Billing Overview
最后更新时间：2025-12-15 16:31:59
Tencent Cloud Agent Development Platform (Tencent Cloud ADP) is an application build platform based on large models. It bills according to the number of tokens used by the user, knowledge base capacity, concurrency number and other resources. Currently, you can obtain a certain amount of free quota just by registering, enabling the service and experiencing it. To purchase or scale out, check Purchase Method or contact your sales.
Activation Method
First-time use of Tencent Cloud ADP requires clicking Activate to activate platform usage permissions.
Free Quota
When a verified master account logs in to the Tencent Cloud ADP platform for the first time, a certain amount of free quota tokens will be obtained, as shown in the figure below:
Resource Category | Free Tier |
DeepSeek-R1-0528 | Upon activating the Tencent Cloud Agent Development Platform (ADP), you will receive a total of 3,000,000 free tokens, valid for 2 months. It will be distributed to your Tencent Cloud account in the form of a resource package and deducted first. |
DeepSeek-V3-0324 | |
Application Service - Knowledge Base Storage | Upon activation, you will receive a total free quota of 3 million characters of knowledge-base storage, valid for 6 months. Upon expiration, the knowledge in the knowledge base will be marked as "Exceeding Quota - Invalid". Knowledge in this state will not take effect in conversations, and any knowledge that remains in this state for 30 days will be automatically deleted. If you need to continue the knowledge base, please purchase additional storage before the expiration date. |
Special Notes
In the following scenarios/functions, interacting will incur token consumption:
Application Configuration:
Knowledge Base Management > Q&A > Import Q&A > Auto-generate Q&A from documents.
Knowledge Base Management > Task Flow > Insert Node > Auto-generate questions.
Knowledge Base Management > Task Flow > Insert Node > Auto-generate answer previews.
During Q&A interactions, token consumption will occur in: application configuration testing, calling the application API after publishing the app, using the app experience link, application evaluation, and application trials.
Token calculation includes: User input + System prompt + Recalled information (documents, Q&A content) + Model output
Notes
1. Token conversion ratio differs by model and language. Typically, 1 token is approximately 1 Chinese character or 3 English letters, but actual counts may vary.
2. For configured task flows that call specialized task-type models, tokens will be consumed accordingly.
3. Tokens are consumed for application configuration triggered by dialogue interactions, welcome messages, output configuration, fallback replies, and replies after sensitive-word filtering
4. In Application Configuration > Knowledge Sources, the number of retrieved documents and Q&A items affects token usage. The higher the set quantity, the more recall fragments are spliced into the input for the large model, and the consumption of token quantity rises accordingly.
Note:
When calculating your bill, the system will deduct usage in the order of Free Quota > Prepaid Resource Packages, meaning free quota is always consumed first.
If your service is suspended due to overdue payment or violations, you will no longer be able to enjoy the free quota. Once the service is restored, you may continue to use any remaining free quota within the validity period.
General free quota: The DeepSeek series provides a total of 500,000 free tokens.
Free quota with earlier expiration dates will be consumed first.
Billing Outline
Billing Scenario | Billing Type | Resource Type | Description |
Application Service | Application Service - Preset Model | Large Model Pay-as-you-go Billing | Multiple preset large model service call prices. |
| | Exclusive Concurrency Package for Large Models | |
| Application Service | Knowledge Base Expansion Package | Purchase a Knowledge Base Expansion Package to upload documents and Q&A to knowledge management for implementing the QA scenario. |
Purchase Instruction:
If you use knowledge management in the application, you need to purchase a knowledge base expansion package.
Product Pricing
Application Service
Large Model Postpaid Billing
Preset models offer a postpaid billing mode. Postpaid billing charges based on the number of tokens generated by actual API calls. Different models have different prices. Charges are calculated based on the total number of input and output tokens actually consumed, and per hour, fees are deducted from your Tencent Cloud account.
Model Version | Unit Price |
DeepSeek-R1-0528 | Input: 0.01051 USD/thousand tokens Output: 0.03032 USD/thousand tokens |
DeepSeek-V3-0324 | Input: 0.00465 USD/thousand tokens Output: 0.00687 USD/thousand tokens |
Note:
Support enabling postpaid mode in System Management - Postpaid Settings. (If enabling postpaid for DeepSeek does not take effect, submit a ticket to contact us.)
Sub-accounts have no feature permissions for postpaid settings by default. The root account can grant permission in user permission management.
Large Model Concurrency Package
Definition: The number of concurrency sessions in progress. The entire process from sending a request to receiving all streaming results is counted as concurrency usage. A dialogue occupies concurrency for about 3-15 seconds.
Payment method: Monthly and Annual Subscriptions.
Exclusive Concurrency Package Price for Large Models
Model Version | Concurrency Limit | Monthly Subscription | Yearly Subscription |
DeepSeek R1-0528 | 1 concurrency | $4,500 | $53,900 |
| 5 concurrency | $22,400 | $268,000 |
DeepSeek V3-0324 | 1 concurrency | $3,300 | $39,500 |
| 5 concurrency | $16,400 | $196,000 |
Purchase Instruction:
Purchase an exclusive concurrency package for independent use without purchasing token resources.
Exclusive concurrency purchase requires confirming resource deployment conditions with the connected customer manager. If there is no connected customer manager, contact customer service inquiry.
Knowledge Base Capacity
Definition: Knowledge Base Capacity is calculated based on the total number of characters in all documents and Q&A content uploaded across all applications. Deleted documents and deleted Q&A entries do not occupy capacity.
Resource Validity: Optional, after expiration renewal is required to continue using the service. Expired knowledge base capacity can only be used for deletion operations.
Billing Method: Annual or monthly subscription.
Knowledge Base Expansion Package | Monthly Subscription | Yearly Subscription |
10 Million Characters | $70 | $750 |
100 million characters | $600 | $6,500 |
Note:
Character count does not include spaces in documents. Images embedded in documents are converted into image links, and each image is counted as approximately 150 characters.
A 30,000-character master's thesis is roughly equivalent to 40,000 characters or a 50-page document. 10 million characters are approximately equivalent to 250 master's theses, and 100 million characters are approximately equivalent to 2,500 master's theses.
Knowledge Base Expansion Package Expiry Instructions:
After expiration, knowledge stored in the Knowledge Base will enter an "over-quota" state. Knowledge in the "over-quota" state cannot be used in conversations, and any content in this state will be automatically deleted after 30 days.
Knowledge in the "over-quota" state will only become available for use in conversations after the expansion package is repurchased and the knowledge is manually republished in bulk.
文档反馈