Frequently asked questions
Answers about base URLs, keys, billing, free trial credit, payments, refunds and data handling on BigModel.io.
Which base URL do I use?
OpenAI-style clients and SDKs use https://bigmodel.io/v1. Anthropic-style clients such as Claude Code and the Anthropic SDK use https://bigmodel.io without /v1, because they add /v1/messages themselves.Quickstart →
How is this different from the official APIs?
The same models behind one key and one bill: prices in US dollars, top-ups with Alipay or USDT, and OpenAI- and Anthropic-compatible endpoints for the tools you already use.
How am I charged?
Per token, in US dollars, from a prepaid balance. Input, output, cache-read and cache-write tokens are each charged at the model's listed price. Failed requests are not charged, and every charge appears in your usage log.Billing and top-ups →
Is there free credit?
Free trial credit when you sign up. It is added to your USD balance and spent at normal model prices. Models marked Free cost $0 per token but, like every request, need a balance above $0.Trial credit and free models →
How can I pay?
Top up on the payment page in the console. The methods available are those shown there, for example Alipay and USDT (TRC20, ERC20, BEP20). Your balance is always in US dollars, and any fee is shown before you pay.
Can I get a refund?
Unused plans can be refunded within 24 hours of purchase. USDT payments are not refundable, and gifted credit has no cash value.Refund policy →
Can I cap what a key spends?
Yes. Each key can have a dollar quota, 5-hour, 1-day and 7-day spend windows, an expiry date and IP rules.Limits and rate limits →
Can I use GLM, Kimi or DeepSeek in Claude Code?
Yes. Set ANTHROPIC_BASE_URL to https://bigmodel.io, ANTHROPIC_AUTH_TOKEN to your key and ANTHROPIC_MODEL to a model ID from the directory.Claude Code guide →
Does my OpenRouter code work?
Mostly, after changing the base URL and key. provider/model IDs, the models fallback array and the HTTP-Referer / X-Title headers are supported; the provider object, transforms and route are ignored.Migrate from OpenRouter →
Are my prompts stored or used for training?
Prompts and outputs of successful requests are passed to the model provider and not kept after delivery. Bodies of failed requests are kept for up to 30 days, with credentials removed, for troubleshooting. Prompts are never used for training. Providers may keep data under their own terms.Privacy policy →
How do I know a model is working?
The status page shows availability, latency and time to first token for each model over 24 hours, 7 days and 30 days.Status →
Can I get an invoice or pay as a company?
Email support@bigmodel.io with your account email and what you need.
How do I contact support?
Send a message from Support in the console, or email support@bigmodel.io. Include the X-Request-ID, time, model and error message so we can find the request.
More answers in the docs FAQ, or email support@bigmodel.io. Docs FAQ