About XHuoAPI
Before you route traffic through an API gateway, it is worth knowing where requests go, whether content is kept, and who to talk to when something fails. The answers are below, with links to the relevant policy where one exists.
What is XHuoAPI?
A model gateway. We do not train or own any models; we forward requests to upstream services and handle billing. One key and one balance cover 70+ models from 9 vendors, including DeepSeek, Qwen, GLM, Kimi and Doubao alongside GPT, Claude, Gemini and Grok. The API follows the OpenAI Chat Completions format, and Claude models also accept the Anthropic Messages format.
Who is it for?
Developers who use models from several vendors and want one key and one bill. If you only need a single vendor, that vendor's own platform or international site is usually simpler.
Where do my requests go?
The gateway runs on servers in Phoenix, Arizona, United States, behind Cloudflare. Request content is forwarded to the upstream service for the model you choose, and where it is processed depends on the model and vendor; models from Chinese vendors, for example, may be processed in mainland China. If your data must stay in a particular region, choose models accordingly or email us first. See the privacy policy.
Do you store my prompts and responses?
Our API logs record only the metadata needed for billing: time, model and token counts. Prompt and response content is not logged. Upstream services handle content under their own privacy policies.
Which upstream handles my request?
Each model may be served from one or more upstream supply pools, and we do not publish which one handles an individual request. What you choose is the tier on each API token: Standard and Official have the full catalogue and priority routing; Saver and Spot cost less, are best-effort, and may queue or fail. Prices and models per tier are on the pricing page.
Is there an SLA or uptime guarantee?
No. We do not offer an uptime commitment or service credits. When an upstream changes, individual models can become temporarily unavailable; the API then returns HTTP 503, and you can retry later or switch models. We periodically test every model in the catalogue and fix or remove the ones that stop working.
Are there rate limits?
Yes. The Saver tier is capped at 10 concurrent requests. When a limit is hit the API returns HTTP 429; retry after a short wait.
How does billing work?
Prepaid and pay-as-you-go, with a minimum top-up of 5 USD and no monthly fee. Per-model prices for every tier are listed on the pricing page, and the console shows your top-ups and per-request usage.
Can I get an invoice?
Not at the moment. We cannot issue VAT invoices or other tax invoices. The top-up and usage records in the console can serve as a record of spending.
Can I get a refund?
Unused top-up balance can be refunded to the original payment method; requests are usually reviewed within 3–7 business days. Usage already consumed and promotional credit are not refundable. If you dispute a charge, please contact us before filing a chargeback. See the refund policy.
Do I need a Chinese phone number?
No. Sign up with an email address; no mainland China phone number or bank card is required.
How do I contact you?
Email support@xhuoapi.ai. We usually reply within one business day. Refunds, chargeback questions and data-deletion requests go to the same address.