Luma Ray3.2 video-to-video vs Sume Omni edit: cost per clip
Editing a 5-second clip costs $1.44 on Luma Ray3.2 at 720p and $0.625 with Gemini Omni Flash edit on Sume. 10 seconds is $2.16 versus $1.25. Math inside.

To restyle or edit an existing 5-second clip, Luma Ray3.2 video-to-video lists $1.44 at 720p; Sume's Gemini Omni Flash 1.1 edit mode costs $0.625 for the same five seconds at 720p. At 10 seconds it is $2.16 on Luma and $1.25 on Sume. The Luma prices are from the Luma API pricing page (read 2026-10-04).
These are two different models doing the same kind of job, so the price gap says nothing about which edit looks better.
| Clip | Luma Ray3.2 (720p) | Sume Omni edit (720p) | Luma / Sume |
|---|---|---|---|
| 5 seconds | $1.44 | 5 x $0.125 = $0.625 | 2.30x |
| 10 seconds | $2.16 | 10 x $0.125 = $1.25 | 1.73x |
Luma's full ladder
Luma lists Ray3.2 video-to-video at $0.72, $1.44 and $2.16 for a 5-second clip at 540p, 720p and 1080p, and $1.08, $2.16 and $4.32 for 10 seconds. The 10-second price is 1.5x the 5-second price at 540p ($1.08 / $0.72) and 2.0x at 1080p ($4.32 / $2.16), so a longer clip is not a linear step.
| Resolution | 5 seconds | 10 seconds |
|---|---|---|
| 540p | $0.72 | $1.08 |
| 720p | $1.44 | $2.16 |
| 1080p | $2.16 | $4.32 |
How Sume prices an edit
Omni edit mode takes a video_url and a prompt that describes the change; resolution is optional and defaults to 720p. Sume bills the provider list x 1.25 per output second as a function of resolution, per the Video Router docs. The list is $0.10 per second at 720p and $0.15 at 1080p, so Sume is $0.125 and $0.1875 per second.
At 1080p a 5-second edit is 5 x $0.1875 = $0.9375, against $2.16 on Luma; 10 seconds is $1.875 against $4.32.
- Native audio is always on for Omni, so the result carries a soundtrack.
- The edit request carries no duration field; the output follows the source clip, so a 10-second source bills 10 seconds.
Steps to compare on your own clip
- Cut the source to the length you actually need; every second is billed.
- Run the same instruction on both services at 720p.
- Judge the result against your brief, then multiply by the retries you needed.
- Only then compare per-clip cost: price x attempts.
A batch of 50 edits
Fifty 5-second edits at 720p cost 50 x $1.44 = $72.00 on Luma and 50 x $0.625 = $31.25 on Sume, a $40.75 difference. At 1080p they cost 50 x $2.16 = $108.00 on Luma and 50 x $0.9375 = $46.88 on Sume. If the edit lands on the first try, the cheaper model wins on price; if it takes three tries, multiply each by three before comparing.
Caveats
Luma's tiers differ by resolution and HDR options, which multiply the price (HDR is 2x). Omni's edit window is 3 to 10 seconds, so longer clips need splitting. Prices were read on 2026-10-04 and promotions change.
Sources
Related posts
More in Comparisons
- MAI-Transcribe-2 timestamps option vs Sume STT: words[] are always on
On Microsoft's MAI-Transcribe you ask for word timestamps in modelOptions. Sume STT returns words[] with start and end on every job: what that changes for you.
- MAI-Voice-2.1 lists Korean, Thai, Vietnamese: Sume's language field
Microsoft's MAI-Voice-2.1 page lists 23 languages including Korean, Thai and Vietnamese. What to check before you plan a non-English voiceover on Sume TTS.
- MAI-Voice-2.1 or Flash for explainer narration: latency barely matters
Microsoft lists about 550 ms vs 45 ms model inference. For narration rendered ahead of time, pick on quality and price, then see how Sume jobs return audio.
- Meshy 7.1 Ultra 4K: when a 3D model is overkill
Meshy 7.1 added Ultra 4K meshes of up to 80 million triangles. Sume has no image-to-3D model; to show a product turning, a short video does it.
Written by Sume