Pricing

Pricing principles

Yomi API offers competitive prices while maintaining service quality:

• International model pricing: Prices for OpenAI, Claude and other international models match their official prices.
• Lower prices for Chinese models: DeepSeek, GLM, Kimi and other Chinese models cost less than their official prices.
• Transparent pricing: All prices are published in the Model Catalog, with no hidden charges.
• Open and transparent: All prices are published in the Model Catalog, with no hidden charges.

Credit purchases and bonuses

• Currency and payment: Your balance is denominated in US dollars (USD) and paid in Chinese yuan. The exchange rate shown on the payment page applies.
• Payment methods: Alipay, WeChat Pay and other online methods are supported according to the options shown on the payment page.
• Low minimum: Start with a small credit purchase.
• Bonus credit: Larger purchases receive larger bonuses; see the payment page for current tiers.

To add credit, open "Add Credit" in the console User Center sidebar.

How billing works

Pay as you go

• No subscription: No monthly fee, annual fee or minimum spending requirement.
• Pay only for usage: Charges arise only from actual API calls.
• Balance does not expire: Except for credit awarded in specific promotions.

Per-request billing and priority

Text models are billed by tokens consumed. Whether media models such as image and video models support per-request billing depends on the models currently offered in the Model Catalog ("Models" in the console).

Tokens and cost components

• Definition: A token is the basic unit of text processed by a model. One Chinese character is roughly 1–2 tokens; 1,000 tokens correspond to approximately 750 English words or 500 Chinese characters.
• Components: Cost = input (prompt) token cost + output (completion) token cost. Output tokens usually cost more than input tokens.
• Model differences: Unit prices vary by model. Refer to the model and route prices in the Model Catalog.

Model selection and cost control

Choose by budget

• Cost first (batch processing, simple tasks, development and testing): deepseek-v4-flash, glm-5.2, deepseek-v4-pro
• Performance first (complex reasoning, creative writing, coding): claude-fable-5, claude-opus-4-8, gpt-5.6-sol
• Balanced value: claude-sonnet-5, deepseek-v4-pro

Estimate your costs

1. Test a small sample: Start with 5–10 real data samples.
2. Review usage: Check token consumption and spending under "Usage" in the User Center.
3. Calculate an average: Find the average token count per call.
4. Estimate the total: Average tokens × expected number of calls × model unit price.

Service commitments

• High concurrency: Concurrency limits depend on the model and route; see route information in the Model Catalog. Businesses can contact support for higher limits, without a risk of account bans.
• No migration cost: A unified API compatible with the OpenAI format.
• Enterprise service: Professional technical support, reliable service and data security.

Invoices

You can request electronic VAT invoices (general) for your top-ups online: open "Invoices" in the User Center sidebar, where paid top-ups are listed as needing details. For one order, choose "Add details", enter the invoice title and submit; to put several orders on one invoice, tick them and choose "Combine". Once it is issued, download the PDF from the Invoices page.

For a special VAT invoice, contact support.