Billing methods for AI Data Preparation - Data Transmission Service

This topic describes the billing methods and standards for the DTS AI Data Preparation feature.

Billing Methods

Only pay-as-you-go (post-paid) billing is supported.

Billing Items

RAGFlow knowledge base (knowledge base basic service, knowledge base data preparation service).
Built-in models.
Enhanced parsing.
Data preparation tasks.

Billing Standards

RAGFlow Knowledge Base

RAGFlow knowledge base costs depend on the number of basic services and data preparation services.

Note

Unit: USD/instance/hour.
New RAGFlow knowledge base purchases include a 15-day free trial.
Fees on the purchase page are rounded. Actual charges follow the standards in the table below.

Region	Basic Service Billing Standard	Data Preparation Service Billing Standard
China (Hangzhou)	0.25	0.25
China (Shanghai)
China (Qingdao)
China (Beijing)
China (Zhangjiakou)
China (Hohhot)
China (Ulanqab)
China (Shenzhen)
China (Heyuan)
China (Guangzhou)
China (Chengdu)
China (Hong Kong)	0.3773	0.3773
Singapore	0.343	0.343
Malaysia (Kuala Lumpur)	0.2617	0.2617
Indonesia (Jakarta)	0.343	0.343
US East (Virginia)	0.2753	0.2753
US West (Silicon Valley)	0.2785	0.2785
Germany (Frankfurt)	0.3293	0.3293

Built-in Models

Built-in models are currently in public preview and are not charged.

China (Beijing)

Text and Vision Models (Billed by Token)

Model Name	Input Tokens per Request	Input Price (USD/1,000 Tokens)	Output Price (USD/1,000 Tokens)
`qwen3-vl-flash`	0 < Tokens ≤ 32K	0.0000264	0.000258
	32K < Tokens ≤ 128K	0.0000516	0.000516
	128K < Tokens ≤ 256K	0.0001032	0.0010308
`qwen3-vl-plus`	0 < Tokens ≤ 32K	0.0001720236	0.00172023
	32K < Tokens ≤ 128K	0.0002580348	0.0025803456
	128K < Tokens ≤ 256K	0.0005160696	0.0051606912
`qwen-vl-ocr`	Not applicable	0.0008604	0.0008604
`text-embedding-v4`	Not applicable	0.0000864	Not applicable

Audio Models (Billed by Duration)

Model Name	Price (USD/second)
`fun-asr`	0.0000384
`qwen3-asr-flash-filetrans`	0.0000384
`qwen3-asr-flash-realtime`	0.0000564

Singapore

Text and Vision Models (Billed by Token)

Model Name	Input Tokens per Request	Input Price (USD/1,000 Tokens)	Output Price (USD/1,000 Tokens)
`qwen3-vl-flash`	0 < Tokens ≤ 32K	0.00006	0.00048
	32K < Tokens ≤ 128K	0.00009	0.00072
	128K < Tokens ≤ 256K	0.000144	0.001152
`qwen3-vl-plus`	0 < Tokens ≤ 32K	0.00024	0.00192
	32K < Tokens ≤ 128K	0.00036	0.00288
	128K < Tokens ≤ 256K	0.00072	0.00006
`qwen-vl-ocr`	Not applicable	0.000864	0.000864
`text-embedding-v4`	Not applicable	0.000084	Not applicable

Audio Models (Billed by Duration)

Model Name	Price (USD/second)
`fun-asr`	0.000042
`qwen3-asr-flash-filetrans`	0.000042
`qwen3-asr-flash-realtime`	0.000108

Enhanced Parsing

Pricing for Enhanced Parsing varies by version. The Basic version uses a flat rate, while the Medium version uses tiered pricing based on your monthly cumulative parsed page count.

Note

Each image document page consumes at least 2,000 tokens from a state-of-the-art (SOTA) model for data parsing and related preparation tasks.

Version	Monthly Cumulative Parsed Pages	Price Per Page (USD)
Basic	-	0.00289
Medium	0–10,000	0.0362
	10,001–100,000	0.0260
	100,001–500,000	0.0217
	500,001–1,000,000	0.0174
	>1,000,000	0.0145

Data Preparation Tasks

Note

Pricing unit: USD/hour .

Region	Billing rate
China (Hangzhou)	0.6443
China (Shanghai)
China (Qingdao)
China (Beijing)
China (Zhangjiakou)
China (Hohhot)
China (Ulanqab)
China (Shenzhen)
China (Heyuan)
China (Guangzhou)
China (Chengdu)
China (Hong Kong)	0.684
Singapore
US East (Virginia)
US West (Silicon Valley)
Germany (Frankfurt)	0.636
Malaysia (Kuala Lumpur)	0.756
Indonesia (Jakarta)	0.84