan2.5-t2i-preview

0.20 CNY/image

50 images

wan2.2-t2i-plus

0.20 CNY/image

100 images

wan2.2-t2i-flash

0.14 CNY/image

100 images

wanx2.1-t2i-plus

0.20 CNY/image

500 images

wanx2.1-t2i-turbo

0.14 CNY/image

500 images

wanx2.0-t2i-turbo

0.04 CNY/image

500 images

wanx-v1

0.16 CNY/image

500 images

Wanx Image Generation and Editing

Only output is billed. For pricing rules, see Image generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.7-image-pro

0.50 CNY/image

50 images

wan2.7-image

0.20 CNY/image

50 images

wan2.6-image

0.20 CNY/image

50 images

Wanx General Image Editing

Only output is billed. For pricing rules, see Image generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.5-i2i-preview

0.20 CNY/image

50 images

wanx2.1-imageedit

0.14 CNY/image

500 images

Wanx Doodle Painting

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wanx-sketch-to-image-lite

0.06 CNY/image

500 images

Wanx Local Repaint

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)
Model IDOutput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
wanx-x-paintingCurrently available for free trial only.
After the free quota is exhausted, the model cannot be called. We recommend referring to Image editing - Qwen or Image editing - Wan2.1 for alternatives.
500 images

Portrait Style Repaint

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wanx-style-repaint-v1

0.12 CNY/image

500 images

Human Instance Segmentation

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)
Model IDOutput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
image-instance-segmentationCurrently available for free trial only.
After the free quota is exhausted, the model cannot be called.
500 images

Image Erasing and Inpainting

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)
Model IDOutput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
image-erase-completionCurrently available for free trial only.
After the free quota is exhausted, the model cannot be called. We recommend referring to Image editing - Qwen or Image editing - Wan2.1 for alternatives.
500 images

Virtual model

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)
Model IDOutput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
wanx-virtualmodelCurrently available for free trial only.
After the free quota is exhausted, the model cannot be called. We recommend referring to Image editing - Qwen or Image editing - Wan2.1 for alternatives.
500 images each
virtualmodel-v2

Footwear model

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)
Model IDOutput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
shoemodel-v1Currently available for free trial only.
After the free quota is exhausted, the model cannot be called.
500 images

Creative Poster Generation

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)
Model IDOutput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
wanx-poster-generation-v1Currently available for free trial only.
After the free quota is exhausted, the model cannot be called. We recommend referring to Image editing - Qwen or Image editing - Wan2.1 for alternatives.
500 images

Portrait generation - FaceChain

  • facechain-facedetect: Limited-time free.
  • facechain-finetune: billed by the number of training runs. Failed requests are not billed.
  • China (Beijing)

Model ID

Unit price

Free quota(Note)

facechain-facedetect

Limited-time free

Limited-time free

facechain-finetune

2.5 CNY/request

50 requests

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Image generation - third-party models

Kling-Image-Generation

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)

Model ID

Output image resolution

Output price

Free quota(Note)

kling/kling-v3-image-generation

1K

0.2 CNY/image

No free quota

2K

0.2 CNY/image

kling/kling-v3-omni-image-generation

1K

0.2 CNY/image

2K

0.2 CNY/image

4K

0.4 CNY/image

Vidu-Image-Generation

Only output is billed. For pricing rules, see Image generation.
  • China (Beijing)

Model ID

Service deployment region

Output image resolution

Unit price

Free quota(Note)

vidu/vidu-image_reference2image

Chinese mainland

1K

CNY 0.625/image

No free quota

2K

CNY 1/image

4K

CNY 1.46875/image

vidu/vidu-image-pro_reference2image

Chinese mainland

1080P

CNY 1.6875/image

2K

CNY 3.1875/image

4K

CNY 5.125/image

vidu/vidu-image-lite_reference2image

Chinese mainland

1080P

CNY 0.3125/image

2K

CNY 0.34375/image

4K

CNY 0.40625/image

vidu/viduq3-fast_reference2image

Chinese mainland

1K

CNY 0.46875/image

2K

CNY 0.78125/image

4K

CNY 1.09375/image

vidu/viduq2-pro_reference2image

Chinese mainland

1K

CNY 0.9375/image

2K

CNY 0.9375/image

4K

CNY 1.71875/image

vidu/viduq2-fast_reference2image

Chinese mainland

1K

CNY 0.28125/image

Audio generation

Input and output tokens are billed separately.
  • China (Beijing)
Model IDInput price (CNY/million tokens)Output price (CNY/million tokens)Free quota(Note)
Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen-audio-3.1-tts-next6121 million tokens

Music generation

Pricing rule: billed by the duration (in seconds) of output audio. Input is free.
  • China (Beijing)

Model ID

Output price (per second)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

fun-music-preview

CNY 0.005

1,000 seconds

fun-music-v1

CNY 0.002

Speech synthesis (text-to-speech)

Character counting rules: For the models in this section that are billed by character count (input prices are listed per 10,000 characters), the number of characters in the input text is counted as follows:
  • Each Chinese character (including simplified Chinese characters, traditional Chinese characters, Japanese kanji, and Korean hanja) counts as 2 characters.
  • Each other character (such as an English letter, a digit, a punctuation mark, a space, a Japanese kana, or a Korean letter) counts as 1 character.
  • When SSML is used, the SSML tags themselves are not counted. Only the text content to be synthesized is counted.
Examples: 你好 is 4 characters (2+2); 中A文123 is 8 characters (2+1+2+1+1+1); 中文。 is 5 characters (2+2+1); 中 文。 is 6 characters (2+1+2+1).

Qwen-Audio-TTS

Billing rules: qwen-audio-3.1-tts-flash is billed separately for input and output tokens. Other models are billed by the number of input characters, with no charge for output.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput unit price (per million tokens)Output unit price (per million tokens)

Free quota(note)

Valid for 90 days from the date of Model Studio activation or model release, whichever is later

qwen-audio-3.1-tts-flashCNY 1.5CNY 121,000,000 tokens

Model ID

Service deployment scope

Input unit price (per 10,000 characters)

Free quota(note)

Valid for 90 days from the date of Model Studio activation or model release, whichever is later

qwen-audio-3.0-tts-plus

Mainland China

CNY 1.4

10,000 characters

qwen-audio-3.0-tts-flash

Mainland China

CNY 1

10,000 characters

Qwen-TTS

The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
  • Qwen3-TTS-Flash
  • Qwen-TTS
Pricing rule: billed by the number of input text characters. Output is free.
Model IDDeployment scopeInput price (per 10,000 characters)Output priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3-tts-flashChinese mainlandCNY 0.8Free110,000 characters

Qwen-TTS-Realtime

The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
  • Qwen3-TTS-Flash-Realtime
Pricing rule: billed by the number of input text characters. Output is free.

Model ID

Deployment scope

Input price (per 10,000 characters)

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen3-tts-flash-realtime

Chinese mainland

CNY 1

Free

2025 (after November 13, 0:00 UTC+8): 10,000 characters

Qwen-TTS Voice cloning

Pricing rule: billed by the number of new voice clones created. Failed voice creations are not billed and do not consume the free quota. Deleting a voice does not restore the free quota.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Price (per voice clone)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-voice-enrollment

CNY 0.01

1,000 voices/account

CosyVoice

The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Pricing rule: billed by the number of input text characters. Output is free.

Model ID

Input price (per 10,000 characters)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

cosyvoice-v3.5-plus

CNY 1.5

10,000 characters

cosyvoice-v3.5-flash

CNY 0.8

10,000 characters

cosyvoice-v3-plus

CNY 2

10,000 characters

cosyvoice-v3-flash

CNY 1

10,000 characters

cosyvoice-v2

CNY 2

10,000 characters

CosyVoice models currently support only the pay-as-you-go billing method. No separate usage-based plan or resource plan is available for CosyVoice models.

Sambert

Pricing rule: billed by the number of input text characters. Output is free.
  • China (Beijing)

Model ID

Input price (per 10,000 characters)

Free quota(Note)

see Model list

CNY 1

30,000 characters per Alibaba Cloud account per model per month.

MiniMax

Pricing rule: billed by the number of input text characters. Output is free. Voice cloning incurs a one-time fee, which is charged together with the speech synthesis fee the first time the cloned voice is used for synthesis.

Model name

Deployment scope

Speech synthesis price (per 10,000 characters)

Clone one voice

Free quota(Note)

MiniMax/speech-2.8-hd

Chinese mainland

CNY 3.5

CNY 9.9

(charged the first time the cloned voice is used for speech synthesis)

None

MiniMax/speech-02-hd

Chinese mainland

CNY 3.5

MiniMax/speech-2.8-turbo

Chinese mainland

CNY 2

MiniMax/speech-02-turbo

Chinese mainland

CNY 2

Speech recognition (speech-to-text) and translation (speech-to-text in a specified language)

Qwen-LiveTranslate-Flash-Realtime

Pricing rule: billed by input tokens and output tokens. For the token calculation rules of different modalities, see Billing.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
Input: audioInput: imageOutput: textOutput: audio
qwen3.8-livetranslate-flash-realtimeCNY 40CNY 3.3CNY 100CNY 1601 million tokens
qwen3.5-livetranslate-flash-realtimeCNY 40CNY 3.3CNY 100CNY 1601 million tokens
qwen3.5-livetranslate-flash-realtime-2026-05-19CNY 40CNY 3.3CNY 100CNY 1601 million tokens

Qwen-LiveTranslate-Flash

Pricing rule: billed by input tokens and output tokens. For the token calculation rules of different modalities, see Billing.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Input: audio

Input: image

Output: text

Output: audio

qwen3-livetranslate-flash

CNY 10

CNY 4

CNY 10

CNY 40

1 million tokens

qwen3-livetranslate-flash-2025-12-01

CNY 10

CNY 4

CNY 10

CNY 40

1 million tokens

Qwen-Audio-3.1-ASR-Flash-Message

Pricing rule: billed by the number of input and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-audio-3.1-asr-flash-messageCNY 6CNY 4.51,000,000 tokens

Qwen-Audio-3.x-ASR-Flash-Streaming

Pricing rules:
  • qwen-audio-3.1-asr-flash-streaming: billed separately for input and output tokens.
  • qwen-audio-3.0-asr-flash-streaming: billed by the duration (in seconds) of input audio. Output is free.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput priceOutput price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-audio-3.1-asr-flash-streaming6 CNY/million tokens4.5 CNY/million tokens1,000,000 tokens
qwen-audio-3.0-asr-flash-streaming0.00033 CNY/secondFree36,000 seconds (10 hours)

Qwen-Audio-3.x-ASR-Flash-Filetrans

Pricing rules:
  • qwen-audio-3.1-asr-flash-filetrans: billed separately for input and output tokens.
  • qwen-audio-3.0-asr-flash-filetrans: billed by the duration (in seconds) of input audio. Output is free.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput priceOutput price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-audio-3.1-asr-flash-filetrans0.8 CNY/million tokens2.7 CNY/million tokens1,000,000 tokens
qwen-audio-3.0-asr-flash-filetrans0.00022 CNY/secondFree36,000 seconds (10 hours)

Qwen-Audio-3.x-ASR-Flash

Pricing rules:
  • qwen-audio-3.1-asr-flash: billed separately for input and output tokens.
  • qwen-audio-3.0-asr-flash: billed by the duration (in seconds) of input audio. Output is free.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput priceOutput price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen-audio-3.1-asr-flash0.8 CNY/million tokens2.7 CNY/million tokens1,000,000 tokens
qwen-audio-3.0-asr-flash0.00022 CNY/secondFree36,000 seconds (10 hours)

Qwen-ASR

The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
Pricing rule: billed by the duration (in seconds) of input audio. Output is free.
Model IDInput priceOutput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3-asr-flash-filetrans0.00022 CNY/secondFree36,000 seconds (10 hours)
qwen3-asr-flash36,000 seconds (10 hours)

Qwen-ASR-Realtime

Pricing rule: billed by the duration (in seconds) of input audio. Output is free.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3-asr-flash-realtime0.00033 CNY/second36,000 seconds (10 hours)

Fun-ASR

Audio file recognition

Pricing rule: billed by the duration (in seconds) of input audio. Output is free.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Model IDInput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
fun-asr0.00022 CNY/second36,000 seconds (10 hours)
fun-asr-flash-2026-06-150.00022 CNY/second36,000 seconds (10 hours)

Real-time speech recognition

Pricing rule: billed by the duration (in seconds) of input audio. Output is free.
  • China (Beijing)
  • Singapore
Model IDInput priceFree quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
fun-asr-realtime0.00033 CNY/second36,000 seconds (10 hours)
fun-asr-realtime-2026-02-2836,000 seconds (10 hours)
fun-asr-mtl-realtime-2025-12-1036,000 seconds (10 hours)
fun-asr-flash-8k-realtime
Currently equivalent to fun-asr-flash-8k-realtime-2026-01-28
0.00022 CNY/second36,000 seconds (10 hours)
fun-asr-flash-8k-realtime-2026-01-2836,000 seconds (10 hours)

Paraformer

Audio file recognition

Pricing rule: billed by the duration (in seconds) of input audio. Output is free.
  • China (Beijing)

Model ID

Input price

Free quota(Note)

paraformer-v2

0.00008 CNY/second

36,000 seconds (10 hours)

Automatically issued on the 1st of each month at 0:00 UTC+8

Valid for 1 month

paraformer-8k-v2

Real-time speech recognition

Pricing rule: billed by the duration (in seconds) of input audio. Output is free.
  • China (Beijing)

Model ID

Input price

Free quota(Note)

paraformer-realtime-v2

0.00024 CNY/second

36,000 seconds (10 hours)

Automatically issued on the 1st of each month at 0:00 UTC+8

Valid for 1 month

paraformer-realtime-8k-v2

Voice Chat

Real-time Voice Chat

Real-time voice chat models support both text and audio input and output, billed separately by input tokens and output tokens. Audio tokens are calculated based on duration: Total tokens = Audio duration (seconds) × 12.5. Durations less than 1 second are rounded up to 1 second. In multi-turn conversations, similar to text-based LLMs, the model maintains a complete conversation context to ensure coherent dialogue. Historical conversation content is processed and billed as input for subsequent turns, so the input token count increases progressively with each turn. The specific billing rules for each content type are as follows:
  • User input audio and text: Counted as context and billed as input in each subsequent turn. Audio is billed as audio tokens, and text is billed as text tokens.
  • User-configured instructions: Billed as text tokens once per turn.
  • Model output text: Counted as context using text tokens and billed as input in each subsequent turn.
  • Model output audio: Billed as audio tokens only once at output time and not counted as context.
As the number of conversation turns increases, the accumulated context tokens grow progressively. We recommend controlling the number of turns in a single session or starting a new session at appropriate times to optimize costs.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Deployment region

Input price (per million tokens)

Output price (per million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Text

Audio

Text

Audio

qwen-audio-3.1-realtime-plus

China (mainland)

CNY 5

CNY 40

CNY 40

CNY 150

1,000,000 tokens

qwen-audio-3.0-realtime-plus

China (mainland)

CNY 5

CNY 40

CNY 40

CNY 150

1,000,000 tokens

qwen-audio-3.0-realtime-flash

China (mainland)

CNY 1.5

CNY 6

CNY 4.5

CNY 12

1,000,000 tokens

Video generation

You are not charged for input. You are charged for output based on the total duration of successfully generated videos (in seconds). Formula: Cost = Video unit price × Video duration (seconds). Notes:
  • Some models charge by output video resolution. Prices differ for resolutions such as 480P, 720P, and 1080P.
  • Some models charge by output video edition. Prices differ for editions such as Standard Edition and Professional Edition.
  • Some models charge by output video aspect ratio. Prices differ for aspect ratios such as 1:1 and 3:4.
  • Some models use a flat rate, regardless of resolution, edition, or aspect ratio.
  • Failed requests incur no cost and do not consume your free quota.

HappyHorse-Text-to-video

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)

Model ID

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

happyhorse-1.1-t2v

480P

0.45 CNY/second

10 seconds

720P

0.9 CNY/second

1080P

1.2 CNY/second

happyhorse-1.0-t2v

720P

0.9 CNY/second

10 seconds

1080P

1.6 CNY/second

HappyHorse-Image-to-video - first frame

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)

Model ID

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

happyhorse-1.1-i2v

480P

0.45 CNY/second

10 seconds

720P

0.9 CNY/second

1080P

1.2 CNY/second

happyhorse-1.0-i2v

720P

0.9 CNY/second

10 seconds

1080P

1.6 CNY/second

HappyHorse-Reference-to-video

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)

Model ID

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

happyhorse-1.1-r2v

480P

0.45 CNY/second

10 seconds

720P

0.9 CNY/second

1080P

1.2 CNY/second

happyhorse-1.0-r2v

720P

0.9 CNY/second

No free quota

1080P

1.6 CNY/second

HappyHorse-Video editing

The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)
  • Japan (Tokyo)
Pricing rule: both input and output videos are billed by video duration (seconds). Failed requests are not billed and do not consume the free quota.

Model ID

Output video resolution

Input and output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

happyhorse-1.0-video-edit

720P

0.9 CNY/second

10 seconds

1080P

1.6 CNY/second

Wan 3.0-Video Generation

Pricing rule: both input and output videos are billed by video duration (seconds). Failed requests are not billed and do not consume the free quota. Billing formula: billable duration = input video duration + output video duration.
  • The billable duration of the input video is the actual input video duration in seconds.
  • The billable duration of the output video is the duration (in seconds) of successfully generated videos.
  • The free quota is 30 seconds combined for input and output video duration.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
  • Japan (Tokyo)
  • Germany (Frankfurt)
  • US (Virginia)
  • China (Hong Kong)

Model ID

Output video resolution

Input and output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan3.0-video-prime

480P

0.45 CNY/second

30 seconds

720P

0.9 CNY/second

1080P

1.8 CNY/second

wan3.0-video

480P

List price 0.3 CNY/second (Limited-time 30% off)

30 seconds

720P

List price 0.6 CNY/second (Limited-time 30% off)

1080P

List price 1.2 CNY/second (Limited-time 30% off)

Wanx-Text-to-Video

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.7-t2v-2026-06-12

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.7-t2v-2026-04-25

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.7-t2v

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.6-t2v

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.5-t2v-preview

480P

0.3 CNY/second

50 seconds

720P

0.6 CNY/second

1080P

1 CNY/second

wan2.2-t2v-plus

480P

0.14 CNY/second

50 seconds

1080P

0.70 CNY/second

wanx2.1-t2v-turbo

480P

0.24 CNY/second

200 seconds

720P

0.24 CNY/second

wanx2.1-t2v-plus

720P

0.70 CNY/second

200 seconds

Wanx-Image-to-Video

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Output video type

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.7-i2v-2026-04-25

Audio video

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.7-i2v

Audio video

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

Wanx-Image-to-Video-First-Frame

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Output video type

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.6-i2v-flash

Audio video

audio=true

720P

0.3 CNY/second

50 seconds

1080P

0.5 CNY/second

Silent video

audio=false

720P

0.15 CNY/second

1080P

0.25 CNY/second

wan2.6-i2v

Audio video

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.5-i2v-preview

Audio video

480P

0.3 CNY/second

50 seconds

720P

0.6 CNY/second

1080P

1 CNY/second

wan2.2-i2v-flash

Silent video

480P

0.10 CNY/second

50 seconds

720P

0.20 CNY/second

1080P

0.48 CNY/second

wan2.2-i2v-plus

Silent video

480P

0.14 CNY/second

50 seconds

1080P

0.70 CNY/second

wanx2.1-i2v-turbo

Silent video

480P

0.24 CNY/second

200 seconds

720P

0.24 CNY/second

wanx2.1-i2v-plus

Silent video

720P

0.70 CNY/second

200 seconds

Wanx-Image-to-Video-First-Last-Frame

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.2-kf2v-flash

480P

0.10 CNY/second

50 seconds

720P

0.20 CNY/second

1080P

0.48 CNY/second

wanx2.1-kf2v-plus

720P

0.70 CNY/second

200 seconds

Wanx-Reference-to-Video

Pricing rule: both input and output videos are billed by video duration (seconds). Failed requests are not billed and do not consume the free quota. Billing formula: billable duration = input video duration (up to 5 seconds) + output video duration.
  • The billable duration of the input video does not exceed 5 seconds. For calculation rules, see Billing and rate limiting.
  • The billable duration of the output video is duration (in seconds) of successfully generated videos.
  • China (Beijing)
  • US (Virginia)
  • Singapore
  • Germany (Frankfurt)

Model ID

Output video type

Output video resolution

Input and output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.7-r2v-2026-06-12

Audio video

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.7-r2v

Audio video

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

wan2.6-r2v-flash

Audio video

audio=true

720P

0.3 CNY/second

50 seconds

1080P

0.5 CNY/second

Silent video

audio=false

720P

0.15 CNY/second

1080P

0.25 CNY/second

wan2.6-r2v

Audio video

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

Wanx-Video-Editing

The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
Pricing rule: both input and output videos are billed by video duration (seconds). Failed requests are not billed and do not consume the free quota.

Model ID

Output video resolution

Input and output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.7-videoedit

720P

0.6 CNY/second

50 seconds

1080P

1 CNY/second

Pricing rule: input is free. Output video is billed by video duration (seconds). Failed requests are not billed and do not consume the free quota.

Model ID

Deployment scope

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wanx2.1-vace-plus

Chinese mainland

720P

0.70 CNY/second

50 seconds

Wanx-Digital Human

  • wan2.2-s2v-detect: Input is billed while output is free. Input is billed by the number of images processed. Each input image is billed once as long as the request succeeds, regardless of the detection result.
  • wan2.2-s2v: Input is free while output is billed. Output is billed by the duration (in seconds) of successfully generated videos. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Unit price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.2-s2v-detect

Input image: CNY 0.004/image

200 images

wan2.2-s2v

Output video:

  • 480P: CNY 0.5/second

  • 720P: CNY 0.9/second

100 seconds

Wanx-Image-to-Motion

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Output video mode

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.2-animate-move

Standard modewan-std

0.4 CNY/second

50 seconds

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Professional modewan-pro

0.6 CNY/second

Wanx-Video-Face-Swap

Only output is billed. For pricing rules, see Video generation.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Output video mode

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

wan2.2-animate-mix

Standard modewan-std

0.6 CNY/second

50 seconds

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Professional modewan-pro

0.9 CNY/second

VideoRetalk

Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

videoretalk

0.08 CNY/second

1800 seconds (30 minutes)

Video Style Repaint

Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video resolution

Output price

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

video-style-transform

540P

0.2 CNY/second

600 seconds

720P

0.5 CNY/second

Video generation - third-party models

Pika-Text-to-Video
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video type

Output video resolution

Output price

Free quota(Note)

pixverse/pixverse-c1-t2v

Audio video

audio=true

360P

0.24 CNY/second

No free quota

540P

0.3 CNY/second

720P

0.39 CNY/second

1080P

0.71 CNY/second

Silent video

audio=false

360P

0.18 CNY/second

540P

0.24 CNY/second

720P

0.3 CNY/second

1080P

0.56 CNY/second

pixverse/pixverse-v6-t2v

Audio video

audio=true

360P

0.21 CNY/second

No free quota

540P

0.27 CNY/second

720P

0.36 CNY/second

1080P

0.68 CNY/second

Silent video

audio=false

360P

0.15 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.53 CNY/second

pixverse/pixverse-v5.6-t2v

Audio video

audio=true

360P

0.47 CNY/second

No free quota

540P

0.47 CNY/second

720P

0.53 CNY/second

1080P

0.7 CNY/second

Silent video

audio=false

360P

0.21 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.44 CNY/second

pixverse/pixverse-v5.6-it2v

Audio video

audio=true

360P

0.47 CNY/second

No free quota

540P

0.47 CNY/second

720P

0.53 CNY/second

1080P

0.7 CNY/second

Silent video

audio=false

360P

0.21 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.44 CNY/second

Pika-Image-to-Video-First-Frame
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video type

Output video resolution

Output price

Free quota(Note)

pixverse/pixverse-c1-it2v

Audio video

audio=true

360P

0.24 CNY/second

No free quota

540P

0.3 CNY/second

720P

0.39 CNY/second

1080P

0.71 CNY/second

Silent video

audio=false

360P

0.18 CNY/second

540P

0.24 CNY/second

720P

0.3 CNY/second

1080P

0.56 CNY/second

pixverse/pixverse-v6-it2v

Audio video

audio=true

360P

0.21 CNY/second

No free quota

540P

0.27 CNY/second

720P

0.36 CNY/second

1080P

0.68 CNY/second

Silent video

audio=false

360P

0.15 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.53 CNY/second

pixverse/pixverse-v5.6-it2v

Audio video

audio=true

360P

0.47 CNY/second

No free quota

540P

0.47 CNY/second

720P

0.53 CNY/second

1080P

0.7 CNY/second

Silent video

audio=false

360P

0.21 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.44 CNY/second

Pika-Image-to-Video-First-Last-Frame
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video type

Output video resolution

Output price

Free quota(Note)

pixverse/pixverse-c1-kf2v

Audio video

audio=true

360P

0.24 CNY/second

No free quota

540P

0.3 CNY/second

720P

0.39 CNY/second

1080P

0.71 CNY/second

Silent video

audio=false

360P

0.18 CNY/second

540P

0.24 CNY/second

720P

0.3 CNY/second

1080P

0.56 CNY/second

pixverse/pixverse-v6-kf2v

Audio video

audio=true

360P

0.21 CNY/second

No free quota

540P

0.27 CNY/second

720P

0.36 CNY/second

1080P

0.68 CNY/second

Silent video

audio=false

360P

0.15 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.53 CNY/second

pixverse/pixverse-v5.6-kf2v

Audio video

audio=true

360P

0.47 CNY/second

No free quota

540P

0.47 CNY/second

720P

0.53 CNY/second

1080P

0.7 CNY/second

Silent video

audio=false

360P

0.21 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.44 CNY/second

Pika-Reference-to-Video
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Input reference type

Output video type

Output video resolution

Output price

Free quota(Note)

pixverse/pixverse-v6-r2v-omni

With video reference

Audio video

audio=true

360P

0.42 CNY/second

No free quota

540P

0.54 CNY/second

720P

0.72 CNY/second

1080P

1.36 CNY/second

Silent video

audio=false

360P

0.3 CNY/second

540P

0.42 CNY/second

720P

0.54 CNY/second

1080P

1.06 CNY/second

Without video reference

Audio video

audio=true

360P

0.21 CNY/second

540P

0.27 CNY/second

720P

0.36 CNY/second

1080P

0.68 CNY/second

Silent video

audio=false

360P

0.15 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.53 CNY/second

Model ID

Output video type

Output video resolution

Output price

Free quota(Note)

pixverse/pixverse-v6-r2v

Audio video

audio=true

360P

0.21 CNY/second

No free quota

540P

0.27 CNY/second

720P

0.36 CNY/second

1080P

0.68 CNY/second

Silent video

audio=false

360P

0.15 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.53 CNY/second

pixverse/pixverse-c1-r2v

Audio video

audio=true

360P

0.24 CNY/second

No free quota

540P

0.3 CNY/second

720P

0.39 CNY/second

1080P

0.71 CNY/second

Silent video

audio=false

360P

0.18 CNY/second

540P

0.24 CNY/second

720P

0.3 CNY/second

1080P

0.56 CNY/second

pixverse/pixverse-v5.6-r2v

Audio video

audio=true

360P

0.47 CNY/second

No free quota

540P

0.47 CNY/second

720P

0.53 CNY/second

1080P

0.7 CNY/second

Silent video

audio=false

360P

0.21 CNY/second

540P

0.21 CNY/second

720P

0.27 CNY/second

1080P

0.44 CNY/second

PixVerse-Video-Lip-Sync
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output price

Free quota(Note)

pixverse/pixverse-lipsync

0.12 CNY/second

No free quota

PixVerse-Video-Motion-Control
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video resolution

Output price

Free quota(Note)

pixverse/pixverse-motioncontrol

360P

0.27 CNY/second

No free quota

540P

0.30 CNY/second

720P

0.36 CNY/second

PixVerse-Video-Upscale
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output price

Free quota(Note)

pixverse/pixverse-upscale

0.15 CNY/second

No free quota

Kling-Video-Generation
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video type

Output video resolution

Output price

Free quota(Note)

kling/kling-v3-turbo-video-generation

Audio video

720P

0.8 CNY/second

No free quota

1080P

1.0 CNY/second

kling/kling-v3-video-generation

Silent video

720P

0.6 CNY/second

No free quota

1080P

0.8 CNY/second

Audio video

720P

0.9 CNY/second

1080P

1.2 CNY/second

kling/kling-v3-omni-video-generation

Silent video (without reference video)

720P

0.6 CNY/second

No free quota

1080P

0.8 CNY/second

Silent video (with reference video)

720P

0.9 CNY/second

1080P

1.2 CNY/second

Audio video (without reference video)

720P

0.9 CNY/second

1080P

1.2 CNY/second

Vidu-Text-to-video
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video resolution

Output price

Free quota(Note)

vidu/viduq3-pro_text2video

540P

0.3125 CNY/second

No free quota

720P

0.78125 CNY/second

1080P

0.9375 CNY/second

vidu/viduq3-turbo_text2video

540P

0.25 CNY/second

No free quota

720P

0.375 CNY/second

1080P

0.4375 CNY/second

vidu/viduq2_text2video

540P

0.1125 CNY/second

No free quota

720P

0.21875 CNY/second

1080P

0.375 CNY/second

Vidu-Image-to-video - first frame
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video resolution

Output price

Free quota(Note)

vidu/viduq3-pro-fast_img2video

720P

0.375 CNY/second

No free quota

1080P

0.46875 CNY/second

vidu/viduq3-pro_img2video

540P

0.3125 CNY/second

No free quota

720P

0.78125 CNY/second

1080P

0.9375 CNY/second

vidu/viduq3-turbo_img2video

540P

0.25 CNY/second

No free quota

720P

0.375 CNY/second

1080P

0.4375 CNY/second

vidu/viduq2-pro_img2video

540P

0.15625 CNY/second

No free quota

720P

0.34375 CNY/second

1080P

0.71875 CNY/second

vidu/viduq2-turbo_img2video

540P

0.0875 CNY/second

No free quota

720P

0.25 CNY/second

1080P

0.46875 CNY/second

vidu/viduq2-pro-fast_img2video

720P

0.1 CNY/second

No free quota

1080P

0.2 CNY/second

Vidu-Image-to-video - first and last frames
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video resolution

Output price

Free quota(Note)

vidu/viduq3-pro_start-end2video

540P

0.3125 CNY/second

No free quota

720P

0.78125 CNY/second

1080P

0.9375 CNY/second

vidu/viduq3-turbo_start-end2video

540P

0.25 CNY/second

No free quota

720P

0.375 CNY/second

1080P

0.4375 CNY/second

vidu/viduq2-pro_start-end2video

540P

0.15625 CNY/second

No free quota

720P

0.34375 CNY/second

1080P

0.71875 CNY/second

vidu/viduq2-turbo_start-end2video

540P

0.0875 CNY/second

No free quota

720P

0.25 CNY/second

1080P

0.46875 CNY/second

Vidu-Reference-to-video
Only output is billed. For pricing rules, see Video generation.
  • China (Beijing)

Model ID

Output video resolution

Output price

Free quota(Note)

vidu/viduq3-ad_reference2video

720P

0.75 CNY/second

No free quota

1080P

0.90625 CNY/second

vidu/viduq3-drama_reference2video

720P

0.875 CNY/second

No free quota

1080P

0.875 CNY/second

vidu/viduq3-mix_reference2video

720P

0.78125 CNY/second

No free quota

1080P

0.9375 CNY/second

vidu/viduq3_reference2video

540P

0.3125 CNY/second

No free quota

720P

0.625 CNY/second

1080P

0.78125 CNY/second

vidu/viduq3-turbo_reference2video

540P

0.15625 CNY/second

No free quota

720P

0.3125 CNY/second

1080P

0.40625 CNY/second

vidu/viduq2-pro_reference2video

540P

0.25 CNY/second

No free quota

720P

0.3125 CNY/second

1080P

0.78125 CNY/second

vidu/viduq2_reference2video

540P

0.21875 CNY/second

No free quota

720P

0.28125 CNY/second

1080P

0.71875 CNY/second

MiniMax - Video generation

Billing rules: Input billing is based on the number of images and video duration; output billing is based on the duration of successfully generated videos. Specifically, billable video duration = input video duration + output video duration, and the price is determined by the output video resolution.
  • China (Beijing)

Model ID

Output video resolution

Input and output video unit price

Free quota(Note)

MiniMax/MiniMax-H3

2K

USD 0.11/second

No free quota

768P

USD 0.07/second

Input material pricing

Model ID

Input material type

Input price

MiniMax/MiniMax-H3

Audio

Free

Image

Free for up to 5 images, USD 0.03/image beyond that

Video

Price determined by output video resolution, see the table above

World models

HappyOyster-Adventure

Billing rules:
  • World creation: Billed by the number of world creations. Cost = Number of world creations × Unit price.
  • World experience: Billed by the output video duration (seconds). Cost = Video duration × Unit price. 480P only.
  • Failed requests incur no cost.
None of the following models offer a free quota in any region.
  • China (Beijing)
  • Singapore
  • US (Virginia)

Model ID

Billing item

Unit price

happyoyster-1.0-adventure

World creation

0.05 CNY/call

World experience (480P)

0.2 CNY/second

HappyOyster-Directing

Billing rules:
  • World creation: Billed by the number of world creations. Cost = Number of world creations × Unit price. Whether a reference image is used affects the unit price: world creation without and with a reference image are priced separately.
  • Instruction injection: Billed by the number of injected instructions. Cost = Number of instructions × Unit price.
  • World experience: Billed by the output video duration (seconds). Cost = Video duration × Unit price. Unit prices differ by resolution (480P/720P).
  • Failed requests incur no cost.
None of the following models offer a free quota in any region.
  • Singapore
  • US (Virginia)

Model ID

Deployment scope

Billing item

Unit price

happyoyster-1.0-directing

International

World creation (without reference image)

2.269918662 CNY/call

World creation (with reference image)

0.056146398 CNY/call

Instruction injection

0.112292796 CNY/call

World experience (480P)

0.393025 CNY/second

World experience (720P)

0.625631 CNY/second

HappyOyster-Acting

Billing rules:
  • World creation: Billed by the number of world creations. Cost = Number of world creations × Unit price.
  • Instruction injection: Billed by the number of injected instructions. Cost = Number of instructions × Unit price.
  • World experience: Billed by the output video duration (seconds). Cost = Video duration × Unit price. Unit prices differ by resolution (480P/720P).
  • Failed requests incur no cost.
None of the following models offer a free quota in any region.
  • Singapore
  • US (Virginia)

Model ID

Deployment scope

Billing item

Unit price

happyoyster-1.0-acting
(Invitational preview)

International

World creation

0.056146398 CNY/call

Instruction injection

0.056146398 CNY/call

World experience (480P)

0.224586 CNY/second

World experience (720P)

0.336878 CNY/second

3D model generation - third-party models

Tripo-3D model generation Pricing rule: input is free, output is billed per request.
  • China (Beijing)

Model ID

3DTask type

Output specification

Output price

Tripo/Tripo-H3.1

Text-to-3D

Standard + no texture

0.7 CNY/request

Standard + SD texture

1.4 CNY/request

Standard + HD texture

2.1 CNY/request

HD + no texture

2.1 CNY/request

HD + SD texture

2.8 CNY/request

HD + HD texture

3.5 CNY/request

Single-image-to-3D / multi-image-to-3D

Standard + no texture

1.4 CNY/request

Standard + SD texture

2.1 CNY/request

Standard + HD texture

2.8 CNY/request

HD + no texture

2.8 CNY/request

HD + SD texture

3.5 CNY/request

HD + HD texture

4.2 CNY/request

Tripo/Tripo-P1.0

Text-to-3D

No texture

2.1 CNY/request

SD texture

2.8 CNY/request

HD texture

3.5 CNY/request

Single-image-to-3D / multi-image-to-3D

No texture

2.8 CNY/request

SD texture

3.5 CNY/request

HD texture

4.2 CNY/request

Text embedding

You are charged only for input tokens. Output tokens are not charged. If the model supports batch calls, the unit price for both input and output tokens is 50% of the real-time inference price.
  • China (Beijing)
  • Singapore
Model IDInput price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen3.7-text-embedding
50% batch inference discount
CNY 0.51 million tokens
qwen3.7-text-embedding-flash
50% batch inference discount
CNY 0.1251 million tokens
text-embedding-v4
50% batch inference discount
CNY 0.51 million tokens
text-embedding-v3
50% batch inference discount
CNY 0.5500,000 tokens
text-embedding-v2
50% batch inference discount
CNY 0.7500,000 tokens
text-embedding-v1
50% batch inference discount
CNY 0.7500,000 tokens
text-embedding-async-v2CNY 0.720 million tokens
text-embedding-async-v1CNY 0.720 million tokens

Multimodal embedding

Pricing rule: billed by input tokens. Output is free.
  • China (Beijing)

Model ID

Input price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

Text

Image/video

qwen3-vl-embedding

CNY 0.7

CNY 1.8

1 million tokens

qwen2.5-vl-embedding

1 million tokens

tongyi-embedding-vision-plus

CNY 0.5

CNY 0.5

1 million tokens

tongyi-embedding-vision-flash

CNY 0.15

CNY 0.15

1 million tokens

multimodal-embedding-v1

CNY 0.7

CNY 0.9

1 million tokens

Text reranking

Text Reranking

Pricing rule: billed by input tokens. Output is free.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore

Model ID

Input price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

qwen3.7-text-rerank

Text input: CNY 0.5

1 million tokens

qwen3-vl-rerank

Text input: CNY 0.5

Image input: CNY 0.5

1 million tokens

qwen3-rerank

Text input: CNY 0.5

1 million tokens

gte-rerank-v2

Text input: CNY 0.8

1 million tokens

Industry models

Intent understanding

You are charged for input tokens and output tokens.
  • China (Beijing)

Model ID

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

tongyi-intent-detect-v3

CNY 0.4

CNY 1

1 million tokens

Role play

You are charged for input tokens and output tokens.
The following models offer a free quota only in China (Beijing). No free quota is available in other regions.
  • China (Beijing)
  • Singapore
  • US (Virginia)
  • Germany (Frankfurt)
  • Japan (Tokyo)
Model IDInput price (per 1 million tokens)Output price (per 1 million tokens)Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later
qwen-plus-character
Session Cache discount
CNY 0.8CNY 21 million tokens
qwen-flash-character
Session Cache discount
CNY 0.25CNY 1.51 million tokens
qwen-flash-character-2026-02-26
Session Cache discount
CNY 0.18CNY 1.51 million tokens

UI interaction

You are charged for input tokens and output tokens.
  • China (Beijing)

Model ID

Input price (per 1 million tokens)

Output price (per 1 million tokens)

Free quota(Note)

Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later

gui-plus

CNY 1.5

CNY 4.5

1 million tokens

gui-plus-2026-02-26

Decision model

You are charged for input tokens.
  • China (Beijing)
  • Singapore

Model ID

Input price (per 1 million tokens)

Free quota

decision-model-preview

Limited-time free

Token consumption and cost control

The following items explain token consumption in common scenarios and how to reduce model call costs.
  • Billing for file reads by URL: When a model reads a file through a URL, the transfer of the file itself does not consume tokens. However, after the file content is parsed, the parsed text is converted into input tokens and billed accordingly. The number of tokens consumed depends on the length of the parsed text, not the size of the original file. For large files, consider using knowledge base chunking and indexing to reduce token consumption instead of passing the full file content directly.
  • Troubleshooting unusually high Credits consumption: Common causes of an unexpectedly high Credits cost for a single request include an oversized input token count and the use of Agent mode, which adds overhead from a large system prompt, tool and function definitions, and thinking-mode content. To reduce cost, try compressing historical messages, starting a new conversation, turning off thinking mode, or switching to a lightweight model (for example, a model in the Flash series).
  • Recommended models for low-frequency calls: For low-frequency call scenarios such as heartbeat detection, use a low-cost model, such as one in the Qwen Flash series.
  • Deprecated models: Models in the Qwen2.5 series and other models marked as deprecated in this document are no longer available for calling, and their pricing information can no longer be queried. Migrate to the corresponding current-generation models.

Error codes

If a model call fails and returns an error message, see Error codes for resolution.
Token Plan
Model Playground
Fine-tuning
Model Compression
Statistics and Monitoring
Support