Skip to main content

Image generation models

Openference exposes image models on POST /v1/images/generations. Stable Diffusion models (Cloudflare) use pixel size and optional diffusion controls. Each successful image consumes plan requests at the model’s ratio (see Model Request Ratios). Use the model ids below in your request body. Aliases are also accepted (e.g. sdxl-lightningSDXL Lightning).

SDXL Lightning (sdxl-lightning)

Task: text-to-image
Best for: fast previews and drafts (few diffusion steps)

SDXL Base 1.0 (sdxl-base)

Task: text-to-image
Best for: higher-quality still images (more steps than Lightning)

Errors

See POST /v1/images/generations and Common errors for the full list.