OpenAPI Image API

Image API

This page explains the API from the image perspective.

Images are the lightest content workflow: the call chain is the same, but the parameter set is smaller and the result is more direct.

Create

POST /creations

Minimum shared fields:

• type=image
• prompt
• language
• ratio and quality fields returned by /catalog

Nano Banana 2

When your catalog selection points to Nano Banana 2, the current frontend maps it to the Gemini image route:

• image_generation.provider=gemini
• image_generation.model=nano-banana-2
• extra_config.output.aspect_ratio supports 16:9 / 9:16 / 1:1
• extra_config.output.quality still uses 1K / 2K / 4K

GPT-Image-2

When your catalog selection points to GPT-Image-2, the current frontend maps it to the Azure OpenAI image route:

• image_generation.provider=azure_openai
• image_generation.model=gpt-image-2
• extra_config.output.aspect_ratio supports 3:2 / 2:3 / 1:1
• image_generation.quality supports low / medium / high
• The frontend display names are now Basic / Standard / Premium
• Current frontend credit display: low=5, medium=8, high=20
• The frontend no longer shows launch or discount badges
• Premium / high is gated like Nano Banana 2 4K; non-members see the member icon and upgrade flow

Manage

Image workflows usually care about:

• whether generation succeeded
• the main image URL
• the thumbnail
• any additional image artifacts

Edit

Image workflows usually do not expose complex scene-level editing first.

More common options are:

• regenerate with a new prompt
• regenerate with a new model
• regenerate with a new ratio or quality

Practical Guidance

• use the same public POST /creations flow as other content types; do not route image to a separate legacy endpoint
• do not reuse video-specific duration or voice controls
• image is now model-aware, so do not assume one shared ratio/quality matrix across all image models
• the image create surface no longer exposes style as an explicit control
• return a directly renderable image URL first