Ideogram 4.5 carousel text slides: what 5 and 10 slides cost

A five-slide text carousel on Ideogram 4.5 costs $0.15, $0.30 or $1.10 at the list prices, depending on whether you pick low, medium or high quality, and a ten-slide carousel costs exactly double. Those figures come from the Fal list price that the Sume image docs quote for ideogram/ideogram-v4.5: $0.03, $0.06 or $0.22 per image by quality, whatever the size. It is a provider list price, so read the live number for your workspace from GET /v1/images/models before you budget a campaign.

I did not find a current Instagram page that states a slide limit I could cite, so the 5 and 10 below are planning numbers, not an Instagram rule.

The math, slide by slide

A mixed plan is usually the sensible one: ten slides at low is $0.30, then re-render the two slides where letters went wrong at high for $0.44 more. That totals $0.74, against $2.20 for ten at high. Failed generations are not billed, per the image docs, so a refused call does not add to the total.

Ask the endpoint for its ratios, then call it

The images docs say to check an endpoint's supported_parameters before pinning a tier or ratio, and that 4:5 is the Instagram portrait ratio (1080 by 1350), not 4:3. The script below reads the model's endpoint listing, prints what it says, then sends one 4:5 slide at low quality. Run it once on a single slide before you loop over ten.

The response in the docs example returns a data list with an image url per item; if your call comes back as a queued job instead, the envelope carries a request_id you poll the usual way.

import json, os, urllib.request

API = "https://api.sume.com/v1"

def post(path, body, key):
    req = urllib.request.Request(
        f"{API}{path}",
        data=json.dumps(body).encode(),
        headers={
            "Authorization": f"Bearer {os.environ['SUME_API_KEY']}",
            "Content-Type": "application/json",
            "Idempotency-Key": key,
        },
        method="POST",
    )
    with urllib.request.urlopen(req) as res:
        return json.load(res)

def get(path):
    req = urllib.request.Request(
        f"{API}{path}",
        headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
    )
    with urllib.request.urlopen(req) as res:
        return json.load(res)

MODEL = "ideogram/ideogram-v4.5"

endpoints = get(f"/images/models/{MODEL}/endpoints")
print(json.dumps(endpoints, indent=2)[:600])

res = post(
    "/images",
    {
        "model": MODEL,
        "prompt": "Carousel slide 1 of 5, bold sans-serif headline: 3 hooks to test",
        "aspect_ratio": "4:5",
        "quality": "low",
        "resolution": "1K",
    },
    "carousel-slide-1-v1",
)
for image in res.get("data", []):
    print(image["url"])
print(res.get("request_id"))

Keep a text check in the loop

Text slides are where image models still slip, so budget a review step rather than assuming the first pass is right. Read every slide at phone size, compare the letters against your intended copy, and regenerate only the slides that fail. Because the idempotency key identifies one operation and payload, change the prompt or the key when you regenerate, or you will be handed the earlier result back.

If a slide needs a small fix rather than a redo, Ideogram 4.5 also accepts input_references: it edits the first image and uses up to four more as references (five in total), and an edit without aspect_ratio keeps the source shape. That route is priced the same way, per image by quality, so a fix costs another slide's worth.

Sume does not publish to Instagram. The output is a set of hosted 4:5 images you upload yourself, in the order you intend. Check Instagram's current help for how many slides your account may add, and whether per-slide captions are available to you, before you plan the sequence.

Planning a campaign budget

Multiply before you commit. A weekly carousel of ten slides at medium is $0.60 at list price, so four a month is $2.40 and fifty-two a year is $31.20; the same plan at high is $114.40. Those totals assume one generation per slide with no retries, which is optimistic for text-heavy images.

A more honest number includes retries. If one slide in three needs a second pass at the same quality, ten slides cost the equivalent of about thirteen. Treat that ratio as an assumption to replace with your own after the first month, because it depends on how much copy sits on each slide and how unusual the words are.

The cheapest habit is to write the slide copy first, keep it short, and let the model draw only the headline. Long paragraphs inside an image are the usual cause of retries, and body text is better added in the caption, where you can edit it after publishing.