DataWorks Data Agent offers core capabilities, including the Agent, a coding assistant, Chat BI, and AI shortcuts. These capabilities are deeply integrated with various DataWorks product modules to cover the entire data lifecycle—from data integration and development to operations, governance, and analysis—delivering an efficient, trustworthy, and intelligent data development experience. It offers a hybrid billing model that combines subscription with pay-as-you-go. You can choose an edition and the number of seats, and each edition includes a complimentary quota of tokens .
Effective April 15, 2026, DataWorks Data Agent will become a paid service. Data Agent covers the entire data lifecycle, including data integration, development, operations, governance, and analysis, and provides an efficient, reliable, and intelligent data development experience. The activation URL will be announced in early April.
-
Starting April 15, 2026, users who do not activate the service will lose access to Data Agent features.
-
Starting April 15, 2026, for users who do not activate the service, all their configuration data and session history from the public preview period will be deleted 15 days after the commercial launch. Please back up your data in advance.
-
Effective April 15, 2026, metering for token consumption will begin. If you activate the service before this date, tokens consumed before this date will not be deducted from your complimentary quota or be billed.
Billing overview
DataWorks Data Agent uses a billing model that combines subscription and pay-as-you-go. Your subscription includes a complimentary token quota, and usage beyond this quota is automatically billed on a pay-as-you-go basis:
-
Subscription: To use Data Agent, you must subscribe to an edition. Each edition includes a specific complimentary token quota. For details, see Editions and pricing.
-
Trial Edition: Free, for feature evaluation.
-
Personal Edition: ¥69/month, for individual developers.
-
Team Edition: ¥149/seat/month, for enterprise teams.
-
-
Pay-as-you-go: Pay-as-you-go billing is enabled by default when you subscribe to any edition.
-
Token billing: After you use up the complimentary token quota for a seat, you are billed on a pay-as-you-go basis for the tokens consumed per request. For details, see Pay-as-you-go.
-
A seat is a license that lets one Alibaba Cloud account use Data Agent features. You can bind one seat to one Alibaba Cloud account. All three editions include at least one seat. The Trial Edition and Personal Edition come with one fixed seat that can only be used by the main account. The Team Edition lets you purchase multiple seats, which can be assigned to and used by both the main account and RAM users.
After exhausting the complimentary token quota for their seats, Team Edition users can use pay-as-you-go billing or switch to an enterprise-specific custom model. Currently, using a custom model incurs no additional feature fees. Note that the coding assistant, Agent-based table creation in AI shortcuts, and data service API generation in AI shortcuts do not support enterprise-specific custom models. These features will consume tokens from your complimentary quota or be billed on a pay-as-you-go basis.
Editions and pricing
DataWorks Data Agent offers three editions for different scenarios, from individual evaluation to enterprise team collaboration. The following table compares their pricing, quotas, and features.
|
Item |
Trial Edition |
Personal Edition |
Team Edition |
|
Target audience |
Evaluators |
Individual developers |
Enterprise teams |
|
Price |
Free |
¥69/month |
¥149/seat/month |
|
Supported accounts |
main account only |
main account only |
main account and sub-accounts |
|
Minimum seats |
1 seat |
1 seat |
1 seat minimum; scalable |
|
Features |
|||
|
Coding assistant |
|
|
|
|
Intelligent agent (for Data Studio, Data Integration, etc.) |
|
|
|
|
Chat BI |
|
|
|
|
AI shortcuts |
|
|
|
|
Personal rules |
|
|
|
|
MCP server |
|
|
|
|
Team Edition exclusive features |
|||
|
Operation audit |
|
|
|
|
Seat assignment and management |
|
|
|
|
Enterprise-level rules |
|
|
|
|
Enterprise custom model integration |
|
|
(Free for a limited time) |
|
Complimentary quota (Chinese mainland) |
|||
|
Monthly complimentary token quota (Limited-time offer) |
1 million |
20 million |
50 million/seat |
|
Code completion (Limited-time offer) |
Unlimited |
Unlimited |
Unlimited |
|
Fast Apply (Limited-time offer) |
Unlimited |
Unlimited |
Unlimited |
|
Complimentary quota (international region, including China (Hong Kong)) |
|||
|
Monthly complimentary token quota (Limited-time offer) |
0.5 million |
2 million |
6 million/seat |
|
Code completion (Limited-time offer) |
Unlimited |
Unlimited |
Unlimited |
|
Fast Apply (Limited-time offer) |
Unlimited |
Unlimited |
Unlimited |
You receive the complimentary quota only on your first activation or renewal each month. If you unsubscribe and reactivate the service within the same month, you will not receive the quota again.
Pay-as-you-go
Activating any version enables the pay-as-you-go feature by default. Once you exhaust the complimentary token quota for each seat, the system bills you on a pay-as-you-go basis for the token consumption of each request.
Billing rules:pay-as-you-go fee = input token count * input unit price + output token count * output unit price
Cache discount: For requests that hit the context cache, input tokens receive a discount. For details, see Appendix: context cache discount.
Qwen-Max
Chinese mainland
All prices are in CNY per 1 million tokens.
|
Model name |
Mode |
Input tokens |
Input price |
Output price Chain of thought and answer |
|
qwen3-max |
non-thinking and thinking modes |
0 < tokens ≤ 32K |
3 |
12 |
|
32K < tokens ≤ 128K |
4.8 |
19.2 |
||
|
128K < tokens ≤ 252K |
8.4 |
33.6 |
||
|
qwen3-max-2026-01-23 |
non-thinking and thinking modes |
0 < tokens ≤ 32K |
3 |
12 |
|
32K < tokens ≤ 128K |
4.8 |
19.2 |
||
|
128K < tokens ≤ 252K |
8.4 |
33.6 |
More models
|
Model name |
Mode |
Input tokens |
Input price |
Output price |
|
qwen-max |
non-thinking mode only |
non-tiered pricing |
2.88 |
11.52 |
International (including China (Hong Kong))
All prices are in CNY per 1 million tokens.
|
Model name |
Mode |
Input tokens |
Input price |
Output price Chain of thought and answer |
|
qwen3-max |
non-thinking and thinking modes |
0 < tokens ≤ 32K |
10.5684 |
52.842 |
|
32K < tokens ≤ 128K |
21.1368 |
105.6852 |
||
|
128K < tokens ≤ 252K |
26.4216 |
132.1068 |
||
|
qwen3-max-2026-01-23 |
non-thinking and thinking modes |
0 < tokens ≤ 32K |
10.5684 |
52.842 |
|
32K < tokens ≤ 128K |
21.1368 |
105.6852 |
||
|
128K < tokens ≤ 252K |
26.4216 |
132.1068 |
More models
|
Model name |
Mode |
Input tokens |
Input price |
Output price |
|
qwen-max |
non-thinking mode only |
non-tiered pricing |
14.0916 |
56.3652 |
Qwen-plus
Chinese mainland
All prices are in CNY per 1 million tokens.
|
Model name |
Input token range |
Input price |
Output price |
|
|
Non-thinking mode |
Thinking mode (chain of thought + answer) |
|||
|
qwen3.5-plus |
0 < tokens ≤ 128K |
0.96 |
5.76 |
5.76 |
|
128K < tokens ≤ 256K |
2.40 |
14.40 |
14.40 |
|
|
256K < tokens ≤ 1M |
4.80 |
28.80 |
28.80 |
|
|
qwen-plus |
0 < tokens ≤ 128K |
0.96 |
2.40 |
9.60 |
|
128K < tokens ≤ 256K |
2.88 |
24.00 |
28.80 |
|
|
256K < tokens ≤ 1M |
5.76 |
57.60 |
76.80 |
|
International (including China (Hong Kong))
All prices are in CNY per 1 million tokens.
|
Model name |
Input token range |
Input price |
Output price |
|
|
Non-thinking mode |
Thinking mode (chain of thought + answer) |
|||
|
qwen3.5-plus |
0 < tokens ≤ 256K |
3.5232 |
21.1368 |
21.1368 |
|
256K < tokens ≤ 1M |
4.404 |
26.4216 |
26.4216 |
|
|
qwen-plus |
0 < tokens ≤ 256K |
3.5232 |
10.5684 |
35.2284 |
|
256K < tokens ≤ 1M |
10.5684 |
31.7052 |
105.6852 |
|
Qwen-Turbo
Chinese mainland
All prices are in CNY per 1 million tokens.
|
Model name |
Mode |
Input price |
Output price |
|
|
Non-thinking mode |
Thinking mode (chain of thought + answer) |
|||
|
qwen-turbo |
non-thinking and thinking modes |
0.36 |
0.72 |
3.6 |
International (including China (Hong Kong))
All prices are in CNY per 1 million tokens.
|
Model name |
Mode |
Input price |
Output price |
|
|
Non-thinking mode |
Thinking mode (chain of thought + answer) |
|||
|
qwen-turbo |
non-thinking and thinking modes |
0.4404 |
1.7616 |
4.404 |
Qwen Coder
Chinese mainland
All prices are in CNY per 1 million tokens.
|
Model name |
Input token range |
Input price |
Output price |
|
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 |
|
|
qwen3-coder-plus-2025-07-22 |
0 < token ≤ 32K |
4.8 |
19.2 |
|
32K < token ≤ 128K |
7.2 |
28.8 |
|
|
128K < token ≤ 256K |
12 |
48 |
|
|
256K < token ≤ 1M |
24 |
240 |
|
|
qwen3-coder-flash |
0 < token ≤ 32K |
1.2 |
4.8 |
|
32K < token ≤ 128K |
1.8 |
7.2 |
|
|
128K < token ≤ 256K |
3 |
12 |
|
|
256K < token ≤ 1M |
6 |
30 |
International (including China (Hong Kong))
All prices are in CNY per 1 million tokens.
|
Model name |
Input token range |
Input price |
Output price |
|
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 |
|
|
qwen3-coder-plus-2025-07-22 |
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 |
|
|
qwen3-coder-flash |
0 < token ≤ 32K |
2.6424 |
13.2108 |
|
32K < token ≤ 128K |
4.404 |
22.0176 |
|
|
128K < token ≤ 256K |
7.0452 |
35.2284 |
|
|
256K < token ≤ 1M |
14.0916 |
84.5484 |
Qwen3
Chinese mainland
All prices are in CNY per 1 million tokens.
|
Model |
Mode |
Input price |
Output price |
|
|
Non-thinking mode |
Thinking mode (chain of thought + answer) |
|||
|
qwen3-235b-a22b-instruct-2507 |
Non-thinking mode only |
2.4 |
9.6 |
- |
|
qwen3-235b-a22b |
Non-thinking and thinking modes |
2.4 |
9.6 |
24 |
International (including China (Hong Kong))
All prices are in CNY per 1 million tokens.
|
Model |
Mode |
Input price |
Output price |
|
|
Non-thinking mode |
Thinking mode (chain of thought + answer) |
|||
|
qwen3-235b-a22b-instruct-2507 |
Non-thinking mode only |
2.0256 |
8.1024 |
- |
|
qwen3-235b-a22b |
Non-thinking and thinking modes |
6.1644 |
24.66 |
73.98 |
Qwen2.5
Chinese mainland
All prices are in CNY per 1 million tokens.
|
Model name |
Input price |
Output price |
|
qwen2.5-72b-instruct |
CNY 4.8 |
CNY 14.4 |
International (including China (Hong Kong))
All prices are in CNY per 1 million tokens.
|
Model name |
Input price |
Output price |
|
qwen2.5-72b-instruct |
CNY 12.33 |
CNY 49.32 |
Qwen1.5
This model is currently available only in the Chinese mainland region.
|
Model |
Input price |
Output price |
|
qwen1.5-1.8b-chat |
Free for a limited time |
|
DeepSeek
This model is currently available only in Chinese mainland.
|
Model name |
Input price |
Output price Chain of thought and answer |
|
deepseek-r1-0528 |
CNY 4.8 |
CNY 19.2 |
GLM
This model is currently available only in the Chinese mainland region.
|
Model |
Mode |
Input tokens |
Input price (per 1M tokens) |
Output price (per 1M tokens) Chain of thought and answer |
|
glm-5 |
Non-thinking and thinking modes |
0 < token ≤ 32K |
CNY 4.8 |
CNY 21.6 |
|
32K < token ≤ 198K |
CNY 7.2 |
CNY 26.4 |
||
|
glm-4.7 |
Non-thinking and thinking modes |
0 < token ≤ 32K |
CNY 3.6 |
CNY 16.8 |
|
32K < token ≤ 166K |
CNY 4.8 |
CNY 19.2 |
DataWorks fine-tuned models
Chinese mainland
All prices are in CNY per 1 million tokens.
|
Model name |
Input tokens per request |
Input price |
Output price |
|
dw-qwen3-coder-30b-a3b-instruct |
0 < tokens ≤ 32K |
1.8 |
7.2 |
|
32K < tokens ≤ 128K |
2.7 |
10.8 |
|
|
128K < tokens ≤ 200K |
4.5 |
18 |
|
|
dw-qwen2.5-coder-32b-instruct |
No tiered pricing |
2.4 |
7.2 |
|
dw-qwen2-7b-instruct |
No tiered pricing |
1.2 |
2.4 |
International (including China (Hong Kong))
All prices are in CNY per 1 million tokens.
|
Model name |
Input tokens per request |
Input price |
Output price |
|
dw-qwen3-coder-30b-a3b-instruct |
0 < tokens ≤ 32K |
3.9636 |
19.8156 |
|
32K < tokens ≤ 128K |
6.6048 |
33.0264 |
|
|
128K < tokens ≤ 200K |
10.5684 |
52.842 |
Activation process
-
Go to one of the following purchase pages:
-
For existing DataWorks users, click Data Agent Quick Activation.
-
For users who do not have DataWorks, click DataWorks Bundle Purchase.
-
-
Select a Data Agent edition (Trial Edition, Personal Edition, or Team Edition).
-
If you select the Team Edition, set the number of seats. A minimum of 1 seat is required.
-
Complete the payment. Service activation is immediate.
Modification rules
Go to the DataWorks Data Agent activation page to change your edition, or add or remove seats for the Team Edition.
|
Operation type |
Supported editions |
Effective time |
Fee |
|
Version Upgrade |
Trial Edition → Personal Edition → Team Edition |
Effective immediately |
You are charged the price difference between the new and original editions, prorated for the rest of the current billing cycle. |
|
Version Downgrade |
Team Edition → Personal Edition → Trial Edition |
Takes effect at the start of the next billing cycle |
The features and quotas of your original edition remain active for the current billing cycle. No refund is provided for this cycle. |
|
Seat Addition |
Team Edition only |
Effective immediately |
The fee for the additional seats is prorated for the rest of the current billing cycle. |
|
Seat Reduction |
Team Edition only |
Takes effect at the start of the next billing cycle |
The original number of seats remains active for the current billing cycle. No refund is provided for this cycle. |
|
Unsubscription |
Personal Edition, Team Edition |
Effective immediately |
You receive a prorated refund for the rest of the current billing cycle. |
FAQ
-
Q: What is a seat?
A: A seat licenses one Alibaba Cloud account to use Agent features. All three editions include at least one seat. The Trial Edition and Personal Edition each include one fixed seat which only the main account can use. The Team Edition allows you to purchase multiple seats, which you can then assign to both the main account and RAM users.
-
Q: Is pay-as-you-go enabled automatically?
A: Yes. When you subscribe to any edition, pay-as-you-go is enabled by default. Once you exhaust your complimentary quota, the system automatically switches to pay-as-you-go billing. No further action is required.
Appendix: Context cache discount
|
Item |
Explicit cache |
Implicit cache |
|
Cost of input tokens for cache creation |
125% of the input token price |
100% of the input token price |
|
Cost of input tokens on a cache hit |
10% of the input token price |
20% of the input token price |