Content policy
What the platform allows, what needs enabling, and what is forbidden unconditionally.
Where we stand
Most hosted models filter heavily at the prompt layer, and a lot of legitimate creative work — figure studies, adult narrative, dark subject matter — gets caught in the crossfire.
SpicyAPI offers a wider selection of model capability: part of the catalogue is marked mature-capable. That does not mean unlimited. It means the boundary is drawn at law and platform responsibility rather than at a conservative default filter.
Three layers
Absolute limits: never permitted
No account tier, no region, no setting relaxes these:
- Sexualised content involving minors — including fictional characters that appear to be minors
- Sexualised depictions of real people without their consent
- Non-consensual or violent harm presented sexually
- Sexual content involving real animals
Requests violating these are rejected and recorded. The first item is reported to the relevant authorities as required by law and permanently terminates the account, with the material preserved separately under the applicable statutory obligation.
Requires enabling: mature-capable models
When a request asks for mature mode ("mature": true), every check runs server-side and any one failing rejects the request:
| Check | Failure |
|---|---|
| The model itself supports it | 400, "this model does not support that mode" |
| This key has not disabled mature capability | 40305 |
`mature` requests; it never relaxes
Sending true from the client only states intent — it cannot relax any check. Every decision is made server-side; nothing you send changes the outcome.
Nor do we silently downgrade to standard mode when the checks fail, because that would leave you with a result you did not ask for and no explanation of why.
Registered accounts can access mature-capable models by default, and the platform currently imposes no additional third-party age gate. That product choice does not alter the legal boundary: mature use is limited to consenting adults and lawful purposes. Integrators must identify and implement any age-assurance, consent-record, real-person and minor-protection duties that apply in their jurisdictions.
Standard models: no extra gate
Models not marked mature-capable follow their own upstream policy. No enabling step, and no need to send mature.
The per-key switch is separate
New keys allow mature requests by default, but you can disable that capability per key. Once disabled, the server rejects mature: true even if the request asks for it.
That separation lets you distinguish an internal key from one handed to an outside partner: if the latter leaks, its capability surface is still limited.
Where we operate
We do not serve mainland China. This is a legal compliance requirement, not a product decision. Hong Kong, Macau and Taiwan are separate jurisdictions and are not covered by it.
Other restricted regions come from a mix of law and upstream supplier contracts — some model providers explicitly prohibit serving particular territories in their own terms. The authoritative list is in the terms of service.
Requests from a restricted region return 40303. The check is enforced at the edge, at the origin, at sign-up and at payment.
Upstream models refuse too
Even with your account enabled and nothing near the absolute limits, the specific upstream model may still refuse under its own policy. Each model's catalogue page states where it sits.
You will see this as a task reaching failed with the funds already released. A different prompt may pass; retrying the same content will not.
Your responsibilities
You are responsible for how content generated through the API is distributed and used. If your product serves end users:
- You must restrict access to adults who may lawfully access the content and implement age assurance or verification wherever local law requires it.
- Real-person uses require the necessary consent; sexualised content involving minors or minor-appearing people remains prohibited.
- You must comply with the law where you are and where your users are.
- The capability boundary we offer is not the legal boundary in your jurisdiction.
This page describes the platform's technical and policy boundaries and is not legal advice. Your obligations depend on your business and location; assess them yourself or consult a lawyer.
Reporting
To report content that violates the above, email abuse@spicyapi.ai with the task ID or the artefact's object key. Reports involving minors are prioritised.