Sync.so $0.05 a second vs Sume H3 Max lip sync: rates side by side

Sync.so's API plans list $0.05 a second on Hobbyist and Creator, $0.0475 on Growth and $0.04 on Scale, each with a monthly fee. Sume's MiniMax H3 Max lip sync has no monthly fee and derives to $0.0625 a second at 480p, $0.10 at 768p and $0.20 at 1080p, rounded up to whole seconds. The two are not the same job: Sync.so re-syncs the mouth in a video you already have, while Sume's route animates a still with your audio.

Sync.so figures are from its pricing page and docs, read 2026-10-06. Sume's come from the lip-sync and models docs, read the same day: the list per second is $0.05, $0.08 and $0.16 at the three resolutions, multiplied by Sume's 1.25 margin.

What do the published rates add up to?

The table prices a single 10-second result at each published rate. Sync.so's monthly fee is shown separately because it does not scale with the clip.

Why the cheaper number is not automatically the better buy

Inputs differ. Sync.so's docs list MP4 video plus WAV or MP3 audio, delivered by public URL, upload or asset id, with lengths up to 30 minutes depending on plan. Sume's route takes either a public HTTPS still (aspect ratio 0.4 to 2.5) or a ready avatar, and the audio must be Sume-hosted, no larger than 10 MB, between 5 and 14.8 seconds. If you start from footage of a real speaker whose words changed, you need a video re-sync. If you start from a photo or a generated avatar, you need a still-plus-audio route.

Length differs too. Sume refuses a duration_seconds outside 5 to 14.8 instead of clamping it, so a clip never ends shorter than the audio without telling you. A longer line becomes several jobs; the 20-second line post shows the split.

How does billing behave when a job fails?

Sume reserves the amount at admission, captures it on completion, and refunds it on failure (lip-sync docs on models, read 2026-10-06). The refund post covers the detail. Sync.so's page lists a free trial of three generations a month, up to 20 seconds each, per its docs, which is a fair way to test its output before paying.

Concurrency and queueing also differ. Sync.so lists 1, 3, 6 and 15 concurrent jobs across its four plans. Sume admits more jobs than it can run at once and holds them as queued, with processing concurrency set by plan, so a burst of submits is not a failure (Generation admission, read 2026-10-06). For a launch week with many short clips, that behavior matters as much as the second-rate.

The decision rule is short. If you hold footage and need a different language or script on the mouth, price Sync.so on its plan. If you hold a still or an avatar and a short voice line, price Sume per second at the resolution you will publish. If your volume is large enough that Scale's $0.04 matters, compare after counting the $249 fee, and compare the output at the same resolution on a real sample, not on the rate card.

One more check belongs on the list: output frame rate and file handling. Sync.so states its prices at 25 fps, and Sume's lip-sync doc says to measure the first real H3 Max clip before you lock a timeline frame rate, so a mixed edit should confirm both before assembly. Keep the same audio file for both samples so the only variable is the model.