Ideogram V3 is $0.075 at every quality on Sume; 4.5 low is half

Ideogram V3 on Sume costs $0.075 per image at low, medium and high quality, because its pricing is a single static line, while Ideogram 4.5 costs $0.0375 at low, $0.075 at medium and $0.275 at high. So sending quality: low to V3 saves nothing, and moving to 4.5 low halves the price of a draft.

The Sume figures come from the pricing code and catalog read on 2026-10-10. The fal Ideogram V4.5 page, read the same day, lists $0.03, $0.06 and $0.22 for low, medium and high. Multiplying by Sume's 1.25 margin gives the billed ladder, as the Image API docs describe.

The ladder

V3 lists a quality field with low, medium and high, but its price does not move. The V3 list rate in the code is a static $0.06, billed $0.075. The 4.5 row prices by quality, so the field changes the charge.

What that means for a draft run

A hundred drafts cost $7.50 on V3 at any quality. On 4.5 at low they cost $3.75, and at medium $7.50. A hundred finals at high cost $27.50 on 4.5, and $7.50 on V3. So V3 is cheaper if you always want the top setting and 4.5 is cheaper if you draft low.

A common two-stage plan is a hundred low drafts at $3.75 and ten high finals at $2.75, a total of $6.50 for the set on 4.5. The same ten finals on V3 are $0.75, but V3 has no cheaper draft tier, so the hundred drafts cost $7.50.

Other differences to check

V3 lists 10 references and 15 ratios and no resolution field. 4.5 lists up to 5 references, 15 ratios and a resolution of 1K or 2K. 4.5 lists no output formats, so omit output_format. Neither row is a quality claim: it is a contract comparison, and the images are yours to judge.

• Use V3 when you want one predictable price.
• Use 4.5 low for cheap drafts and high for finals.
• Omit output_format on 4.5.
• Check the endpoints route for the live price before a big run.

Why the field is still accepted on V3

The V3 descriptor lists quality because the underlying model supports the setting, and Sume forwards it. The billing line is a separate thing: it is a single static amount. That is why a request with quality: low is valid and still costs $0.075. If you build a cost estimator, key it off the endpoints route and not off the presence of a quality field.

A frequent mistake is to assume any row with a quality field has a ladder. Of the rows in the catalog that list quality, GPT Image 2, GPT Image 2.5 and Ideogram 4.5 price by it, and Ideogram V3 does not.

Choosing between the two rows

Pick Ideogram V3 when you want a single price and up to 10 references. Pick Ideogram 4.5 when you want a cheap low-quality draft tier, a 1K or 2K resolution field, or a higher ceiling at the top setting and can live with up to 5 references. Both list 15 aspect ratios, so a ratio is not the deciding factor.

Check the output yourself. Ideogram's own page for the 4.5 model, read on 2026-10-10, shows text rendering in its example, but the page is a vendor claim, and your headline is the test.

Whichever row you choose, write down the date and the price you read. A pricing line that is right today can move, and a note with a date lets the next person see whether a surprise in the wallet came from usage or from a price change. Repeat the check each month, since it takes a minute with the endpoints route.

For a team, put the ladder in a shared table so nobody has to guess the cost of a quality setting. Add a line for the date you read it and the source route.