Ideogram 4.5 headline variants from one ad: 4 edits for about $0.30
Four headline edits of one approved ad on ideogram/ideogram-v4.5 at medium quality cost about $0.30 by Sume's catalog line. Python that runs them in parallel.

Four headline variants of one approved ad cost about $0.30 at medium quality on Sume's Ideogram 4.5 row. That comes from the catalog line: the Fal list price for medium is $0.06 per image, and the docs and code price it at list times 1.25, which is $0.075 per image, so four images is $0.30. Read usage.cost on each response for the real figure, since the docs treat that as the billed amount.
The point of Ideogram 4.5 for this job is that an edit without aspect_ratio keeps the source geometry, so the layout stays put while the words change.
Run four edits at once
Each variant starts from the same source, so the edits do not depend on each other and can run in parallel. The source must be a public HTTPS URL. With references, the first image is the one that gets edited.
import asyncio, os, requests
KEY = os.environ.get("SUME_API_KEY")
SRC = os.environ.get("AD_URL")
if not KEY or not SRC:
raise SystemExit("set SUME_API_KEY and AD_URL")
HEADS = ["Spring sale", "Two days left", "Free shipping", "New arrivals"]
def edit(i: int, text: str) -> str:
body = {"model": "ideogram/ideogram-v4.5", "quality": "medium",
"prompt": f'Change the headline to read "{text}". Keep everything else.',
"input_references": [{"type": "image_url", "image_url": {"url": SRC}}]}
h = {"Authorization": f"Bearer {KEY}", "Idempotency-Key": f"ad-head-{i}"}
r = requests.post("https://api.sume.com/v1/images", headers=h, json=body, timeout=60)
if r.status_code != 200:
return f"{r.status_code} {r.text[:120]}"
d = r.json()
return f"{d['data'][0]['url']} cost={d.get('usage', {}).get('cost')}"
async def main() -> None:
out = await asyncio.gather(*(asyncio.to_thread(edit, i, t) for i, t in enumerate(HEADS)))
print(*out, sep="\n")
asyncio.run(main())Cost per quality, from the catalog
| Quality | Fal list price per image | Billed per image (list x 1.25) | Four edits |
|---|---|---|---|
| low | $0.03 | $0.0375 | $0.15 |
| medium (default) | $0.06 | $0.075 | $0.30 |
| high | $0.22 | $0.275 | $1.10 |
What to watch
- A
202means the edit outran the 30-second wait. Poll thestatus_urlinstead of resending. - Failed or cancelled generations are not billed. Completed ones are, even if you do not like the result.
- Keep the instruction short and name only the headline. Longer instructions give the model more room to change other things.
- Save each output yourself. Do not feed a variant back as a new source unless you want to stack edits.
Sources
Related posts
More in Models
- Ideogram 4.5 low, medium or high: a text grid to pick the tier
Run one headline through Ideogram 4.5 at low, medium and high on Sume, read the billed cost of each, and keep the cheapest tier whose text you can read.
- Ideogram 4.5 vs Nano Banana 2 for multi-turn edits: cost per turn
Ideogram 4.5 claims clean multi-turn edits. Compare it with Nano Banana 2 on Sume: price per turn, six-turn totals and a drift test you can run.
- Kling 3.0 15-second clip: $2.10 silent or $3.15 with sound on Sume
Kling 3 on Sume makes 4 to 15 second clips at 720p or 1080p, with audio priced at $0.21 a second and silence at $0.14. Limits, ratios and how to try it.
- Longest single AI video clip by API in 2026: 30 seconds, two Sume ids
Seedance 2.5 and Wan 3.0 make a 30-second clip in one job on Sume. Kling 3 and MiniMax H3 stop at 15 s and Omni at 10 s. Limits and per-second prices.
Written by Sume