GPT Image aspect ratios: Firefly's 10 vs Sume's 17

Adobe's September 2026 Firefly update says GPT Image 2 now supports 10 aspect ratios on web and Boards, up from 3. Sume's image documentation lists 17 normalized ratios plus auto, and a model-specific image_size that accepts custom pixel sizes for ChatGPT Image 2.5.

The Firefly change

The Adobe Community post gives the count, not the list, so I do not name which ten ratios Firefly offers.

The ratios Sume documents

The Image API reference lists these normalized values for aspect_ratio: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, 1:2, 2:1, 1:4, 4:1, 1:8, 8:1, 9:21 and 21:9. That is 17. Use auto to let the provider choose, which the docs recommend on edit and image-to-image calls so the result matches the source shape.

The 17 are the normalized vocabulary, not a promise that every model accepts every one. Each model's supported_parameters.aspect_ratio lists its own values, so read GET /v1/images/models for the model you plan to call.

Custom pixels on ChatGPT Image 2.5

For openai/gpt-image-2.5 and openai/gpt-image-2.5-sunburst, image_size accepts named presets, auto or custom pixels. Both edges must be multiples of 16, the longest edge is at most 3840, the aspect ratio is at most 3:1, and the total is between 655,360 and 8,294,400 pixels.

Do not put custom pixels in size. The docs reserve size for resolution tiers, and explicit pixel size there is rejected in v1.

Choosing between ratios and pixels

• Use a ratio when the platform cares about shape, such as 9:16 for vertical video covers.
• Use image_size when you need an exact deliverable, such as 1080 by 1350, and the model supports it.
• Check the catalog first so a rejected parameter does not cost a retry.