Ideogram 4.5 high costs 7x low: when it pays

Ideogram 4.5 (ideogram/ideogram-v4.5) lists at $0.03 for low, $0.06 for medium and $0.22 for high quality per image, so high is about 7.3 times low. The Sume docs say the price follows quality and is the same at 1K or 2K, which makes quality, not size, the knob that moves your bill.

List prices from the docs

The Sume Image API docs cite the Fal list price for these three tiers. Treat the figures as list prices and read the live price for your account from the model's endpoint record at GET /v1/images/models/{model_id}/endpoints.

How the request works

Without input_references the model generates from text. With them it edits the first image and uses up to 4 more as references, 5 in total. Quality is low|medium|high, and an omitted value defaults to medium. resolution is 1K or 2K. An edit without aspect_ratio keeps the source image's shape.

curl -X POST "https://api.sume.com/v1/images" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "ideogram/ideogram-v4.5",
    "prompt": "A poster that reads OPEN LATE in bold block letters",
    "quality": "low",
    "resolution": "1K"
  }'

A plan for when to pay $0.22

• Explore at low: 100 candidate ideas cost $3.00 at list.
• Promote the few that work to medium or high only when you will publish them.
• Because 2K costs the same as 1K at a given quality, choose resolution by where the image will run, not by price.
• Keep high for finals where fine text or detail has to hold up; there is no documented rule that it always reads better, so compare on your own prompts.

Worked example

Ten finals at high cost $2.20 at list. Doing 100 drafts at low first adds $3.00, so a draft-then-final run is $5.20 for ten publishable images, against $22.00 if every one of 100 drafts were rendered at high.