Luma Ray 3.2 video-to-video vs Sume H3 Max Recast: 10 s cost
Luma lists 10 s Ray 3.2 video-to-video at $2.16 (720p) and $4.32 (1080p). Sume's H3 Max Recast is $3.75 (768p) and $5.625 (1080p), per source second.

On Luma's API, a 10 second Ray 3.2 video-to-video job is $1.08 at 540p, $2.16 at 720p and $4.32 at 1080p, while Sume's h3-max-recast is $3.75 at 768p and $5.625 at 1080p for 10 source seconds. Luma is cheaper per clip on both tiers, but the two products do different things.
Recast, per the Video Router docs, swaps the people in a source video_url for one to four reference photos, one per person, at 768p or 1080p, with duration equal to the source length (5 to 30 seconds). Luma's page lists video-to-video as a general mode; read its docs before assuming it replaces people with reference photos. The Luma figures are from Luma's API pricing page.
Rates and totals
Sume's list is $0.30 per second at 768p and $0.45 at 1080p, multiplied by 1.25 to $0.375 and $0.5625 per second.
| Resolution tier | Luma Ray 3.2 V2V | Sume h3-max-recast |
|---|---|---|
| 540p | $1.08 | not offered |
| 720p / 768p | $2.16 | $3.75 |
| 1080p | $4.32 | $5.625 |
Longer sources
Sume bills recast per source second, so a 30 second source is 30 x $0.375 = $11.25 at 768p and 30 x $0.5625 = $16.875 at 1080p. Luma's page lists 5 and 10 second tiers only for this mode, so a 30 second source is not directly priced there.
Budget the Sume side by source length, since there is no duration to choose: a 12 second clip is 12 x $0.375 = $4.50 at 768p.
Sources
Related posts
More in Comparisons
- MAI-Transcribe-2 'one hour in 10 seconds': a short caption clip
Microsoft lists 1hr audio to 10 sec for MAI-Transcribe-2. What that ratio omits for a 20-second clip, and the Sume STT, detach and caption limits that apply.
- MAI-Transcribe-2-Streaming partials vs Sume STT word timings
MAI-Transcribe-2-Streaming sends partials after about 100 ms in 60 languages. Sume STT is a job that always returns word timings. Which fits what.
- Make an AI avatar say exact words: Tavus echo mode vs Sume
Tavus echo mode sends text or audio straight to the avatar for playback, skipping perception and speech recognition. Sume's avatar video renders your script.
- Micro-drama lead: Avatar 1.0 or Seedance 2.5? Pick by dialogue
Pick Sume Avatar 1.0 for talking to camera and Seedance 2.5 for a lead who moves through places. Cost per minute: $14.70 against $34.67.
Written by Sume