Thanksgiving takeout pre-order clip: Wan 3.0 or Seedance 2.5
A vertical 8-second takeout pre-order clip with a spoken order deadline: $1.32 on Wan 3.0 or $4.94 on Seedance 2.5, voice and caption included, from docs.

For a restaurant selling Thanksgiving takeout dinners, a vertical 8-second clip with a spoken order deadline costs $1.319 on Wan 3.0 or $4.9414 on Seedance 2.5, including a 400-character voice-over, one Timeline render and one caption job. The only difference between the two builds is the clip: $1.00 on Wan, $4.6224 on Seedance.
Other posts cover a bakery's pie graphics and the holiday-hours bulk run. This one is the pre-order clip for a full dinner, and it states what each model price buys.
The visual: one plated-table still
Use a photo of the packed dinner box on a counter. Wan 3.0 at 720p is $0.125 per second, so 8 seconds is $1.00. Seedance 2.5 bills tokens: ceil(width x height x seconds x 24 / 1024) at $0.0214 per 1,000 tokens at 720p before the 1.25 multiplier. For 720 x 1280 and 8 seconds that is 172,800 tokens, $3.698 list, $4.6224 charged (Video Router). Both accept 9:16 and 8 seconds is inside both models' duration ranges.
The voice: deadline and pickup
Write a script of about 400 characters: what is in the box, the price you decide, the last day to order and the pickup window. TTS 1.0 is $0.0475 per 1,000 characters, so 400 characters is $0.019. Keep dates in words: 'order by Sunday the 22nd' reads more reliably than '10/22' in a voice model. Read the script out loud at normal speed first; if it runs past 8 seconds, cut words instead of speeding it up beyond speed 1.2.
Join and caption
Timeline takes the voice as the audio spine and the clip as one slot: $0.10. Captions from script_text align your exact wording to the speech: $0.20 (video captions). Together they add $0.319 for the voice, render and captions on either model.
| Line | Basis | USD |
|---|---|---|
| Clip: Wan 3.0 / Seedance 2.5 | $0.125 per s / tokens | $1.00 / $4.6224 |
| TTS, 400 characters | $0.0475 per 1,000 | $0.019 |
| Timeline render | $0.10 per output minute | $0.10 |
| Caption job | $0.20 per job | $0.20 |
| Total Wan / Seedance | $1.319 / $4.9414 |
Which one to run
Run one clip on each with the same photo and prompt, look at the food and the steam, and pick. If Wan's version reads well, you save $3.6224 on every clip. Re-read the catalog before a large run: the figures above are Sume's list on 2026-10-04.
Related posts
More in Use cases
- Threads 5 minutes and 1 GB: one render for feed and Threads
Threads video can run up to 5 minutes and 1 GB. Render one vertical master with Timeline 1.0, at $0.10 per output minute, and probe it before you post.
- Thumbnail A/B test: five GPT Image 2.5 variants at medium vs high
Five 16:9 thumbnail variants cost about $0.04 at medium and $0.16 at high in GPT Image 2.5 output tokens. A budget for 20 videos and the review step.
- TikTok ad rejected for a fake play button? The interactive rule
TikTok bans fake play buttons, fake close buttons, dead carousel dots and fake CTA pop-ups inside ad video. How to check an AI-made clip for them before review.
- TikTok AI label source: detection, creator label or TikTok's own tool
TikTok says its AIGC labels now show whether a label came from detection, a creator label, or TikTok's AI tools. What each means for a clip you made elsewhere.
Written by Sume