Wan 3.0 for 500 prompts: 480p, 720p or 1080p costs 1x, 2x or 4x
On Sume, a 500-prompt Wan 3.0 library costs $250.00 at 480p, $500.00 at 720p and $1,000.00 at 1080p for 8-second clips. Resolution sets the bill.

A 500-prompt library of 8-second Wan 3.0 clips costs $250.00 at 480p, $500.00 at 720p and $1,000.00 at 1080p on Sume. The price doubles with each resolution step, so the resolution you pick matters more than any choice of prompt.
Teams that lose the sora-2 endpoints (OpenAI removed the Videos API on 2026-09-24, with no replacement listed) often re-render their whole stored library once. At 500 prompts, a one-step resolution decision moves the total by hundreds of dollars.
The three tiers
Wan 3.0 is priced per second by resolution, with a list price of $0.05, $0.10 and $0.20. Sume bills the list price times 1.25 and rounds up to the cent for each clip.
| Clip | 480p per clip | 720p per clip | 1080p per clip |
|---|---|---|---|
| 8 s | $0.50 | $1.00 | $2.00 |
| 12 s | $0.75 | $1.50 | $3.00 |
| 500 clips at 8 s | $250.00 | $500.00 | $1,000.00 |
| 500 clips at 12 s | $375.00 | $750.00 | $1,500.00 |
The arithmetic, once
At 480p an 8-second clip is 8 x $0.05 = $0.40 list. Multiply by 1.25 and you get $0.50, which is 50 cents. Multiply by 500 clips: $250.00. The 720p and 1080p tiers repeat the step with $0.10 and $0.20 per second, giving 100 and 200 cents.
Because every clip is a whole number of cents already, there is no rounding drift across 500 clips at these lengths. That is not true for every length; see the rounding effect on 3-second Wan clips.
Longer clips scale the same way
Wan 3.0 accepts 2 to 30 seconds, so a library that held long shots can stay on one model. A 12-second 720p clip is 150 cents, a 20-second one is 250 cents, and a 30-second 1080p clip is 750 cents, which is $7.50. The price is linear in seconds at each tier, so a 30-second clip costs the same as three 10-second clips at the same resolution, and you can split or join by whichever editing step is simpler for you.
A 480p pass is a sketch, not a preview of the final
Sume's v1 video models do not accept a seed, so a 480p render is a different generation from the 1080p render of the same prompt. Use the cheap tier to judge whether a prompt is usable, not to approve a specific take.
A reasonable plan for a 500-prompt library is to render all 500 at 480p for $250.00, drop the prompts that fail, and re-render only the survivors at the final resolution. If 150 survive and go to 1080p, the second pass costs $300.00, for a total of $550.00, against $1,000.00 if you had rendered everything at 1080p.
- Set
resolutionexplicitly on every request; do not rely on a default. - Wan 3.0 accepts 2 to 30 seconds, so the length column above can go beyond 12.
- Read the live numbers from
GET /v1/videos/modelsbefore a large run.
Sources
Related posts
More in Pricing
- Wan 3.0 per finished minute with 1, 2 or 3 480p drafts per 30 s
A finished minute is two 30-second blocks. Cost per minute on Sume with one to three 480p drafts per block and a 720p or 1080p final, plus the timeline render.
- Wan 3.0 by Alibaba region: Singapore 10 cents, China 8.25 at 720p
Alibaba lists Wan 3.0 at $0.10 a second at 720p in Singapore and $0.082513 in China (Beijing) and other regions, 17.5% less. Sume's own price table is separate.
- A weekly 30-minute solo podcast: 52 transcripts for the year
52 half-hour episodes are 26 hours: $14.04 on MAI-Transcribe-2-Streaming's intro price and $15.60 on Sume STT, in three 10-minute jobs each.
- A weekly bedtime story: 52 narrations at 5,000 characters
A 5,000-character story each week for a year is 260,000 characters: $5.72 on MAI-Voice-2.1, $3.90 on Flash, $12.48 on Sume as 52 jobs.
Written by Sume