Q: What is a Token, and how is Token usage calculated?
A: A Token is the smallest unit of text processed by a large language model. It can be understood as a fragment of a word.
Chinese: One Chinese character uses approximately 1.5–2 Tokens.
English: One English word uses approximately 1–1.5 Tokens.
Punctuation marks, spaces, and line breaks also consume Tokens.
Example: “你好，世界！” uses approximately 7–10 Tokens.
Total Token usage = Input Tokens + Output Tokens
Input Tokens: System prompt + conversation history + current user input
Output Tokens: Content generated by the model
Q: How does billing work, and what are the prices?
A: MaiToken uses a prepaid billing model. You add credits to your account first, and the corresponding number of credits is automatically deducted as you use the service.
Models share a unified credit balance.
There is no minimum spending requirement. Services will be suspended if the account balance is insufficient.
Q: Are Input Tokens and Output Tokens priced the same?
A: No. Output Tokens usually cost two to four times more than Input Tokens because generating content requires more computing resources than processing input. Exact prices vary by model. Refer to the pricing page for details.
Q: Why is my Token usage higher than expected?
A: Common causes and optimization recommendations include:
The system prompt is too long → Shorten it to fewer than 500 Tokens.
Too much conversation history is included → Truncate the conversation by retaining only the most recent 5–10 rounds, or compress earlier messages into a summary.
max_tokensis set too high → Adjust it according to actual requirements. For customer service scenarios, 256–512 Tokens are usually sufficient.enable_thinkingis enabled → Disable deep thinking for simple tasks to reduce thinking Token usage.Requests are retried frequently → Check whether a code bug is causing duplicate API calls.
Images are included in the input → Multimodal image input can consume a large number of Input Tokens. Control the number and resolution of images.
You can view usage details by day, API Key, and model under “Usage Records” in the console to identify the source of Token consumption.
Q: How do I view usage and billing information?
A:
Real-time usage: Console → “Usage Records” → filter by day, week, month, API Key, or model.
Billing details: Console → “Wallet Management” → “Transactions.”
Balance: The current available balance is displayed in the lower-left corner of the console.
Spending alerts: Configure SMS or email notifications when the balance falls below a specified amount.
Q: How do I add funds or make a payment?
A:
Go to Console → “Wallet Management” → “Select Recharge Amount” → “Recharge.”
Supported payment methods: Alipay, WeChat Pay, bank transfer, and corporate remittance.
Monthly billing is available to enterprise customers with a signed agreement.
Processing time: Online payments are credited immediately; corporate remittances take 1–3 business days.
Q: Can I request an invoice?
A: Yes.
Go to Console → “Wallet Management” → “Request Invoice.”
The invoice amount is based on the amount already consumed.
Supported invoice types: Electronic general VAT invoice and special VAT invoice.
General VAT invoice: Issued within 10 business days after the application is submitted.
Special VAT invoice: General taxpayer qualification documents are required, and the invoice will be issued within 5–10 business days.
Q: Can I get a refund for unused credits?
A:
Within seven days of activation, if no API calls have been made → A full refund is available.
If the credits have been partially used → The used amount will be deducted, and the remaining balance may be refunded after accounting for any discount difference.
If abnormal charges are caused by a platform failure → The corresponding charges will be refunded in full.
Refund process: Submit a support ticket → verification, approximately 10 minutes → approval → refund to the original payment method within 3–5 business days after approval.