100 draft clips at 5 seconds: Omni 360p $19, Wan 480p $32, Grok $7
A hundred 5-second drafts then ten finals. Draft on Omni 360p, Wan or H3 480p, or Grok from stills, then rerun winners on Omni 1080p: totals for each route.

A cheap draft round is how most teams spend video credits well: generate many small clips, pick ten, and only pay full price for those. The draft tier you pick matters less than whether its look matches the final model, so the routes below keep the draft and the final in the same model family where Sume allows.
All rows are Sume billed rates (list x 1.25, rounded up per job), 5 seconds each, 100 drafts and 10 finals.
Four draft routes
| Draft route | One draft | 100 drafts | + 10 Omni 1080p finals | Total |
|---|---|---|---|---|
| Omni 360p | $0.19 | $19.00 | $9.40 | $28.40 |
| Wan 3.0 480p | $0.32 | $32.00 | $9.40 | $41.40 |
| MiniMax H3 480p | $0.32 | $32.00 | $9.40 | $41.40 |
| Grok Imagine (from a still) | $0.07 | $7.00 | $9.40 | $16.40 |
Which route is honest
Only the first route is a true draft: the same prompt on the same model, at 360p, then the same prompt at 1080p. The motion and framing will be close, though not identical, because Sume models do not accept a seed, so you cannot pin a draft to its final. Treat the draft as a screening tool for the prompt, not a preview of the exact take.
Wan 3.0 and H3 at 480p cost 13 cents more per draft, but they are the right drafts when the final will run on that same model. Grok Imagine drafts are the cheapest, yet they start from a still and do not carry sound, so use them to test a first frame, not a script.
Same prompt, two tiers
Change only resolution. Keep the prompt and the aspect ratio fixed so you compare like with like.
# draft
{"model":"gemini-omni-flash-1.1","prompt":"Product turntable, soft studio light","duration":5,"resolution":"360p","aspect_ratio":"9:16"}
# final
{"model":"gemini-omni-flash-1.1","prompt":"Product turntable, soft studio light","duration":5,"resolution":"1080p","aspect_ratio":"9:16"}Sources
Related posts
More in Pricing
- 100 product shots on GPT Image 2.5: $1.65 from text, $2.09 as edits
100 product shots on ChatGPT Image 2.5 at 1024x1024 cost $1.65 from text or $2.09 as one-reference edits at medium on Sume.
- 100 rendered 45-second videos: Sume timeline fees are up to $7.50
Sume Timeline renders are $0.10 per output minute, so 100 videos of 45 seconds is 75 output minutes and up to $7.50. Footage and voice cost extra.
- 100 three-second bumpers: Omni Flash 360p $11.25 vs Wan 480p $18.75
Three-second AI bumpers cost $0.1125 each on Omni Flash 1.1 at 360p and $0.1875 on Wan 3.0 at 480p, so 100 are $11.25 and $18.75 on Sume.
- 100 videos of 11 seconds: which models fit and the total cost
100 eleven-second videos cost $137.50 on Wan 3.0 at 720p and $295.54 on Seedance 2.5 at 480p; Omni Flash stops at 10 seconds, so it cannot do them.
Written by Sume