Developer / API

Image-to-text (vision)Describe or read an image with a vision model: submit mode=text, a vision model id, and the image URL as init_image.
Video engines (face-swap, talking heads, character swap)Face-swap, talking-head, reenactment, and whole-person character-swap engines - what each needs, and how to call them.
Generating audio (music and sound)POST /v1/generate with mode=audio: prompt versus lyrics, the params object, and licence gating.