Ideogram 4.5 Precise Edit API: mask and reference limits
Ideogram 4.5 Precise Edit takes 4 reference images, 3 when a mask is sent. Sume lists Ideogram V3, and mask_url there is only for ChatGPT Image 2.5.
Ideogram's 4.5 Precise Edit endpoint takes up to 4 reference_images, but at most 3 when you also send a mask, because the mask uses one slot. Sume's image catalog lists Ideogram V3 (ideogram-v3), not 4.5, and its mask_url field is documented for ChatGPT Image 2.5 only.
Ideogram's limits are from its Precise Edit 4.5 reference page; Sume's from Image generation and editing and the image catalog code, read 2026-09-30.
What are Ideogram 4.5's edit limits?
The endpoint is POST /v2/image/precise-edit/ideogram-4-5, multipart only.
| Field | Rule |
|---|---|
image | Required, max 25MB, JPEG, PNG or WEBP; output keeps its width and height |
reference_images | Optional, max 4, 25MB each; never edited themselves |
mask | Optional; black is the area to edit, white is kept; same size as image; needs both black and white |
| Mask plus references | At most 3 references |
Does Sume offer Ideogram 4.5?
In the catalog code I read, the Ideogram entry is ideogram-v3, marked edit-capable. Check the live catalog before relying on any id; the 4.0 request shape is in Ideogram 4.0 API.
How many references and what about masks on Sume?
Edit-capable ids share a reference ceiling of 10 in the catalog code, advertised through the input_references descriptor. mask_url is described as an optional public HTTPS mask URL for ChatGPT Image 2.5 edits. Do not carry Ideogram's mask polarity or its 3-with-mask rule over; they are Ideogram's.
What happens if I send an unsupported field?
A request that sets a parameter the selected model does not list is rejected with 400 unsupported_parameter rather than silently dropped. Read the model's supported_parameters first.
Sources
Related posts
More in Models
- How many reference images? 10 on Image 1.0, 16 on GPT Image 2.5
Image 1.0 image_urls takes 1 to 10 public HTTPS URLs; ChatGPT Image 2.5 takes up to 16 references. Text-only models reject references.
- Image quality default on Sume: low on Image 1.0, high on GPT 2.5
Image 1.0 defaults quality to low; ChatGPT Image 2.5 on POST /v1/images defaults to high when omitted. Values, and when to raise quality.
- gpt-image-2 or openai/gpt-image-2? Model ids on Sume's Image API
Sume accepts bare Image Router ids such as gpt-image-2 and nano-banana-2 as aliases for openai/gpt-image-2 style ids. The catalog lists org/slug ids.
- Kling 2.1 API sunset: 410 Gone, and the kling-3 id on Sume
Krea retired Kling 1.0 through 2.1: endpoints return 410 Gone after Sep 15, 2026. Sume lists kling-3 in its video catalog; here is what to change.
Written by Sume