text: prompt textimage_url: image referencevideo_url: video reference (triggers video-input pricing)audio_url: audio reference{
"type": "text",
"text": "The astronaut stands up and looks back at Earth",
"image_url": {
"url": "https://example.com/reference-photo.jpg"
},
"video_url": {
"url": "https://example.com/input-video-clip.mp4"
},
"audio_url": {
"url": "https://example.com/background-music.mp3"
},
"role": "first_frame"
}