Data Agent fees
DataWorks Data Agent integrates capabilities such as Data Agent, AI Assistant, and Chat BI into the full data lifecycle of integration, development, operations, governance, and analysis. The available features depend on your DataWorks edition. Before you use DataWorks Data Agent, you must activate the pay-as-you-go service. You can also activate a Credit Plan subscription as needed: the Credits quota within a seat is used to deduct the consumption of the bound account in Data Agent and Chat BI, and usage after the quota is exhausted is billed on a pay-as-you-go basis. During the subscription validity period, all members (regardless of whether a seat is bound) can use code completion an unlimited number of times free of charge.
Billing logic
The feature availability of DataWorks Data Agent is determined by your DataWorks edition. Resource consumption is billed on a pay-as-you-go basis, and you can additionally subscribe to a Credit Plan to obtain a Credits quota and code completion benefits.
Feature availability
The Data Agent capabilities supported vary by DataWorks edition:
-
Basic capabilities: Supported by DataWorks Basic Edition, Standard Edition, Professional Edition, and Enterprise Edition, including Data Agent, AI Assistant, Chat BI, Coding Assistant, and AI shortcuts, as well as personal knowledge base, scheduled tasks, extensions, quota governance, seat management, and more.
-
Enterprise-level governance capabilities: Supported only by DataWorks Standard Edition and above, including model access control, enterprise-level knowledge base, and operation audit.
For the complete feature scope of each edition, see DataWorks: Features by edition.
Pay-as-you-go (mandatory)
Before you use DataWorks Data Agent, you must activate the pay-as-you-go service. During usage, model invocations, sandbox runtime, web searches, and voice input are metered by token usage, sandbox runtime, number of web search calls, and audio duration respectively, and are converted into Credits consumption. For the unit prices and supported scope of each resource, see Pay-as-you-go.
Credit Plan subscription (optional)
In addition to pay-as-you-go, you can activate a Credit Plan subscription as needed to enjoy the following benefits:
-
Seat Credits quota: When an account bound to a seat uses Data Agent and Chat BI, the Credits quota within the seat is deducted first; consumption after the quota is exhausted is billed on a pay-as-you-go basis.
-
Free code completion: During the subscription validity period, all members can use code completion an unlimited number of times free of charge without consuming Credits, regardless of whether a seat is bound.
Credit Plan pricing
Credit Plan is an optional subscription plan of DataWorks Data Agent, available in four editions: Trial, Personal, Team, and Team Pro. The editions differ mainly in price, seat configuration, and Credits quota.
Credit Plan does not directly determine the feature availability of Data Agent. Whether related features are available is determined by your DataWorks edition.
Subscription editions and prices:
|
Comparison item |
Trial Edition |
Personal Edition |
Team Edition |
Team Pro Edition |
|
Positioning |
Feature evaluation |
Individual developers |
Enterprise teams |
Enterprise teams with intensive usage |
|
Seat price |
Free |
CNY 69/month |
CNY 149/seat/month |
CNY 1,490/seat/month |
|
Eligible accounts |
Primary account only |
Primary account only |
Primary account + RAM users |
Primary account + RAM users |
|
Minimum seats |
1 |
1 |
1 (minimum), scalable |
1 (minimum), scalable |
|
Complimentary quota |
||||
|
Monthly complimentary Credits quota |
100 Credits |
1,500 Credits |
3,500 Credits/seat |
35,000 Credits/seat |
|
Code completion |
Unlimited |
Unlimited |
Unlimited |
Unlimited |
Impact of the plan change and rights of existing customers
Starting from September 18, 2026, seats in a Credit Plan are used only to allocate and manage Credits quotas and no longer carry feature entitlements. The availability of Data Agent capabilities is determined by your DataWorks edition. For more information, see Feature details of DataWorks editions.
Impact of the change: For customers who use DataWorks Basic Edition and have subscribed to Data Agent Team Edition, after switching to the new plan, enterprise-level management and control capabilities such as the enterprise knowledge base and operation audit are no longer available. To continue using these capabilities, upgrade to DataWorks Standard Edition or higher.
Rights of existing customers: Existing customers can continue to use their original benefits until the end of the current subscription month. The new plan takes effect from the next subscription month.
Credits quota issuance and validity
-
Issuance and validity: The monthly complimentary Credits quota is issued on a subscription-month basis starting from the date of a new purchase, renewal, or upgrade. The quota of the current period takes effect on the day of the operation and expires at 23:59 on the corresponding day of the next month. After each subscription cycle ends, the unused quota is cleared and is not carried over to the next cycle; the quota of the next cycle is issued at 00:00 the following day. For example, if a new purchase, renewal, or upgrade is completed on June 10, the quota of the current period is valid from June 10 to 23:59 on July 10, and the quota of the next cycle is issued at 00:00 on July 11.
-
Reactivation after unsubscribing: Within the same quota validity period, unsubscribing and then reactivating does not grant the Credits quota again. However, if the tenant has no Credits consumption records within that validity period, the corresponding quota is still granted upon reactivation.
Code completion billing
How code completion is billed depends on whether the tenant has activated a Credit Plan subscription:
-
Without a subscription: Billed based on actual usage.
-
With a subscription: During the subscription validity period, all members of the tenant can use code completion an unlimited number of times free of charge and without consuming Credits, regardless of whether a seat is bound.
Pay-as-you-go
Before you use DataWorks Data Agent, you must activate pay-as-you-go. The system meters the actual resource consumption, such as model invocation, sandbox runtime, web search, and voice input, and displays the consumption in Credits.
For Data Agent and Chat BI consumption that meets the Credit Plan deduction conditions, the system first deducts from the valid quota within the seat. Consumption after the quota is exhausted is billed on a pay-as-you-go basis. Other features are billed according to their own rules, except for the code completion benefit.
Credits and billing
Credits are the billing unit of DataWorks Data Agent. Different resources are metered by their corresponding rules and unit prices, and are converted into Credits. The unit price is CNY 0.05/Credit.
The billing details of each resource are as follows:
|
Billable item |
Metering basis |
|
Model invocation |
Number of input and output tokens |
|
Sandbox |
Runtime, in minutes |
|
Web search |
Number of calls |
|
Voice input |
Duration of input audio, in seconds |
Model invocation
Cache discount: For requests that hit the context cache, input tokens are eligible for a discount. For details, see Generic discount.
The context cache price of qwen3.8-max has been reduced as part of this adjustment. For details, see Appendix: Context cache discount.
Qwen Max
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-max |
Non-thinking and thinking modes |
0<Token≤1M |
14.4 |
43.2 |
|
qwen3.7-max |
Non-thinking and thinking modes |
0<Token≤1M |
14.4 |
43.2 |
|
qwen3-max |
Non-thinking and thinking modes |
0<Token≤32K |
3 |
12 |
|
32K<Token≤128K |
4.8 |
19.2 |
||
|
128K<Token≤256K |
8.4 |
33.6 |
More models
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-max |
Non-thinking mode only |
No tiered pricing |
2.88 |
11.52 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-max |
Non-thinking and thinking modes |
0<Token≤1M |
17.9856 |
53.958 |
|
qwen3.7-max |
Non-thinking and thinking modes |
0<Token≤1M |
22.4832 |
67.4484 |
|
qwen3-max |
Non-thinking and thinking modes |
0<Token≤32K |
10.5684 |
52.842 |
|
32K<Token≤128K |
21.1368 |
105.6852 |
||
|
128K<Token≤256K |
26.4216 |
132.1068 |
More models
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-max |
Non-thinking mode only |
No tiered pricing |
14.0916 |
56.3652 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-max |
Non-thinking and thinking modes |
0<Token≤1M |
14.4 |
43.2 |
|
qwen3.7-max |
Non-thinking and thinking modes |
0<Token≤1M |
14.4 |
43.2 |
|
qwen3-max |
Non-thinking mode only |
0<Token≤32K |
10.7916 |
53.958 |
|
32K<Token≤128K |
21.5832 |
107.916 |
||
|
128K<Token≤256K |
26.9796 |
134.8956 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-max |
Non-thinking and thinking modes |
0<Token≤1M |
14.4 |
43.2 |
|
qwen3.7-max |
Non-thinking and thinking modes |
0<Token≤1M |
14.4 |
43.2 |
|
qwen3-max |
Non-thinking mode only |
0<Token≤32K |
3 |
12 |
|
32K<Token≤128K |
4.8 |
19.2 |
||
|
128K<Token≤256K |
8.4 |
33.6 |
Qwen Plus
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
|
Non-thinking mode |
Thinking mode(Chain of thought + response) |
|||
|
qwen3.7-plus |
0<Token≤256K |
2.4 |
9.6 |
9.6 |
|
256K<Token≤1M |
7.2 |
28.8 |
28.8 |
|
|
qwen-plus |
0<Token≤128K |
0.96 |
2.4 |
9.6 |
|
128K<Token≤256K |
2.88 |
24 |
28.8 |
|
|
256K<Token≤1M |
5.76 |
57.6 |
76.8 |
|
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
|
Non-thinking mode |
Thinking mode(Chain of thought + response) |
|||
|
qwen3.7-plus |
0<Token≤256K |
3.5976 |
14.3892 |
14.3892 |
|
256K<Token≤1M |
10.7916 |
43.1664 |
43.1664 |
|
|
qwen-plus |
0<Token≤256K |
3.5232 |
10.5684 |
35.2284 |
|
256K<Token≤1M |
10.5684 |
31.7052 |
105.6852 |
|
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
|
Non-thinking mode |
Thinking mode(Chain of thought + response) |
|||
|
qwen3.7-plus |
0<Token≤256K |
2.4 |
9.6 |
9.6 |
|
256K<Token≤1M |
7.2 |
28.8 |
28.8 |
|
|
qwen-plus |
0<Token≤128K |
3.5976 |
10.7916 |
35.9724 |
|
128K<Token≤256K |
3.5976 |
10.7916 |
35.9724 |
|
|
256K<Token≤1M |
10.7916 |
32.3748 |
107.916 |
|
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
|
Non-thinking mode |
Thinking mode(Chain of thought + response) |
|||
|
qwen3.7-plus |
0<Token≤256K |
2.4 |
9.6 |
9.6 |
|
256K<Token≤1M |
7.2 |
28.8 |
28.8 |
|
|
qwen-plus |
0<Token≤128K |
0.96 |
2.4 |
9.6 |
|
128K<Token≤256K |
2.88 |
24 |
28.8 |
|
|
256K<Token≤1M |
5.76 |
57.6 |
76.8 |
|
Qwen Turbo
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input price (per million tokens) |
Output price (per million tokens) |
|
|
Non-thinking mode |
Thinking mode(Chain of thought + response) |
|||
|
qwen-turbo |
Non-thinking and thinking modes |
0.36 |
0.72 |
3.6 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input price (per million tokens) |
Output price (per million tokens) |
|
|
Non-thinking mode |
Thinking mode(Chain of thought + response) |
|||
|
qwen-turbo |
Non-thinking and thinking modes |
0.4404 |
1.7616 |
4.404 |
Qwen Flash
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-flash |
Non-thinking and thinking modes |
0<Token≤1M |
0.96 |
3.24 |
|
qwen3.6-flash |
Non-thinking and thinking modes |
0<Token≤256K |
1.44 |
8.64 |
|
256K<Token≤1M |
5.76 |
34.56 |
||
|
qwen3.5-flash |
Non-thinking and thinking modes |
0<Token≤128K |
0.24 |
2.4 |
|
128K<Token≤256K |
0.96 |
9.6 |
||
|
256K<Token≤1M |
1.44 |
14.4 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-flash |
Non-thinking and thinking modes |
0<Token≤1M |
1.3128 |
4.1124 |
|
qwen3.6-flash |
Non-thinking and thinking modes |
0<Token≤256K |
2.24826 |
13.48956 |
|
256K<Token≤1M |
8.99304 |
35.97096 |
||
|
qwen3.5-flash |
Non-thinking and thinking modes |
0<Token≤1M |
0.8808 |
3.5232 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-flash |
Non-thinking and thinking modes |
0<Token≤1M |
0.96 |
3.24 |
|
qwen3.6-flash |
Non-thinking and thinking modes |
0<Token≤256K |
1.44 |
8.64 |
|
256K<Token≤1M |
5.76 |
34.56 |
||
|
qwen3.5-flash |
Non-thinking and thinking modes |
0<Token≤128K |
0.8988 |
3.5976 |
|
128K<Token≤256K |
0.8988 |
3.5976 |
||
|
256K<Token≤1M |
0.8988 |
3.5976 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
qwen3.8-flash |
Non-thinking and thinking modes |
0<Token≤1M |
1.3128 |
4.1124 |
|
qwen3.6-flash |
Non-thinking and thinking modes |
0<Token≤256K |
1.44 |
8.64 |
|
256K<Token≤1M |
5.76 |
34.56 |
||
|
qwen3.5-flash |
Non-thinking and thinking modes |
0<Token≤128K |
0.24 |
2.4 |
|
128K<Token≤256K |
0.96 |
9.6 |
||
|
256K<Token≤1M |
1.44 |
14.4 |
Qwen Coder
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen3-coder-plus |
0<Token≤32K |
4.8 |
19.2 |
|
32K<Token≤128K |
7.2 |
28.8 |
|
|
128K<Token≤256K |
12 |
48 |
|
|
256K<Token≤1M |
24 |
240 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen3-coder-plus |
0<Token≤32K |
8.8068 |
44.0352 |
|
32K<Token≤128K |
15.8532 |
79.2636 |
|
|
128K<Token≤256K |
26.4216 |
132.1068 |
|
|
256K<Token≤1M |
52.842 |
528.4248 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen3-coder-plus |
0<Token≤32K |
4.8 |
19.2 |
|
32K<Token≤128K |
7.2 |
28.8 |
|
|
128K<Token≤256K |
12 |
48 |
|
|
256K<Token≤1M |
24 |
240 |
Qwen OCR
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-vl-ocr |
0<Token≤1M |
0.36 |
0.6 |
|
qwen3.5-ocr |
0<Token≤1M |
0.6 |
2.4 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-vl-ocr |
0<Token≤1M |
0.6168 |
1.4088 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-vl-ocr |
0<Token≤1M |
0.36 |
0.6 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-vl-ocr |
0<Token≤1M |
0.36 |
0.6 |
Qwen Translation
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-mt-plus |
0<Token≤1M |
2.16 |
6.48 |
|
qwen-mt-turbo |
0<Token≤1M |
0.84 |
2.34 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-mt-plus |
0<Token≤1M |
21.666 |
64.908 |
|
qwen-mt-turbo |
0<Token≤1M |
1.4088 |
4.3152 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-mt-plus |
0<Token≤1M |
2.16 |
6.48 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen-mt-plus |
0<Token≤1M |
2.16 |
6.48 |
Qwen Omni
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Modality |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen3.5-omni-plus |
audio |
63.6 |
255.6 |
|
context |
8.4 |
255.6 |
|
|
image |
8.4 |
255.6 |
|
|
vision |
8.4 |
255.6 |
|
|
qwen3.8-omni-flash |
- |
0.96 |
3.24 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Modality |
Input price (per million tokens) |
Output price (per million tokens) |
|
qwen3.5-omni-plus |
audio |
98.928 |
395.688 |
|
context |
12.588 |
395.688 |
|
|
image |
12.588 |
395.688 |
|
|
vision |
12.588 |
395.688 |
|
|
qwen3.8-omni-flash |
- |
1.3128 |
4.1124 |
DeepSeek
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
deepseek-v4-pro |
14.4 |
28.8 |
|
deepseek-v4-flash |
1.2 |
2.4 |
deepseek-v4-flash-0731, deepseek-v4.1-flash, and deepseek-v4.1-flash-1017 are billed at time-based prices for off-peak hours and peak hours. Off-peak hours are 22:00-08:00 (UTC+8). All other hours are peak hours.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) |
||
|
Off-peak hours |
Peak hours |
Off-peak hours |
Peak hours |
|
|
deepseek-v4-flash-0731 |
1.8 |
3.6 |
5.4 |
10.8 |
|
deepseek-v4.1-flash |
1.2 |
2.4 |
4.8 |
9.6 |
|
deepseek-v4.1-flash-1017 |
1.2 |
2.4 |
4.8 |
9.6 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
deepseek-v4-pro |
21.5832 |
43.1664 |
|
deepseek-v4-flash |
1.7988 |
3.5976 |
deepseek-v4-flash-0731, deepseek-v4.1-flash, and deepseek-v4.1-flash-1017 are billed at time-based prices for off-peak hours and peak hours. Off-peak hours are 22:00-08:00 (UTC+8). All other hours are peak hours.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) |
||
|
Off-peak hours |
Peak hours |
Off-peak hours |
Peak hours |
|
|
deepseek-v4-flash-0731 |
1.9248 |
3.8496 |
5.7756 |
11.55 |
|
deepseek-v4.1-flash |
1.3128 |
2.6256 |
5.25 |
10.5 |
|
deepseek-v4.1-flash-1017 |
1.3128 |
2.6256 |
5.25 |
10.5 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
deepseek-v4-flash-0731, deepseek-v4.1-flash, and deepseek-v4.1-flash-1017 are billed at time-based prices for off-peak hours and peak hours. Off-peak hours are 22:00-08:00 (UTC+8). All other hours are peak hours.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) |
||
|
Off-peak hours |
Peak hours |
Off-peak hours |
Peak hours |
|
|
deepseek-v4.1-flash |
1.2 |
2.4 |
4.8 |
9.6 |
|
deepseek-v4-flash-0731 |
1.8 |
3.6 |
5.4 |
10.8 |
|
deepseek-v4.1-flash-1017 |
1.2 |
2.4 |
4.8 |
9.6 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
deepseek-v4-pro |
14.4 |
28.8 |
|
deepseek-v4-flash |
1.2 |
2.4 |
deepseek-v4-flash-0731, deepseek-v4.1-flash, and deepseek-v4.1-flash-1017 are billed at time-based prices for off-peak hours and peak hours. Off-peak hours are 22:00-08:00 (UTC+8). All other hours are peak hours.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) |
||
|
Off-peak hours |
Peak hours |
Off-peak hours |
Peak hours |
|
|
deepseek-v4.1-flash |
1.2 |
2.4 |
4.8 |
9.6 |
|
deepseek-v4-flash-0731 |
1.8 |
3.6 |
5.4 |
10.8 |
|
deepseek-v4.1-flash-1017 |
1.2 |
2.4 |
4.8 |
9.6 |
GLM
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
glm-5.2 |
9.6 |
33.6 |
|
GLM-5.3 |
9.6 |
33.6 |
|
GLM-5.3-Flash |
0.96 |
3.36 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
glm-5.2 |
9.6 |
33.6 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
glm-5.2 |
9.6 |
33.6 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
glm-5.2 |
9.6 |
33.6 |
DataWorks fine-tuned models
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
dw-qwen2.5-coder-32b-instruct |
No tiered pricing |
2.4 |
7.2 |
|
dw-qwen3-coder-30b-a3b-instruct |
0<Token≤32K |
1.8 |
7.2 |
|
32K<Token≤128K |
2.7 |
10.8 |
|
|
128K<Token≤256K |
4.5 |
18 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
dw-qwen3-coder-30b-a3b-instruct |
0<Token≤32K |
3.9636 |
19.8156 |
|
32K<Token≤128K |
6.6048 |
33.0264 |
|
|
128K<Token≤256K |
10.5684 |
52.842 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
dw-qwen3-coder-30b-a3b-instruct |
0<Token≤32K |
1.8 |
7.2 |
|
32K<Token≤128K |
2.7 |
10.8 |
|
|
128K<Token≤256K |
4.5 |
18 |
Kimi
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
kimi-k3 |
Non-thinking and thinking modes |
No tiered pricing |
24 |
120 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
kimi-k3 |
Non-thinking and thinking modes |
No tiered pricing |
26.25 |
131.2512 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
kimi-k3 |
Non-thinking and thinking modes |
No tiered pricing |
24 |
120 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per million tokens.
|
Model ID |
Mode |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) Chain of thought + response |
|
kimi-k3 |
Non-thinking and thinking modes |
No tiered pricing |
24 |
120 |
Embedding
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per million tokens.
|
Model ID |
Modality |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
text-embedding-v3 |
- |
0<Token≤1M |
0.6 |
0 |
|
text-embedding-v4 |
- |
0<Token≤1M |
0.6 |
0 |
|
qwen3-vl-embedding |
context |
0<Token≤1M |
0.84 |
0 |
|
image |
0<Token≤1M |
2.16 |
0 |
|
|
vision |
0<Token≤1M |
2.16 |
0 |
|
|
tongyi-embedding-vision-flash |
context |
0<Token≤1M |
0.18 |
0 |
|
image |
0<Token≤1M |
0.18 |
0 |
|
|
vision |
0<Token≤1M |
0.18 |
0 |
|
|
tongyi-embedding-vision-plus |
context |
0<Token≤1M |
0.6 |
0 |
|
image |
0<Token≤1M |
0.6 |
0 |
|
|
vision |
0<Token≤1M |
0.6 |
0 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per million tokens.
|
Model ID |
Modality |
Input tokens per request |
Input price (per million tokens) |
Output price (per million tokens) |
|
text-embedding-v3 |
- |
0<Token≤1M |
0.6168 |
0 |
|
text-embedding-v4 |
- |
0<Token≤1M |
0.6168 |
0 |
Sandbox
The sandbox provides a secure code execution environment for the Agent and is billed by actual runtime. The unit of measurement is minutes . Runtime shorter than one minute is billed as one minute. The unit prices in the following tables are per CU per minute. Unit prices vary by region:
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per CU per minute.
|
Region |
Unit price (CNY/CU/minute) |
|
China (Beijing) |
0.0083333 |
|
China (Hangzhou) |
0.0083333 |
|
China (Shanghai) |
0.0083333 |
|
China (Shenzhen) |
0.0083333 |
|
China (Ulanqab) |
0.00745 |
|
China (Chengdu) |
0.00683 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per CU per minute.
|
Region |
Unit price (CNY/CU/minute) |
|
China (Hong Kong) |
0.015166 |
|
Singapore |
0.015166 |
|
Malaysia (Kuala Lumpur) |
0.014213 |
|
Indonesia (Jakarta) |
0.015166 |
|
Japan (Tokyo) |
0.017374 |
|
South Korea (Seoul) |
0.012710 |
EMEA
Applicable regions: Germany (Frankfurt), UK (London).
Prices are in CNY per CU per minute.
|
Region |
Unit price (CNY/CU/minute) |
|
Germany (Frankfurt) |
0.015166 |
|
UK (London) |
0.011442 |
Americas
Applicable regions: US East 1 (Virginia), US West 1 (Silicon Valley), Mexico.
Prices are in CNY per CU per minute.
|
Region |
Unit price (CNY/CU/minute) |
|
US East 1 (Virginia) |
0.012095 |
|
US West 1 (Silicon Valley) |
0.016318 |
|
Mexico |
0.008438 |
Web search
Web search is based on the iQS service and provides the Agent with real-time internet retrieval capabilities. It is billed by the number of calls. Web search is supported on the China site but not yet on the International site.
The web search unit price is CNY 0.012/call.
Voice input
Voice input is billed by input audio duration. The unit of measurement is seconds. Output is not billed. Currently, only some regions are supported:
Chinese mainland
Applicable regions: China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu).
Prices are in CNY per second.
|
Region |
Unit price (CNY/second) |
|
China (Beijing), China (Ulanqab), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu) |
0.000264 |
Asia Pacific (including China (Hong Kong))
Applicable regions: China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul).
Prices are in CNY per second.
|
Region |
Unit price (CNY/second) |
|
China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), South Korea (Seoul) |
0.000312 |
Voice input is not yet supported in the EMEA and Americas regions.
Activation
1. Go to the activation page
-
If you have not activated DataWorks: go to DataWorks Bundle Purchase.
-
If you have already activated DataWorks: go to Data Agent Quick Activation.
2. Confirm the region and DataWorks edition
Select the region you want to use and confirm the DataWorks edition activated in that region. If you have not activated DataWorks, select a DataWorks edition and purchase duration as needed.
The feature availability of Data Agent is determined by the DataWorks edition. For the capabilities supported by each edition, see DataWorks: Features by edition.
3. Activate pay-as-you-go (mandatory)
In the Data Agent area of the page, select Pay-As-You-Go. Pay-as-you-go is the base billing method of DataWorks Data Agent.
4. Activate a Credit Plan subscription as needed (optional)
To obtain the seat Credits quota and the free code completion benefit, select Subscription in the Data Agent area and complete the following configurations:
|
Parameter |
Description |
|
Credit Plan subscription |
Select the Trial, Personal, Team, or Team Pro edition as needed. |
|
Seats |
The Trial and Personal editions are fixed at 1 seat. For the Team and Team Pro editions, you can set the number of seats as needed, with a minimum of 1. |
|
Purchase duration |
Select the subscription duration of the Credit Plan. The available options are subject to the page display. |
|
Auto-renewal upon expiration |
Confirm whether to enable auto-renewal as needed, and review the renewal and deduction instructions on the page. |
The Credit Plan subscription edition is not equivalent to the DataWorks edition and does not change the feature scope supported by the current edition. If you do not need a subscription, you can skip this step and activate only pay-as-you-go.
5. Confirm the order and complete activation
Review the configuration list, purchase duration, auto-renewal settings, and fees, click Confirm Order and Pay, and complete the activation as prompted on the page. If the order incurs payable fees, you must complete the corresponding payment.
Subscription adjustment and unsubscription
You can go to the Purchased Resources page in the DataWorks console to adjust the subscription edition of your Data Agent Credit Plan, add or remove seats of the Team or Team Pro edition, or unsubscribe. The following operations apply only to Credit Plan subscriptions.
|
Operation |
Adjustment |
Effective time |
Fees and quota |
|
Upgrade the subscription edition |
Adjust the current subscription to a higher edition. From low to high, the editions are Trial, Personal, Team, and Team Pro. |
Immediate |
Pay the price difference between the old and new subscription editions based on the remaining time of the current subscription cycle. |
|
Add seats |
Increase the number of seats of the Team or Team Pro edition |
Immediate |
Fees for the additional seats are prorated based on the remaining time of the current subscription cycle. |
|
Downgrade the subscription edition |
Adjust the current subscription to a lower edition |
Effective in the next subscription cycle |
The quota of the original edition remains available within the current subscription cycle. The subscription fee of the current cycle is not refunded. |
|
Remove seats |
Decrease the number of seats of the Team or Team Pro edition |
Effective in the next subscription cycle |
The original number of seats is retained within the current subscription cycle. The seat fee of the current cycle is not refunded. |
|
Unsubscribe |
Cancel a Personal, Team, or Team Pro edition subscription |
Immediate |
Currently, the corresponding subscription fee is refunded based on the remaining time of the current subscription cycle, rather than calculated based on the remaining Credits quota. This will be adjusted later to a refund calculated based on the Credits already used. |
-
The subscription cycle is based on the start and end times agreed in the order, which is not equivalent to a calendar month.
-
The pay-as-you-go service currently does not support unsubscription. If you no longer use DataWorks Data Agent, you can set the tenant quota to 0 in Quota Management. This operation is not equivalent to unsubscribing from a Credit Plan; to stop the subscription, you must still unsubscribe separately.
Appendix: Context cache discount
Context caching has two modes: explicit cache and implicit cache. Input tokens that hit the cache are eligible for a discount. The discount ratios vary by model, as described below.
Generic discount
The generic discount ratios apply to all models except qwen3.8-max:
|
Item |
Explicit cache |
Implicit cache |
|
Token billing for cache creation |
125% of input token unit price |
100% of input token unit price |
|
Input token billing for cache hits |
10% of input token unit price |
20% of input token unit price |
qwen3.8-max specific discount
The cache discount for qwen3.8-max does not follow the generic ratios above. The discount for input tokens that hit the cache is lower than the generic discount, as shown in the following table.
|
Item |
Explicit cache |
Implicit cache |
|
Token billing for cache creation |
125% of input token unit price |
100% of input token unit price |
|
Input token billing for cache hits |
8.33% of input token unit price; 8.5% in the Asia Pacific (including China (Hong Kong)) region |
12.5% of input token unit price |