The backend for NSFW creator platforms and white-label generators

What the platform layer needs

• Breadth under one key. Wan 2.7 (t2v / i2v / r2v / edit) and Wan 3.0 video, Seedance 2.5, the HappyHorse family (NSFW-permitted), Qwen-Image 3.0, Z-Image Turbo, Wan 2.7 image (edit, inpaint, grids). One contract, one auth header, one polling loop. Full list: /model.
• Metering you can bill on. Every task returns reserved_cost at submit and actual_cost at settlement, in USD decimal strings. Charge credits, tiers or pass-through — your call.
• Predictable supply. No model in the catalog "grows" a safety checker overnight; policy changes are announced 7 days ahead. Prepaid wallet in USD, topped up in crypto/stablecoins — the payment path has no card network in it (see why that matters on /compare/civitai).
• A policy you can quote. Your reviewers, payment partners and users will ask what is allowed. Link them to /content-policy and pass the same red lines through your terms.

Architecture sketch

1. Your API → ours. Keep your own auth and quotas; call POST /ent/v2/generate server-side with your key. Send a stable Idempotency-Key per user job so retries never double-charge.
2. Job queue → poll. Store task_id; poll GET /ent/v2/tasks/{id} with 2 s → 5 s back-off from a worker; states are pending → processing → success | failed.
3. Results → your storage. result_urls point at our CDN; copy to your bucket for long-term serving and your own moderation pass.
4. Uploads. User media must reach us as public HTTPS URLs — either from your storage or via our upload endpoint.
5. Pricing surface. GET /ent/v2/prices returns the effective price list for your key; mirror it in your admin so margins are visible.

Moderation and age assurance

Our input/output screening blocks the absolute prohibitions (minors in any style, real people, non-consent, bestiality). It is a backstop. As a platform you must implement age assurance appropriate to your jurisdictions, run your own moderation on what you serve, and give users a reporting path. Repeat flags on your key escalate to account review — so catching abuse first protects your supply.

Related

• All models · Pricing · pricing.json · API docs
• Content Policy · NSFW policy tracker
• 18models vs ModelsLab · 18models vs Replicate

Prices are USD base list prices from /pricing; the effective price for your key is returned by GET /ent/v2/prices. NSFW output is permitted under the Content Policy — fictional adults only, minors and real people prohibited and screened.

FAQ

Can I resell 18models generation inside my own product?

Yes. Outputs are yours to serve to your users under our Terms; you set your own pricing. You are responsible for age assurance and downstream moderation; our screening is a backstop.

How do I price generation for my users?

Every task returns reserved_cost and actual_cost in USD, so you can meter per user or per credit. Base prices: images from $0.017, video from $0.048/s; per-key effective prices via GET /ent/v2/prices.

Where are generated files stored?

Results are archived to our CDN and returned as result_urls; copy them to your own storage for long-term hosting. Media inputs must be public HTTPS URLs (or use our upload endpoint).

What happens if a user submits prohibited content?

Input screening rejects it before a paid request is made; output screening blocks prohibited results. Repeat abuse escalates to account review — build your own reporting path so you can act first.