OffRail API
Build NSFW-capable image, video, chat, and audio generation with one managed API, one key, and usage-based billing.
OffRail is a managed API gateway for NSFW-capable generation. It gives your application one authentication method and a consistent set of routes for image, video, chat, and audio models while keeping upstream access, routing, metering, and task delivery behind the API boundary.
Private preview — Self-service account access and API key provisioning are not open yet. These docs describe the API contract for developers who receive a preview key.
One public API surface
| Task | Route | Delivery | Billing unit |
|---|---|---|---|
| Chat | /v1/chat/completions | JSON or stream | Input/output tokens |
| Image | /v1/images/generations | Result or task | Generated image |
| Video | /v1/videos | Asynchronous task | Generated second |
| Audio | /v1/audio/speech | Binary audio payload | Input characters |
Choose a model from the live public catalogue, send its ID to the matching route, and use the same OffRail API key as you move between modalities.
What OffRail manages
- Authentication — one OffRail API key across public generation routes.
- Routing — eligible upstream selection remains behind the public model ID.
- Fallback — supported routes can move to another eligible mapping when an upstream fails.
- Metering — each response or task is measured using its modality-specific billing unit.
- Delivery — synchronous responses, streams, binary audio, and asynchronous video tasks follow documented contracts.
Request lifecycle
- Select a public model and its API route.
- Authenticate with
Authorization: Bearer $OFFRAIL_API_KEY. - Send the request using the documented schema.
- Receive the result immediately or track the asynchronous task.
Start building
- Quickstart — configure a preview key and send your first request.
- Overview — understand routing, delivery, and billing across modalities.
- Image generation — call the image generation route.
- Video generation — create and track asynchronous video tasks.
- Speech generation — generate audio from text.
- API reference — inspect endpoint schemas and responses.