VEED Lip Sync 2.0 at $0.07 a second vs Sume still plus audio

VEED lists Lip Sync 2.0 at $0.07 per second and describes it as video-to-video lip synchronization, so you give it an existing video. Sume's MiniMax H3 Max lip sync takes a still image plus audio and makes a talking clip, at the fal list rate times 1.25: $0.0625, $0.10 or $0.20 per second at 480p, 768p and 1080p. If you need to re-sync a recorded video, Sume's route does not do that.

What each product starts from

VEED's page names Fabric 1.0 for animating an image with audio, and Lip Sync 2.0 for video-to-video. Sume's docs describe its H3 Max route as the public still-plus-audio talking-clip surface, with audio between 5 and 14.8 seconds, Sume-hosted and up to 10 MB, and duration_seconds required in that range.

Reading the Sume numbers

The Sume page prices the clip as ceil(duration_seconds) times the fal list for each resolution, times 1.25, and its own example is a 5-second clip at 768p reserving $0.50. The fal figures in it carry a 2026-09-22 date, so the live catalog is the authority.

Steps: animate a still with a recorded line

• Host your audio on Sume and keep it between 5 and 14.8 seconds.
• POST to /v1/minimax/h3-max/lip-sync with image_url, audio_url, duration_seconds and resolution.
• Send an Idempotency-Key and poll the job.
• If the audio is longer, cut it first; the API refuses duration values outside 5 to 14.8 instead of clamping.

curl -X POST https://api.sume.com/v1/minimax/h3-max/lip-sync \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: lipsync-001" \
  -d '{"image_url":"https://example.com/face.png","audio_url":"https://media.sume.com/artifacts/artf_demo/line.mp3","duration_seconds":5.84,"resolution":"768p"}'

What Sume does not do

Sume does not re-sync lips on an existing video through this route. If you have footage and new dialogue, VEED's video-to-video product is the closer match, and I found no equivalent in the Sume docs I read.

Audio length is the practical limit

The 5 to 14.8 second window is the constraint that will bite first on Sume. Longer dialogue has to be split into windows of that size, each a separate job, and the clips joined afterward in Timeline at $0.10 per rounded output minute. A 30-second speech is at least three jobs. At 768p and $0.10 a second that is about $3.00 plus the join. Compare that with VEED's $2.10 for 30 seconds of video-to-video at $0.07, if you already have the footage.

Resolution choices

Sume's route accepts 480p, 768p (the default) and 1080p, with no 2K, and the speed_tier field is accepted and ignored for Fabric body compatibility. VEED's Fabric lists a range from $0.08 to $0.20 a second depending on resolution and speed tier. If you only need a preview, 480p on Sume at $0.0625 a second is the cheapest clip in the table; for delivery, check that 768p is sharp enough for your placement.

Verdict

Existing footage: VEED. A talking clip from a photo: either; Sume's 480p price is lowest, VEED's Fabric lists a range, and VEED's Fabric lists a range up to $0.20 a second. Sume's route only accepts audio from 5 to 14.8 seconds.