Question 1

How to register an account? What to do after registration?

Accepted Answer

How to RegisterClick "Register" in the top right corner of the page and use your email address and a password to create an account (some environments may require an email verification code; please follow the instructions on the registration page). Before registering, please read the "Privacy Policy" and check the box to agree before submitting.What to Do After RegistrationLogin: Once registered, log in using your email and password;Top-up: Go to the "Console" page to top up your balance, which can be used for API calls;Create API Key: Go to "User Center → API Keys" to create a key starting with sk-;Call: Point your gateway address to this site and use your API Key for authentication to call the Claude model.

Question 2

Do new users get a bonus upon registration?

Accepted Answer

Yes. After successful registration, new users can receive a trial balance bonus (default ¥1, subject to on-site configuration), which can be directly used for API calling trials. The bonus balance is the same as recharged balance, and will be deducted in real time based on token usage. You can check it in “User Center → Balance.” Once the trial balance is used up, please recharge with the relevant package to continue using the service.

Question 3

How to Recharge? What Payment Methods Are Supported?

Accepted Answer

Recharge StepsLog in to your account and go to the "Console" page;Enter the recharge amount and click "Buy Now" to place the order;Complete the payment as prompted on the payment page (Alipay, subject to what is actually displayed);After successful payment, the balance will be credited instantly. You can verify it under "Console → Balance".If the payment is successful but the balance is not credited, please keep the payment receipt screenshot and contact customer service. We will verify and reissue it as soon as possible.

Question 4

What are the recharge packages? What are the bonus rules?

Accepted Answer

We currently offer several recharge packages, for example: PackageBalance CreditedDescription Recharge ¥10¥10Basic tier Recharge ¥50¥50Includes bonus points, better value Higher-denomination packages usually include more bonus points (e.g., the ¥50 tier includes 30 points). Package names, prices, and credited points are subject to the real-time display on the "Pricing" page; please verify before completing your payment.

Question 5

How is my balance billed? How are costs calculated?

Accepted Answer

This site charges based on token usage, with real-time deductions applied after successful calls. The formula for a single request is: Cost (CNY) = (Input tokens × Input unit price + Output tokens × Output unit price + Cached hit tokens × Cache read unit price + Cache write tokens × Cache write unit price) × Channel multiplier ÷ 1,000,000 Unit Price: Each model has its own input / output / cache read / cache write unit price (CNY / million tokens), which can be found on the "Pricing" page or via GET /api/v1/relay/models; Multiplier: Different channels or time periods may have multiplier adjustments. Currently, the multiplier for all models is 1.0 (meaning billing is at the original price); Cache: Tokens that hit the prompt cache are charged at the cache read unit price (which is significantly lower than the input price), while tokens written to the cache are charged at the cache write unit price; Details: The token count and cost for every call are recorded in the usage logs, which can be verified under "User Center → API Usage".

Question 6

Which models are supported? What are the prices?

Accepted Answer

Currently available models via the gateway:Flagship models from Claude / OpenAI / Gemini are all supported(Unit price: CNY / million tokens; for cache write prices and real-time multipliers, please refer to the response from GET /api/v1/relay/models.)The integration of models such as OpenAI / Gemini follows official documentation; the actual available models are subject to the response from GET /api/v1/relay/models. Calling a disabled model will return a 400 error.

Question 7

How to Create and Use an API Key?

Accepted Answer

Creating a KeyAfter logging in, click "Create" under "Console → API Keys";The system generates a key in the sk-xxxxxxxx format. Please copy and save it immediately after creation;You can view / copy the created key in the key list.Using the KeyWhen calling the API, include it in the request header: Authorization: Bearer sk-your-key or x-api-key: sk-your-key;The key is linked to your account balance. If the key is invalid, disabled, or the balance is insufficient, you will receive a 401 / 403 error;Do not expose your key in public repositories or frontend code. If you suspect it has been leaked, disable it immediately and create a new one.

Question 8

How to call the API? Authentication header and request examples

Accepted Answer

API Endpoints MethodPathDescription POST/api/v1/relayRelay forwarding entry, request body follows Anthropic /v1/messages payload GET/api/v1/relay/modelsList of models and unit prices (no authentication required) GET / POST/api/v1/relay/keysAPI key management (authentication required) GET/api/v1/relay/usageUsage details (authentication required) curl Example curl -X POST http://127.0.0.1:8080/api/v1/relay \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-your_key" \ -d '{ "model": "claude-sonnet-5", "max_tokens": 1024, "messages": [ {"role": "user", "content": "Introduce yourself in one sentence"} ] }' For production environments, please replace 127.0.0.1:8080 with your gateway domain. You may also include parameters such as system prompts or temperature; see the "Documentation" for a full field specification.

Question 9

Is streaming output supported? How do I use it?

Accepted Answer

Yes. Add "stream": true to the request body, and the server will return responses in chunks using SSE (Server-Sent Events), with event types consistent with Anthropic: event: message_start data: {"type": "message_start", "message": {"id": "msg_...", "model": "claude-sonnet-5"}} event: content_block_delta data: {"type": "content_block_delta", "delta": {"type": "text_delta", "text": "Hello"}} event: message_delta data: {"type": "message_delta", "delta": {"stop_reason": "end_turn"}, "usage": {"output_tokens": 18}} event: message_stop data: {"type": "message_stop"} Clients should parse the JSON from the data: lines, ignoring the event: lines and empty lines; receiving message_stop indicates the stream has ended. Billing is settled server-side based on the usage of the entire request, regardless of whether streaming is used.

Question 10

How is platform stability ensured? What to do in case of upstream failure?

Accepted Answer

The platform ensures stability through the following measures: Multi-channel Disaster Recovery: Supports configuring multiple upstream channels with automatic failover based on priority; in case of a single channel failure, the system automatically switches to a backup channel (recommended for production environments). Error Code Normalization: Upstream anomalies are mapped to clear HTTP status codes (e.g., 500 / 529), making them easy for clients to handle. Call Logging: Every call records the model, tokens, costs, and status, facilitating troubleshooting and reconciliation. Timeout Control: Implements timeouts for upstream requests to prevent hanging processes. If a call hits an upstream failure (500 / 529), we recommend retrying later (exponential backoff is suggested). If the issue persists, please provide your request_id and contact customer support; we will investigate and switch channels immediately.

Question 11

How to view usage and billing details?

Accepted Answer

After logging in, you can view your recent call records in "Console → Usage Logs". Each record includes:Model: The model slug used for this call;Tokens: Input tokens, output tokens, cache read / cache write tokens;Cost: The amount deducted for this call (CNY);Status: success, no_quota (insufficient balance), etc.It is recommended to pass a custom request ID via metadata.request_id when making calls, which will help you precisely reconcile your accounts in the usage logs.

Question 12

What happens if my balance is low? How does billing work?

Accepted Answer

Billing Rules Fees are deducted in real-time based on actual token usage after a successful API call (see "How is the balance charged?" for the formula); Deductions occur after the server completes the upstream request and receives usage data, with amounts accurate to the cent; Failed or intercepted requests will not be charged. Insufficient Balance When your account balance is insufficient to cover a call, the interface will return 403 Insufficient balance, please top up, and you will not be charged for this call (a "no_quota" record will be logged). You can resume API calls immediately once the top-up is processed. We recommend monitoring your balance and topping up in time to avoid any impact on your live services.

Question 13

Are there any coupons or referral rebate programs?

Accepted Answer

Yes, the platform offers two types of promotions: Coupons The platform supports coupon types such as flat-off discounts / percentage discounts / bonus points. Methods for obtaining them depend on site events, announcements, and distributions by customer service. If you have valid coupons, you may select them for deduction during top-up. Referral Rebates After inviting friends to register and top up, you can receive a certain percentage of commission rebates. You can view your invitation records, commission details, and withdrawal options in the "User Center → Invitation / Commission" section. Specific rebate percentages are subject to the invitation page and site rules.

Question 14

Is my data secure? Are my requests saved?

Accepted Answer

We take your data security very seriously: Passwords: Stored encrypted; the platform cannot view them in plain text. API Keys: Visible only to you. Please keep them secure and do not share them with third parties. Call Logs: Only billing and reconciliation information such as model used, token usage, cost, and status is recorded; the actual content of the conversations is not stored. Access Control: All interfaces are authenticated; unauthorized requests cannot utilize your keys. Reminder: Please do not submit sensitive personal data such as ID numbers, passwords, or payment information in your request content. For content involving corporate secrets, please use your company's self-hosted environment. See our "Privacy Policy" for detailed terms.

Question 15

Do you support refunds? What is the refund policy?

Accepted Answer

Recharge balance is used to offset API call fees, and refunds are generally not supported without reason. You can contact customer service to apply for handling in the following cases: Payment successful but balance not credited (duplicate payment / payment exception); Long-term inability to use normally due to platform reasons that cannot be fixed; Other situations verified to be the platform's responsibility. Please provide payment voucher screenshots, order number, and account email when applying, and customer service will handle it as soon as possible after verification. Accounts banned due to malicious activity or abuse are not eligible for refunds. For specifics, please refer to the latest platform policy.

Question 16

How can I contact customer support if I have an issue?

Accepted Answer

You can contact us in the following ways: Online Support: The floating customer service icon in the bottom-right corner of the page (WeChat / QQ / Email, depending on what is currently displayed); In-site Messaging: Log in to your user center to submit feedback or questions. To help us resolve your issue faster, please provide the following information when reaching out: The request_id used during the call (found in metadata.request_id) or the exact time of the request; The first few digits of your API Key, the error code, and the returned error message; For billing or recharge issues, please include your order number and a screenshot of the payment receipt. Our customer service hours are generally 9:00 – 22:00. If you leave a message outside of these hours, we will respond as soon as possible.

Question 17

Do you support OpenAI / Gemini compatible interfaces?

Accepted Answer

The current gateway primarily supports the Anthropic Messages API (/v1/messages) compatible format, accessible via POST /api/v1/relay: Anthropic Compatibility: ✅ Supported. The request body, response structure, and streaming events are identical to Anthropic's, allowing for integration with Claude Code, the Anthropic SDK, and others by simply updating the Base URL. Gemini Compatibility: Planned according to official documentation (e.g., /v1beta/models/{model}:generateContent); current availability depends on the actual live environment. OpenAI Compatibility: A native /v1/chat/completions endpoint is not currently provided. Requests in OpenAI format must be converted to the Anthropic messages format before being called. Most aggregation tools that support custom Base URLs (such as one-api or new-api) can handle this format conversion. For integration details, please refer to the "Usage Documentation" (Claude API Integration Guide, Gemini API Integration Guide).