55-second avatar script: $10.12 standard to $30.25 max on Sume

A 55-second script costs $10.12 at standard (10.12), $13.475 at plus and $30.25 at max on Sume's Avatar 1.0 talking video (as of 2026-10-09). It is inside the 60-second maximum, with 5 seconds of margin for the estimator's error.

Why leave margin

Sume accepts a script or multi-scene plan when it estimates the video at 4 to 60 seconds inclusive. The estimate comes from the text, and speaking pace varies by voice and language. If your estimate is 59 seconds and the estimator says 61, the docs tell you to shorten the script or split it into jobs. A planned 55 seconds leaves room.

When the script is 70 seconds

Split it into two jobs at a sentence boundary, 35 seconds each, using one avatar and the same scene. Because pricing is linear in seconds, two 35-second jobs cost exactly what one 70-second job would: 70 x 0.245 = $17.15 at plus. Join the two clips with a timeline job, which is $0.10 per output minute, rounded up, so a 70-second join is $0.20. Total $17.35.

Inline captions are also capped: Sume rejects an estimated duration over 60 seconds when captions are on, so the split is needed for captioned videos too.

Counting words before you submit

A rough pace of 2.5 words per second gives 55 seconds for about 137 words. Treat that as a planning figure, not a rule: voices, languages and punctuation change the pace, and Sume's own estimator decides what is accepted. Write to 130 words, run a preview, and then render. If the estimate comes back above 60 seconds, remove a sentence and not a word at a time.

For Korean scripts, the caption style matters as well as the length: the docs say a Korean script with the slam, punch or tiktok-green styles is rejected with caption_hangul_text_latin_style, so choose a Hangul style when captions are on.

Preview a long script before you pay

For a 55-second render at max ($30.25) the first-frame preview is worth doing. The preview does not start the full render, and the quality you pick at generate-video applies only to the final video. See Avatar video previews for the four routes.

Whichever tier you pick, store the job id and the script together, so a second render of the same script at another tier can be compared against the first.

One more budget check: if the 55-second video is the final of a series of three episodes at plus, the series is 3 x 13.475 = $40.425 before any previews. Decide the tier for the whole series up front so the faces and pacing match across episodes.