All Products
Search
Document Center

AI Guardrails:Activation and billing overview

Last Updated:Mar 31, 2026

Guardrails uses pay-as-you-go billing. Activating the service is free — charges apply only when you make API calls or run moderation checks.

Important

After activation, the Content Compliance check service is enabled by default. To use sensitive content detection or prompt attack detection, enable these features separately in the console.

Activate pay-as-you-go

  1. Go to the Guardrails activation page.

  2. Read and agree to the Terms of Service, then click Buy Now.

Activation is free. Billing starts when you use the service.

Pricing

Charges are based on the connection type, moderation type, and the volume of content checked.

Billing units

Two billing units apply depending on the connection type and model selected:

UnitDefinitionWhen it applies
Segment1 segment = up to 3,000 characters. Content under 3,000 characters counts as 1 segment.Billing is based on the total character count of the custom prompt.
CallEach API request = 1 call, regardless of content length.The selected model does not involve custom prompts.

Which model determines the billing method

The large language model (LLM) you select determines whether you are billed per segment or per call:

LLMBilling methodNotes
Qwen3_FlashPer segment (Basic)Billing is based on the total character count of the custom prompt.
Qwen3Guard-Gen-4BPer call (Basic)No custom prompt involved; each call is one transaction.
Qwen3_PlusPer segment (Advanced)Billing is based on the total character count of the custom prompt.

Price table

Connection typeModeration typeService identifierSupported servicesUnit price
API accessCustom Detection Agent – Per Segment – Basictext_guard_llm_standardAI Input Content Security Check (query_security_check), AI-generated Content Security Check (response_security_check)USD 8.00 / 10,000 segments

API access

Custom Detection Agent - Per Call - Basic

text_guard_llm_standard

  • AI Input Content Security Check (query_security_check)

  • AI-generated Content Security Check (response_security_check)

USD 8.00 per 10,000 calls

Important
  1. This item is billed when you select the Qwen3Guard-Gen-4B large language model.

  2. Because Qwen3Guard-Gen-4B does not involve custom prompts, billing is based on the number of calls. Each call is billed as a single transaction.

API accessCustom Detection Agent – Per Segment – Advancedtext_guard_llm_advancedAI Input Content Security Check (query_security_check), AI-generated Content Security Check (response_security_check)USD 16.00 / 10,000 segments

API/AI Gateway integration

Text moderation - by call - advanced

text_guard_advanced

  • AI input content security check (query_security_check_intl)

  • AI-generated content security check (response_security_check_intl)

USD 6/10,000 calls

API/AI Gateway integration

Prompt attack moderation

  • AI input content security check (query_security_check_intl)

  • AI-generated content security check (response_security_check_intl)

USD 6/10,000 calls

Important

This feature will be commercially billed starting from January 22, 2026. If you do not need this service, disable it in the Guardrails console to avoid incurring fees.

API/AI Gateway integration

Sensitive content moderation - by call - advanced

text_sddp_advanced

  • AI input content security check (query_security_check_intl)

  • AI-generated content security check (response_security_check_intl)

USD 6/10,000 calls

Important

Prompt attack moderation is commercially billed starting January 22, 2026. If you do not need this feature, disable it in the Guardrails console to avoid charges.

Billing cycle and overdue payments

Guardrails uses a 24-hour billing cycle. Bills are generated 10 to 12 hours after the end of each calendar day and deducted automatically from your Alibaba Cloud account balance.

If your account has an overdue payment, the service is paused. After you pay the overdue amount, the service resumes within 5 minutes.

Enable additional detection features

After activation, go to the Guardrails console and navigate to Protection Configuration > Configuration to enable optional detection features based on your needs.

The following features are available:

  • Sensitive content detection — billed separately

  • Prompt injection detection — billed separately

  • Malicious file detection

  • Digital watermarking

Important

When you enable Sensitive content detection or Prompt injection detection, review the billing information in the confirmation window before proceeding.