tencent cloud

Model pricing

Download
Mode fokus
Ukuran font
Terakhir diperbarui: 2026-08-28 18:41:55
Diterjemahkan oleh AI
Attention:
1. The DeepSeek-V4 GA [Vendor Direct] model has been officially launched. Meanwhile, to allocate resources more effectively and improve service stability, we are adjusting the model pricing policy in sync with the vendor by introducing a peak/off-peak pricing mechanism. The off-peak price is half of the peak price. Peak hours are 9:00–12:00 and 14:00–18:00 Beijing Time, with all other hours being off-peak. The new prices will take effect at 00:00 Beijing Time on August 17, 2026.
2. Peak and Off-Peak Period Determination Rules: The billing period for a single request is determined by the time (Beijing Time) when the platform server receives that request. If the request reception time falls within peak hours, that request is billed at the peak rate. If it falls within off-peak hours, it is billed at the off-peak rate. The request processing duration and return time do not affect the period determination.
3. GLM-5.3-Flash Limited-Time Discount Notice: From the date of this notice until 23:59:59 (Beijing Time) on September 10, 2026, GLM-5.3-Flash model call fees will be settled at 50% of the list price. The discount applies only to usage actually incurred during the above period. After the period ends, the model fees will be settled at the list price again.

Language models

Singapore
Guangzhou
Silicon Valley
Model Name
Condition
(token)
Peak/Off-Peak Billing
Input Tokens
(USD / million tokens)
Output Tokens
(USD / million tokens)
Cache
(USD / million tokens)
Hy4 preview
-
-
0.834
2.501
0.042
Hy3
-
-
0.132
0.528
0.033
DeepSeek-V4-Flash 0731 GA (Vendor Direct)
-
OFF-PEAK
0.22
0.66
0.007
PEAK
0.44
1.32
0.014
DeepSeek-V4-Pro 0813 GA (Vendor Direct)
-
OFF-PEAK
0.66
1.98
0.022
PEAK
1.32
3.96
0.044
DeepSeek-V4-Flash-Vision-Exp (Vendor Direct)
-
OFF-PEAK
0.22
0.66
0.007
PEAK
0.44
1.32
0.014
DeepSeek-V4-Flash 0731 GA
-
OFF-PEAK
0.22
0.66
0.007
PEAK
0.44
1.32
0.014
DeepSeek-V4-Pro 0813 GA
-
OFF-PEAK
0.66
1.98
0.022
PEAK
1.32
3.96
0.044
DeepSeek-V4-Flash
-
-
0.14
0.28
0.028
DeepSeek-V4-Pro
-
-
1.74
3.48
0.145
Deepseek-v3.2
-
-
0.57
1.71
0.114
GLM-5.3
-
-
1.4
4.4
0.26
GLM-5.3-Flash
-
-
0.15
0.50
0.03
GLM-5.2
-
-
1.4
4.4
0.26
GLM-5.1
-
-
1.4
4.4
0.26
GLM-5
-
-
1
3.2
0.2
GLM-5V-Turbo
-
-
1.2
4
0.24
GLM-5-Turbo
-
-
1.2
4
0.24
Kimi K3
-
-
3
15
0.3
Kimi K2.7 Code HighSpeed
-
-
1.9
8
0.38
Kimi K2.7 Code
-
-
0.95
4
0.19
kimi-k2.6
-
-
0.858
3.566
0.145
kimi-k2.5
-
-
0.6
3
0.1
MiniMax-M3
Input length (0, 512k]
-
0.3
1.2
0.06
Input length 512k+
-
0.6
2.4
0.12
MiniMax-M2.7
-
-
0.3
1.2
0.06
MiniMax-M2.5
-
-
0.3
1.2
0.03
Hy-MT2-Pro
-
-
0.074
0.295
-
Hy-MT2-Plus
-
-
0.074
0.295
-
Hy-MT2-Lite
-
-
0.044
0.177
-
MiMo-V2.5-Pro
-
-
0.435
0.87
0.0036
Model Name
Condition
(token)
Peak/Off-Peak Billing
Input Tokens
(USD / million tokens)
Output Tokens
(USD / million tokens)
Cache
(USD / million tokens)
Hy4 preview
-
-
0.834
2.501
0.042
Hy3
-
-
0.132
0.528
0.033
DeepSeek-V4-Flash 0731 GA (Vendor Direct)
-
OFF-PEAK
0.22
0.66
0.007
PEAK
0.44
1.32
0.014
DeepSeek-V4-Pro 0813 GA (Vendor Direct)
-
OFF-PEAK
0.66
1.98
0.022
PEAK
1.32
3.96
0.044
DeepSeek-V4-Flash-Vision-Exp (Vendor Direct)
-
OFF-PEAK
0.22
0.66
0.007
PEAK
0.44
1.32
0.014
DeepSeek-V4-Flash 0731 GA
-
OFF-PEAK
0.22
0.66
0.007
PEAK
0.44
1.32
0.014
DeepSeek-V4-Pro 0813 GA
-
OFF-PEAK
0.66
1.98
0.022
PEAK
1.32
3.96
0.044
DeepSeek-V4-Flash
-
-
0.14
0.28
0.028
DeepSeek-V4-Pro
-
-
1.74
3.48
0.145
Deepseek-v3.2
-
-
0.28
0.42
0.056
GLM-5.3
-
-
1.1116
3.8907
0.2779
GLM-5.3-Flash
-
-
0.11116
0.38907
0.03196
GLM-5.2
-
-
1.12
3.92
0.28
GLM-5.1
0 ≤ Input length < 32k
-
0.84
3.36
0.182
Input length ≥ 32k
-
1.12
3.92
0.28
GLM-5
0 ≤ Input length < 32k
-
0.573
2.58
0.115
Input length ≥ 32k
-
0.86
3.154
0.172
GLM-5V-Turbo
0 ≤ Input length < 32k
-
0.7
3.08
0.168
Input length ≥ 32k
-
0.98
3.64
0.252
GLM-5-Turbo
0 ≤ Input length < 32k
-
0.7
3.08
0.168
Input length ≥ 32k
-
0.98
3.64
0.252
Kimi K3
-
-
2.731
13.653
0.2731
Kimi K2.7 Code HighSpeed
-
-
1.9
8
0.38
Kimi K2.7 Code
-
-
0.95
4
0.19
kimi-k2.6
-
-
0.858
3.566
0.145
kimi-k2.5
-
-
0.56
2.94
0.098
MiniMax-M3
Input length (0, 512k]
-
0.3
1.2
0.06
Input length 512k+
-
0.6
2.4
0.12
MiniMax-M2.7
-
-
0.3
1.2
0.06
MiniMax-M2.5
-
-
0.3
1.2
0.03
Hy-MT2-Pro
-
-
0.074
0.295
-
Hy-MT2-Plus
-
-
0.074
0.295
-
Hy-MT2-Lite
-
-
0.044
0.177
-
MiMo-V2.5-Pro
-
-
0.41
0.819
0.003
Model Name
Condition
(token)
Input Tokens
(USD / million tokens)
Output Tokens
(USD / million tokens)
Cache
(USD / million tokens)
GLM-5.2
-
1.4
4.4
0.26

Embedding Model

Model Name
Billing Item
Price (USD/Million tokens)
Kinfra-Text-Embedding-0.6b
Text input
0.07
Kinfra-Text-Embedding-4b
Text input
0.084
Kinfra-VL-Embedding-2b
Text input
0.07
Image input
0.098
Video input
0.21
Kinfra-VL-Embedding-8b
Text input
0.084
Image input
0.126
Video input
0.252
Note:
You can view the token counts for text, image, and video inputs to a multimodal embedding model in the response fields usage.prompt_tokens_details.text_tokens, usage.prompt_tokens_details.image_tokens, and usage.prompt_tokens_details.video_tokens.

Vision Models

Image Generation

The image generation model uses token-based billing. Under the postpaid mode, the settlement cycle is hourly.
Token consumption for the same model may vary under different parameter configurations. Generation costs can be estimated based on the selected model version and parameters by referring to the parameters and costs in the table. The cost is jointly determined by the Token unit price and the Token usage. The specific calculation formula is:
Reference cost = Token usage x Token unit price
For the Token usage of each model under different parameter configurations, see the table below:
Model Name
Function
Resolution
Token Unit Price
Token Usage
Estimated Reference Cost
Hy-Image-3.0
Text-to-image and reference-to-image generation
No distinction
1.6 USD/million tokens
20,000 tokens/image
USD 0.032/image
Seedream-Image-v5.0-pro
Text-to-image and reference-to-image generation
Output image ≤ 2.61 megapixels
1.6 USD/million tokens
28,125 tokens/image
The first input image is free, and each additional image costs 1,875 tokens/image
USD 0.045/image
The first input image is free, and each additional image costs USD 0.003/image
Text-to-image and reference-to-image generation
Output image > 2.61 megapixels
1.6 USD/million tokens
56,250 tokens/image
The first input image is free, and each additional image costs 1,875 tokens/image
USD 0.09/image
The first input image is free, and each additional image costs USD 0.003/image
Seedream-Image-v5.0-lite
Text-to-image and reference-to-image generation
No distinction
1.6 USD/million tokens
21,875 tokens/image
USD 0.035/image
Vidu-Image-q2
Text-to-image generation
1080P
1.6 USD/million tokens
18,750 tokens/image
USD 0.03/image
Text-to-image generation
2K
25,000 tokens/image
USD 0.04/image
Text-to-image generation
4K
31,250 tokens/image
USD 0.05/image
Reference-to-image generation (1 to 3 images)
1080P
25,000 tokens/image
USD 0.04/image
Reference-to-image generation (1 to 3 images)
2K
37,500 tokens/image
USD 0.06/image
Reference-to-image generation (1 to 3 images)
4K
62,500 tokens/image
USD 0.1/image
Reference-to-image generation (4 to 7 images)
1080P
31,250 tokens/image
USD 0.05/image
Reference-to-image generation (4 to 7 images)
2K
50,000 tokens/image
USD 0.08/image
Reference-to-image generation (4 to 7 images)
4k
93,750 tokens/image
USD 0.15/image
Billing example:
1. After successfully calling the HY-Image-3.0 model to generate one image with the text-to-image feature, the actual cost is 1 image × 20,000 tokens/image × USD 1.6/million tokens = USD 0.032.
2. After successfully calling the Vidu-Image-q2 model to generate one 4K image with the reference-to-image feature (4–7 images), the actual cost is 1 image × 93,750 tokens/image × USD 1.6/million tokens = USD 0.15.
The billing example is provided only for reference on how fees are calculated. The specific billable items, unit prices, consumption rules, and final fees are subject to the console display and the billing center invoice.

Video Generation

The video generation model uses the token-based billing method. Under the postpaid mode, the settlement cycle is hourly.
Token consumption for the same model may vary under different parameter configurations. Generation costs can be estimated by referring to the parameters and costs in the table based on the selected model version and parameters. The cost is jointly determined by the Token unit price and the Token usage. The specific calculation formula is:
Reference cost = Token usage x Token unit price
For the Token usage of each model under different parameter configurations, see the table below:
Model Name
Generation Feature
Resolution
Token Unit Price
Token Usage
Equivalent Reference Cost
Hy-Video-1.5
Text/image-to-video generation
720P
1.6 USD/million tokens
30,000 tokens/second
USD 0.048/second
MiniMax-Video-H3
Text/image-to-video and reference-to-video generation
768P
1.6 USD/million tokens
Input video: 50,000 tokens/second
Output video: 50,000 tokens/second
The first 5 input images are free, and each additional image costs 25,000 tokens/image.
Input video: USD 0.08/second
Output video: USD 0.08/second
The first 5 input images are free, and each additional image costs USD 0.04/image.
2K
Input video: 81,250 tokens/second
Output video: 81,250 tokens/second
The first 5 input images are free, and each additional image costs 25,000 tokens/image.
Input video: USD 0.13/second
Output video: USD 0.13/second
The first 5 input images are free, and each additional image costs USD 0.04/image.
Kling-Video-V3
With audio - unspecified voice
720P
1.6 USD/million tokens
78,750 tokens/second
USD 0.126/second
With audio - unspecified voice
1080P
105,000 tokens/second
USD 0.168/second
With audio - unspecified voice
4K
262,500 tokens/second
USD 0.42/second
Silent
720P
52,500 tokens/second
USD 0.084/second
Silent
1080P
70,000 tokens/second
USD 0.112/second
Silent
4K
262,500 tokens/second
USD 0.42/second
Motion control
720P
78,750 tokens/second
USD 0.126/second
Motion control
1080P
105,000 tokens/second
USD 0.168/second
Kling-Video-v3-turbo
With audio
720P
1.6 USD/million tokens
70,000 tokens/second
USD 0.112/second
With audio
1080P
87,500 tokens/second
USD 0.14/second
Kling-Video-v3-omni
Without reference video - with audio
720P
1.6 USD/million tokens
70,000 tokens/second
USD 0.112/second
Without reference video - with audio
1080P
87,500 tokens/second
USD 0.14/second
Without reference video - with audio
4K
262,500 tokens/second
USD 0.42/second
With reference video - silent
720P
78,750 tokens/second
USD 0.126/second
With reference video - silent
1080P
105,000 tokens/second
USD 0.168/second
With reference video - silent
4K
262,500 tokens/second
USD 0.42/second
Without reference video - silent
720P
52,500 tokens/second
USD 0.084/second
Without reference video - silent
1080P
70,000 tokens/second
USD 0.112/second
Without reference video - silent
4K
262,500 tokens/second
USD 0.42/second
PixVerse-Video-v6
With audio
360p
1.6 USD/million tokens
19,444.44 tokens/second
USD 0.03111/second
With audio
540p
25,000 tokens/second
USD 0.04/second
With audio
720p
33,333.33 tokens/second
USD 0.05333/second
With audio
1080p
63,888.89 tokens/second
USD 0.10222/second
Silent
360p
13,888.89 tokens/second
USD 0.02222/second
Silent
540p
19,444.44 tokens/second
USD 0.03111/second
Silent
720p
25,000 tokens/second
USD 0.04/second
Silent
1080p
50,000 tokens/second
USD 0.08/second
PixVerse-Video-c1
With audio
360p
1.6 USD/million tokens
22,222.22 tokens/second
USD 0.03556/second
With audio
540p
27,777.78 tokens/second
USD 0.04444/second
With audio
720p
36,111.11 tokens/second
USD 0.05778/second
With audio
1080p
66,666.67 tokens/second
USD 0.10667/second
Silent
360p
16,666.67 tokens/second
USD 0.02667/second
Silent
540p
22,222.22 tokens/second
USD 0.03556/second
Silent
720p
27,777.78 tokens/second
USD 0.04444/second
Silent
1080p
52,777.78 tokens/second
USD 0.08444/second
Billing example:
After successfully calling the Kling-Video-v3 model to generate a 5-second silent video at 1080P resolution, the actual cost is 5 seconds × 70,000 tokens/second × USD 1.6/million tokens = USD 0.56.
The billing example is provided only for reference on how fees are calculated. The specific billable items, unit prices, consumption rules, and final fees are subject to the console display and the billing center invoice.

3D Generation

The world model uses token-based billing. Under the postpaid mode, the settlement cycle is hourly.
Token consumption for the same model may vary under different parameter configurations. Generation costs can be estimated by referring to the parameters and costs in the table based on the selected model version and parameters. The cost is jointly determined by the Token unit price and the Token usage. The specific calculation formula is:
Reference cost = Token usage x Token unit price
For the Token usage of each model under different parameter configurations, see the table below:
Model Name
Generation Feature
Token Unit Price
Token Usage
Equivalent Reference Cost
Hy-World-2.1-panorama
Text/image-to-360° panorama generation
1.6 USD/million tokens
481,250 tokens/item
USD 0.77/unit
Hy-World-2.1-scene
Text/image-to-3D scene generation
7,687,500 tokens/item
USD 12.3/unit
Billing example:
1. After successfully calling the Hy-World-2.1-panorama model to generate one panoramic image, the actual cost is 1 image × 481,250 tokens/image × USD 1.6/million tokens = USD 0.77.
2. After successfully calling the Hy-World-2.1-scene model to generate one 3D scene, the actual cost is 1 scene × 7,687,500 tokens/scene × USD 1.6/million tokens = USD 12.3.
The billing example is provided only for reference on how fees are calculated. The specific billable items, unit prices, consumption rules, and final fees are subject to the console display and the billing center invoice.


Bantuan dan Dukungan

Apakah halaman ini membantu?

masukan