Agent Metering and Billing

Last Updated on : 2025-03-27 06:44:37download

This topic describes each billing item and the related costs of AI agents.

Billing items

The billing items for the AI agent currently mainly include model fees and intelligent voice fees.

Billing cycle

Each billing item is charged on a daily basis and the fees are settled based on usage. The billing statement is usually generated one day after the end of the current billing cycle, with the exact billing time subject to the system.

Model fees

Available models

For information about how to differentiate model types, you can refer to Available Models.

Billing formula

The model service charges are based on the token usage of the model, with the billing formula as follows:

Model cost = Token usage × Model token unit price

In large language models, tokens are the basic units of text processing. Models typically break down input text into a series of tokens, which are then processed and analyzed. Tokens can be words, characters, subword fragments, or other forms of text segments, depending on the tokenization algorithm used by the model. Therefore, the calculation and processing of tokens may vary according to the specific architecture and design of the model.

Unit Price

Model name Billing item Unit price (per million tokens)
Qwen Qwen-Max Input ¥2.40
Qwen-Max Output ¥9.60
Doubao Doubao-Pro-32k Input ¥0.80
Doubao-Pro-32k Output ¥2.00
Deepseek Deepseek-Chat Input ¥2.00
Deepseek-Chat Output ¥8.00
ChatGPT GPT-4o Input $2.50
GPT-4o Output $10.00
GPT-4o-mini Input $0.15
GPT-4o-mini Output $0.60
Gemini Gemini-1.5-pro Input $1.25
Gemini-1.5-pro Output $2.50
Gemini-2.0-flash Input $0.10
Gemini-2.0-flash Output $0.40
Mistral Mistral Large Input -
Mistral Large Output -
Claude Claude 3.5 Haiku Input $0.80
Claude 3.5 Haiku Output $4.00
Claude 3.7 Sonnet Input $3.00
Claude 3.7 Sonnet Output $15.00
Nova Nova Pro Input -
Nova Pro Output -

Smart voice fees

Billing formula

The smart voice service is divided into two parts: speech input (ASR) and speech output (TTS). The billing formula is as follows:
Smart voice fees = ASR unit price × Input audio duration + TTS unit price × Output audio character count

  • Automatic speech recognition (ASR): The technology that can recognize and understand natural audio language input from humans, by analyzing and processing speech signals, thereby converting audio into text.
  • Text to speech (TTS): The technology that can convert text into speech output, simulating human reading, thereby transforming written information into audio information.

Unit price

ASR vendor Unit price (per hour)
AISpeech ¥5.00
Google $1.44
TTS vendor Unit price (per thousand characters)
AISpeech ¥0.20
Huoshan ¥0.30
Azure $0.015