Terms of Service

WanAPIs (the "Service") is an API gateway that aggregates large language models (LLMs). It exposes an OpenAI-compatible interface and routes requests to upstream providers including OpenAI, Anthropic, Google and DeepSeek.

The Service is operated independently and is not affiliated with any of those upstream providers. Some models are reached via OAuth or enterprise subscriptions, so availability may fluctuate when an upstream provider changes its policies.

You reach the API at api.wanapis.com, and the marketing site and console entry point at wanapis.com.

One email address corresponds to one account. Fraudulent registration and impersonation are prohibited.

Your API key is the sole credential for calling the Service and should be treated like a password. You bear any loss arising from a leaked API key.

We recommend enabling safeguards such as IP restrictions and spending caps to limit the impact of a leaked key.

If we detect abnormal activity on your account — including but not limited to unauthorised use, credential stuffing or brute-force attempts — we may suspend the account immediately and ask you to cooperate with our investigation.

You agree not to use the Service to:

(1) engage in any activity that violates the laws of the People's Republic of China or of your own jurisdiction;

(2) breach the terms of service of any upstream LLM provider, including training competing models, model distillation, or circumventing abuse detection;

(3) generate or distribute prohibited content, including sexual, violent, terrorist, discriminatory or politically sensitive material;

(4) carry out network attacks, credential stuffing, scanning, exploitation, spam or fraud;

(5) use scripts to bypass the Service's rate limits, billing or anti-abuse mechanisms;

(6) resell or sublicense API keys, or sell the Service to third parties, without our written authorisation.

We reserve the right to suspend or terminate accounts that breach these rules, without prior notice.

The Service is billed by tokens consumed. Unit prices vary by model; see the pricing page at docs.wanapis.com.

Because upstream pricing can change, we reserve the right to adjust our unit prices accordingly. Significant changes are announced by email 7 days in advance.

New users receive starter credit worth about USD 0.20 for testing. This credit cannot be withdrawn or transferred, and is not reissued once expired.

Top-ups are made through supported payment channels and are usable as soon as they settle.

Your balance may only be spent on API usage within the Service and does not constitute a debt instrument.

You may request a refund under the following conditions:

(1) within 7 calendar days of your first top-up, provided no credit has been consumed, you may request a full refund;

(2) once any credit has been consumed, that top-up is non-refundable;

(3) balances on accounts suspended or terminated for breaching these Terms are not refundable;

(4) unintended charges caused by a system fault are compensated with equivalent credit once verified.

To request a refund, email [email protected] with "Refund request" in the subject line, together with your registered email address and order number. We process requests within 7 business days.

Refunds are returned to the original payment account; settlement timing depends on the payment channel.

We use commercially reasonable efforts to keep the Service available, but we do not commit to a 99.9% SLA.

Interruptions caused by the following are not treated as our breach, and we accept no liability for them:

(1) upstream LLM provider outages, rate limiting or policy changes;

(2) failures of carriers, data centres, CDNs or other infrastructure;

(3) force majeure, including natural disasters, war and governmental action;

(4) necessary planned maintenance, for which we give advance email notice;

(5) problems in your own network environment or integration code.

Model output is generated by upstream models; the Service only forwards it. We make no express or implied warranty as to the accuracy, legality, completeness, fitness or harmlessness of that output.

Any decision or action you take based on model output, and any direct or indirect consequence of it, is your own responsibility.

To the maximum extent permitted by applicable law, and regardless of the legal theory, our aggregate liability to you shall not exceed the total fees you actually paid to the Service in the preceding 12 months.

We are not liable for indirect loss, lost profits, lost data, loss of goodwill or punitive damages, even if we were advised such loss was possible.

You may close your account at any time from the console. All of your data — except records we must retain by law — is deleted within 30 days, and any balance is handled under the refund policy.

We may terminate your account immediately if:

(1) you materially breach these Terms;

(2) abuse is detected on your account;

(3) required by law or ordered by a judicial or administrative authority.

After termination you can no longer access the Service, but the clauses on fees, limitation of liability and dispute resolution survive.

We may revise these Terms at any time. Minor wording changes are made without separate notice.

Material changes — to the fee structure, the scope of liability or the scope of the Service — are notified 30 days in advance to the email address you registered with.

Before a change takes effect you may decline it and stop using the Service. Continued use constitutes acceptance of the revised Terms.

The formation, validity, interpretation, performance and dispute resolution of these Terms are governed by the laws of the People's Republic of China (excluding Hong Kong, Macau and Taiwan).

Any dispute arising out of or relating to these Terms shall first be resolved through good-faith negotiation. Failing that, either party may bring proceedings before the competent People's Court at the operator's place of business.

If any clause is held invalid or unenforceable, the remaining clauses continue in full force.

If you have any question about these Terms, contact [email protected].