Ideogram 4.5 API cost: $0.0375 to $0.275 per image on Sume

Ideogram 4.5 on Sume (ideogram/ideogram-v4.5) is priced by the quality parameter. The Sume docs give the list as $0.03, $0.06 and $0.22 per image for low, medium and high, and images are billed at list x 1.25: $0.0375, $0.075 and $0.275. Per 100 images that is $3.75, $7.50 and $27.50.

Quality sets the price, size does not

The docs say the price is the same for all sizes; resolution 1K or 2K does not change it. If you omit quality, it defaults to medium. Images are returned with a usage.cost field equal to the billed amount, so you can check any request against this table.

Draft then upgrade

Quality is a 7.3x price spread from low to high ($0.275 / $0.0375). Render 100 candidates at low for $3.75, pick the best 10, and re-render those at high for 10 x $0.275 = $2.75. The total is $6.50, against $27.50 for 100 images all at high, a saving of 76 percent (21 / 27.5). Note the re-render is a new generation and will not reproduce the draft exactly, so this suits concepting, not pixel-matching.

A comparison point and its limits

This post does not compare other models. When you compare image APIs, check how each one prices quality tiers, whether size changes the price, and whether edits with a reference image cost the same as text-only generation.

Edits and references

With input_references, the first reference is the image being edited and up to four more are references. The docs give one per-quality list price for the model; check GET /v1/images/models/ideogram/ideogram-v4.5/endpoints for the current figure before you scale a batch.

A 1,000-image catalogue

A shop with 1,000 product images could render at medium for 1,000 x $0.075 = $75.00. Rendering everything at high is $275.00. A tiered plan (800 at low, 150 at medium, 50 at high) is 800 x $0.0375 + 150 x $0.075 + 50 x $0.275 = $30.00 + $11.25 + $13.75 = $55.00.

That tiered plan is 80 percent cheaper than all-high and 27 percent cheaper than all-medium ($75 to $55). The decision is which images deserve high; it is the same call a photographer makes when choosing which frames to retouch.