AI Image
AI Image uses type: image. Select one complete model bundle from GET /capabilities under image.models.
Nano Banana 2
{ "type": "image", "prompt": "A clean binary-search course cover with a dark technical style", "input_file_ids": [], "image": { "provider": "gemini", "model": "nano-banana-2", "aspect_ratio": "16:9", "output_quality": "2K", "style_id": "style_from_capabilities" }}
GPT Image 2
{ "type": "image", "prompt": "A clear binary-search poster for a developer course", "image": { "provider": "azure_openai", "model": "gpt-image-2", "aspect_ratio": "3:2", "output_quality": "2K", "quality": "medium" }}
Parameter Rules
- provider and model must come from the same advertised bundle.
- aspect_ratio and output_quality must be supported by that bundle.
- quality is a GPT Image native setting and is valid only for azure_openai + gpt-image-2.
- Choose style_id from current capabilities.
- All three products accept input_file_ids; upload references first.
- AI Image does not currently advertise URL Context. Do not claim that it read a webpage from a URL in prompt.
Describe the subject, intended use, composition, audience, and visual direction. Manually review exact typography, figures, trademarks, and regulated labels before publishing. Poll the creation and read outputs only after it is ready.
