Ideogram 4.5 vs Ideogram V3 on the Sume image API
Ideogram V3 lists $0.075 with up to 10 references; Ideogram 4.5 lists $0.075 at medium quality with 5 references and a 1K or 2K resolution field.

Sume lists two Ideogram rows: Ideogram V3 (ideogram/ideogram-v3) and Ideogram 4.5 (ideogram/ideogram-v4.5). At the catalog's default each shows $0.075 per image, but they behave differently. V3 takes up to 10 reference images and has no resolution field. Ideogram 4.5 takes up to 5 references, adds resolution (1K or 2K), and its price depends on quality.
The facts below come from the catalog descriptors and the Image API page, read 2026-10-03.
Catalog comparison
Both rows list the same 15 aspect ratios, including the tall and wide 1:3 and 3:1, which no other Sume image model lists.
| Field | Ideogram V3 | Ideogram 4.5 |
|---|---|---|
| Catalog id | ideogram/ideogram-v3 | ideogram/ideogram-v4.5 |
| Listed price | $0.075 | $0.075 (default medium) |
quality | low, medium, high (maps to the provider's rendering speed) | low, medium, high (changes price) |
resolution | not listed | 1K, 2K |
input_references | 0 to 10 | 0 to 5 |
n | 1 to 4 | 1 to 4 |
output_format | png, jpeg, webp | empty list: the provider picks, omit the field |
How the two behave on edits
With Ideogram 4.5, a request with no input_references generates from text; with references the first image is edited and up to four more are used as references. An edit with no aspect_ratio keeps the source image's shape. The catalog caps Ideogram 4.5 at five references, so a sixth is an input error.
Ideogram V3 has no such split in the catalog: references go in input_references up to 10.
Which id to put in the request
If you need ten references, or you want the quality word to mean speed and not price, V3 is the row. If you want a 2K setting or the three-tier pricing, 4.5 is. Ideogram 4.5's low tier lists at $0.03 per image at the provider before Sume's margin, so a draft-then-final workflow on one id is possible without switching models. Run a short draft with quality: "low", then re-run the winner with quality: "high".
import os
import requests
resp = requests.post(
"https://api.sume.com/v1/images",
headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
json={
"model": "ideogram/ideogram-v4.5",
"prompt": "poster that reads GRAND OPENING in bold slab letters",
"quality": "low",
"resolution": "1K",
},
timeout=60,
)
resp.raise_for_status()
if resp.status_code == 202:
print("still running:", resp.json()["data"]["status_url"])
else:
for image in resp.json()["data"]:
print(image["url"])Sources
Related posts
More in Comparisons
- Instagram's Devanagari caption fonts: what Sume burns for Hindi
Instagram added Devanagari and Bengali-Assamese caption fonts. Sume documents Latin and Hangul faces only, so test Hindi cues on one clip first.
- Reels run to 20 minutes; Sume's audio and caption limits to plan for
Instagram says Reels can reach 20 minutes but over 3 minutes is not recommended. The Sume Timeline audio, avatar and caption limits that matter at that length.
- Story video Restyle lists presets only; edit by prompt with Sume
Meta's Restyle for Story video lists a preset effect and no prompt. Sume's Video Router edit mode takes a prompt and a source clip. What each takes and costs.
- Story Layout holds 6 photos; Sume compose holds a still and a clip
Instagram's Layout splits a Story frame into up to 6 photos. Sume compose does less: one still and one video, stacked or overlaid, returned as one MP4.
Written by Sume