ModelArts Studio (MaaS) Text Generation Model

Input and output of text dialogs are converted into tokens for billing.

Billing Mode

Token-based billing is supported.

Token-based Billing

Model inference services process input by converting it into identifiable tokens. Using a built-in MaaS service charges you based on token count. The specific pricing is subject to the MaaS console.

**Table 1** Billing items
Model	Billing Sub-item		Unit Price (USD/1,000 Tokens)	Description	Billing Formula
DeepSeek-V3.1	Input		0.000539	Input and output tokens used for model inference.	Fee = Number of used input tokens x Unit price of input tokens + Number of used output tokens x Unit price of output tokens
DeepSeek-V3.1	Output		0.001617
Qwen3-32B	Input		0.000270
	Output	Reasoning pattern	0.002695
	Output	Non-reasoning pattern	0.001078

Billing Example

The prices below are for reference only. For the actual prices, see pricing details on the MaaS console.

If you call a DeepSeek-V3.1 built-in service, the number of input tokens is 3 million, and the number of output tokens is 7 million, the fee = 3 million/1,000 x 0.000539 + 7 million/1,000 x 0.001617 = $12.936 USD.

Parent topic: ModelArts Studio (MaaS) Model Service Prices

Previous topic: ModelArts Studio (MaaS) Model Service Prices

Next topic: Model Inference Billing Items