Pricing for detection Agent - AI Guardrails - Alibaba Cloud Documentation Center

Pricing and billing rules for the capabilities available in the AI Guardrails Detection Agent Center.

Billing

The Detection Agent Center includes two types of capabilities: custom detection agents and well-known IP (intellectual property) detection agents. Billing is based on the total character count of your prompts and the selected large language model. The billing rules are as follows:

Custom prompt character count: Prompts are processed in 3,000-character slices. Each slice counts as one billing unit, and a partial slice is billed as a full slice. For example, a 4,000-character prompt is billed as two slices.
Selected large language model: The billable item and unit price vary by model. For details about the mapping between models and unit prices, see Review types and unit prices.

Free quota deduction: If you have received the AI Guardrails Trial Package after activating the service, the system deducts from the free quota first. Billing at pay-as-you-go rates begins only after the free quota is exhausted or expires.
Resource package not fully deducted: If your bill shows charges but your resource package has not been fully deducted, go to Cost Center > Bill Details to check the specific billing breakdown. Confirm whether the resource package covers only certain detection types or usage quotas (which may result in excess usage being billed at pay-as-you-go rates), or whether other uncovered billing items exist.

Review types and unit prices

Review type	Model	Supported service (`service`)	Unit price
Custom Detection Agent - Basic `text_guard_llm_standard`	Qwen3Guard-Gen-4B	AI Input Content Security Check - international edition (`query_security_check_intl`) AI Generated Content Security Check - international edition (`response_security_check_intl`)	USD 8 per 10,000 calls Note When you select the `Qwen3Guard-Gen-4B` large language model, it uses built-in prompts. Therefore, billing is based on the number of calls, not the prompt length. For unit pricing, see Qwen3Guard-Gen-4B.
Custom Detection Agent - Basic `text_guard_llm_standard`	Qwen3.0-Flash	AI Input Content Security Check - international edition (`query_security_check_intl`) AI Generated Content Security Check - international edition (`response_security_check_intl`)	USD 8 per 10,000 slices
Custom Detection Agent - Advanced `text_guard_llm_advanced`	Qwen3.0-Plus	AI Input Content Security Check - international edition (`query_security_check_intl`) AI Generated Content Security Check - international edition (`response_security_check_intl`)	USD 16 per 10,000 slices
Custom Detection Agent - LLM36 - Level 1 `text_guard_llm36_tier_1`	Qwen3.6-Flash	AI Input Content Security Check - international edition (`query_security_check_intl`) AI Generated Content Security Check - international edition (`response_security_check_intl`)	USD 24 per 10,000 slices
Custom Detection Agent - LLM36 - Level 2 `text_guard_llm36_tier_2`	Qwen3.6-Plus	AI Input Content Security Check - international edition (`query_security_check_intl`) AI Generated Content Security Check - international edition (`response_security_check_intl`)	USD 40 per 10,000 slices
Custom Detection Agent - Basic - for-China edition `text_guard_llm_standard_cn`	Qwen3.0-Flash	AI Input Content Security Check - for-China edition (`query_guard_pro_ec`) AI Generated Content Security Check - for-China edition (`response_guard_pro_ec`)	USD 3.1 per 10,000 slices
Custom Detection Agent - Advanced - for-China edition `text_guard_llm_advanced_cn`	Qwen3.0-Plus	AI Input Content Security Check - for-China edition (`query_guard_pro_ec`) AI Generated Content Security Check - for-China edition (`response_guard_pro_ec`)	USD 6.2 per 10,000 slices
Custom Detection Agent - LLM36 - Level 1 - for-China edition `text_guard_llm36_tier_1_cn`	Qwen3.6-Flash	AI Input Content Security Check - for-China edition (`query_guard_pro_ec`) AI Generated Content Security Check - for-China edition (`response_guard_pro_ec`)	USD 9.3 per 10,000 slices
Custom Detection Agent - LLM36 - Level 2 - for-China edition `text_guard_llm36_tier_2_cn`	Qwen3.6-Plus	AI Input Content Security Check - for-China edition (`query_guard_pro_ec`) AI Generated Content Security Check - for-China edition (`response_guard_pro_ec`)	USD 15.5 per 10,000 slices
Custom Detection Agent - Image - Entry `image_guard_vl_basic`	Qwen3.0-VL-Flash	AIGC Input Image Security Check - international edition (`img_query_guard_intl`) AIGC Output Image Security Check - international edition (`img_response_guard_intl`)	USD 6 per 10,000 images
Custom Detection Agent - Image - Basic `image_guard_vl_standard`	Qwen3.0-VL-Plus	AIGC Input Image Security Check - international edition (`img_query_guard_intl`) AIGC Output Image Security Check - international edition (`img_response_guard_intl`)	USD 18 per 10,000 images
Custom Detection Agent - Image - Entry - for-China edition `image_guard_vl_basic_cn`	Qwen3.0-VL-Flash	AIGC Input Image Security Check - for-China edition (`img_query_guard_ec`) AIGC Output Image Security Check - for-China edition (`img_response_guard_ec`)	USD 2.4 per 10,000 images
Custom Detection Agent - Image - Basic - for-China edition `image_guard_vl_standard_cn`	Qwen3.0-VL-Plus	AIGC Input Image Security Check - for-China edition (`img_query_guard_ec`) AIGC Output Image Security Check - for-China edition (`img_response_guard_ec`)	USD 7 per 10,000 images

FAQ

Does AI Guardrails support viewing billing details by instance ID or service type? What is the UID in an instance ID?

AI Guardrails is billed uniformly per Alibaba Cloud account and does not support splitting or viewing billing details by instance ID or specific service type (such as Content Moderation).

In exported billing records, the instance ID follows the format 163245*****00260;text_guard_advanced. The numeric portion before the semicolon is your Alibaba Cloud account UID.